Uncertainty Theory, 3rd Ed., 2009 Liu B.

Uncertainty Theory
Third Edition
Baoding Liu
Uncertainty Theory Laboratory
Department of Mathematical Sciences
Tsinghua University
Beijing 100084, China
liu@tsinghua.edu.cn
http://orsc.edu.cn/liu
3rd Edition
c 2009 by UTLAB
Japanese Translation Version
c 2008 by WAP
2nd Edition c 2007 by Springer-Verlag Berlin
Chinese Version
c 2005 by Tsinghua University Press
1st Edition
c 2004 by Springer-Verlag Berlin
Reference to this book should be made as follows:

Liu B, Uncertainty Theory, 3rd ed., http://orsc.edu.cn/liu/ut.pdf
Contents
Preface ix
1 Probability Theory 1
1.1 Probability Space . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Random Variables . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3 Probability Distribution . . . . . . . . . . . . . . . . . . . . . 7
1.4 Independence . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.5 Identical Distribution . . . . . . . . . . . . . . . . . . . . . . 13
1.6 Expected Value . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.7 Variance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
1.8 Moments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
1.9 Critical Values . . . . . . . . . . . . . . . . . . . . . . . . . . 26
1.10 Entropy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
1.11 Distance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
1.12 Inequalities . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
1.13 Convergence Concepts . . . . . . . . . . . . . . . . . . . . . . 35
1.14 Conditional Probability . . . . . . . . . . . . . . . . . . . . . 40
2 Credibility Theory 45
2.1 Credibility Space . . . . . . . . . . . . . . . . . . . . . . . . . 45
2.2 Fuzzy Variables . . . . . . . . . . . . . . . . . . . . . . . . . 57
2.3 Membership Function . . . . . . . . . . . . . . . . . . . . . . 58
2.4 Credibility Distribution . . . . . . . . . . . . . . . . . . . . . 62
2.5 Independence . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
2.7 Extension Principle of Zadeh . . . . . . . . . . . . . . . . . . 72
2.8 Expected Value . . . . . . . . . . . . . . . . . . . . . . . . . 73
2.9 Variance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
2.10 Moments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
2.11 Critical Values . . . . . . . . . . . . . . . . . . . . . . . . . . 90
2.12 Entropy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
2.13 Distance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
2.14 Inequalities . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
vi Contents

2.16 Conditional Credibility . . . . . . . . . . . . . . . . . . . . . 108
3 Chance Theory 115

3.1 Chance Space . . . . . . . . . . . . . . . . . . . . . . . . . . 115
3.2 Hybrid Variables . . . . . . . . . . . . . . . . . . . . . . . . . 124
3.3 Chance Distribution . . . . . . . . . . . . . . . . . . . . . . . 130
3.4 Expected Value . . . . . . . . . . . . . . . . . . . . . . . . . 132
3.5 Variance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
3.6 Moments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
3.7 Critical Values . . . . . . . . . . . . . . . . . . . . . . . . . . 137
3.8 Distance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
3.9 Inequalities . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
3.11 Conditional Chance . . . . . . . . . . . . . . . . . . . . . . . 147
4 Uncertainty Theory 153

4.1 Uncertainty Space . . . . . . . . . . . . . . . . . . . . . . . . 153
4.2 Uncertain Variables . . . . . . . . . . . . . . . . . . . . . . . 162
4.3 Identification Function . . . . . . . . . . . . . . . . . . . . . 164
4.4 Uncertainty Distribution . . . . . . . . . . . . . . . . . . . . 170
4.5 Independence . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
4.7 Operational Law . . . . . . . . . . . . . . . . . . . . . . . . . 175
4.8 Expected Value . . . . . . . . . . . . . . . . . . . . . . . . . 176
4.9 Variance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
4.10 Moments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
4.11 Critical Values . . . . . . . . . . . . . . . . . . . . . . . . . . 187
4.12 Entropy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
4.13 Distance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
4.14 Inequalities . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
4.16 Conditional Uncertainty . . . . . . . . . . . . . . . . . . . . . 195
5 Uncertain Process 201

5.1 Basic Definitions . . . . . . . . . . . . . . . . . . . . . . . . . 201
5.2 Continuity Concepts . . . . . . . . . . . . . . . . . . . . . . . 202
5.3 Renewal Process . . . . . . . . . . . . . . . . . . . . . . . . . 202
5.4 Stationary Process . . . . . . . . . . . . . . . . . . . . . . . . 204
6 Uncertain Calculus 205

6.1 Canonical Process . . . . . . . . . . . . . . . . . . . . . . . . 205
6.2 Uncertain Integral . . . . . . . . . . . . . . . . . . . . . . . . 207
6.3 Chain Rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208
6.4 Integration by Parts . . . . . . . . . . . . . . . . . . . . . . . 210
Contents vii
7 Uncertain Differential Equation 211

7.1 Uncertain Differential Equation . . . . . . . . . . . . . . . . 211
7.2 Existence and Uniqueness Theorem . . . . . . . . . . . . . . 212
7.3 Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212
7.4 Uncertain Finance . . . . . . . . . . . . . . . . . . . . . . . . 213
7.5 Uncertain Control . . . . . . . . . . . . . . . . . . . . . . . . 216
7.6 Uncertain Filtering . . . . . . . . . . . . . . . . . . . . . . . 216
7.7 Uncertain Dynamical System . . . . . . . . . . . . . . . . . . 217
8 Uncertain Logic 219

8.1 Uncertain Proposition . . . . . . . . . . . . . . . . . . . . . . 219
8.2 Uncertain Formula . . . . . . . . . . . . . . . . . . . . . . . . 220
8.3 Truth Value . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
8.4 Some Laws . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
9 Uncertain Inference 225

9.1 Uncertain Sets . . . . . . . . . . . . . . . . . . . . . . . . . . 225
9.2 Inference Rule . . . . . . . . . . . . . . . . . . . . . . . . . . 226
9.3 Inference Rule with Multiple Antecedents . . . . . . . . . . . 227
9.4 Inference Rule with Multiple If-Then Rules . . . . . . . . . . 227
9.5 General Inference Rule . . . . . . . . . . . . . . . . . . . . . 228
A Measure and Integral 229

A.1 Measurable Sets . . . . . . . . . . . . . . . . . . . . . . . . . 229
A.2 Classical Measures . . . . . . . . . . . . . . . . . . . . . . . . 231
A.3 Measurable Functions . . . . . . . . . . . . . . . . . . . . . . 232
A.4 Lebesgue Integral . . . . . . . . . . . . . . . . . . . . . . . . 234
B Euler-Lagrange Equation 238
C Logic 239
C.1 Traditional Logic . . . . . . . . . . . . . . . . . . . . . . . . . 239
C.2 Probabilistic Logic . . . . . . . . . . . . . . . . . . . . . . . . 240
C.3 Credibilistic Logic . . . . . . . . . . . . . . . . . . . . . . . . 240
C.4 Hybrid Logic . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
C.5 Uncertain Logic . . . . . . . . . . . . . . . . . . . . . . . . . 243
D Inference 244
D.1 Traditional Inference . . . . . . . . . . . . . . . . . . . . . . 244
D.2 Fuzzy Inference . . . . . . . . . . . . . . . . . . . . . . . . . 244
D.3 Hybrid Inference . . . . . . . . . . . . . . . . . . . . . . . . . 245
D.4 Uncertain Inference . . . . . . . . . . . . . . . . . . . . . . . 246
E Maximum Uncertainty Principle 247
F Why Uncertain Measure? 248

viii Contents
Bibliography 251
List of Frequently Used Symbols 267
Index 268
Preface
There are various types of uncertainty. This fact provides a motivation to

study the behavior of uncertain phenomena because “uncertainty is absolute
and certainty is relative” in the real world.
Randomness is a basic type of objective uncertainty, and probability the-
ory is a branch of mathematics for studying the behavior of random phe-
nomena. The study of probability theory was started by Pascal and Fermat
(1654), and an axiomatic foundation of probability theory was given by Kol-
mogoroff (1933) in his Foundations of Probability Theory. Probability theory
has been widely applied in science and engineering. Chapter 1 will provide
the probability theory.
Fuzziness is a basic type of subjective uncertainty initiated by Zadeh
(1965). Credibility theory is a branch of mathematics for studying the be-
havior of fuzzy phenomena. The study of credibility theory was started by
Liu and Liu (2002), and an axiomatic foundation of credibility theory was
given by Liu (2004) in his Uncertainty Theory. Chapter 2 will introduce the
credibility theory.
Sometimes, fuzziness and randomness simultaneously appear in a sys-
tem. A hybrid variable was proposed by Liu (2006) as a tool to describe the
quantities with fuzziness and randomness. Both fuzzy random variable and
random fuzzy variable are instances of hybrid variable. In addition, Li and
Liu (2009) introduced the concept of chance measure for hybrid events. After
that, chance theory was developed steadily. Essentially, chance theory is a
hybrid of probability theory and credibility theory. Chapter 3 will offer the
chance theory.
A lot of surveys showed that the measure of union of events is not neces-
sarily the maximum measure except those events are independent. This fact
implies that many subjectively uncertain systems do not behave fuzziness
and membership function is no longer helpful. In order to deal with this type
of subjective uncertainty, Liu (2007) founded an uncertainty theory in his
Uncertainty Theory, and had it become a branch of mathematics based on
normality, monotonicity, self-duality, countable subadditivity, and product
measure axioms. Chapter 4 is devoted to the uncertainty theory.
For this new edition the entire text has been totally rewritten. More im-
portantly, uncertain process, uncertain calculus, uncertain differential equa-
tion, uncertain logic and uncertain inference are completely new. In addi-
tion, uncertain finance, uncertain control, uncertain filtering and uncertain
dynamical system have been added.
x Preface
The book is suitable for mathematicians, researchers, engineers, design-

ers, and students in the field of mathematics, information science, operations
research, industrial engineering, computer science, artificial intelligence, fi-
nance, and management science. The readers will learn the axiomatic ap-
proach of uncertainty theory, and find this work a stimulating and useful
reference.
A Guide for the Reader

The logic dependence of chapters is illustrated by the figure below. The
readers are not required to read the book from cover to cover. In fact, the
readers may skip over the first three chapters and start at Chapter 4.
... ... ...
... .. ...
......... ......... .........
. .......
. . .......
. . .......
.
.
... ..... . .
... ...... .
... ......
.... .. .
.. .. .... ..
....
...
1
.... .....
............
.
2
.... .....
............
. ...
4
.... ......
.............
.
...... . . .
...... ..
.. ....... .
..
. . ...... ...........
...... . . ......
........ ............. ...... ........
..................... ............ ..............
.. ...... ................ ..... .......
... .. ... ... ... .
.... ....
3
.... .....
...........
. 5
.... .....
...........
. . ... 8
................
...
... ...
... ....
........ ........
.. ..
........... ...........
.... ...... .... ......
..... .....
6
.... .....
..........
. 9
.... ....
...........
.
...
..
........
..
........
..... .......
.....
7
.... .....
...........
.
Acknowledgment
This work was supported by a series of grants from National Natural Science
Foundation, Ministry of Education, and Ministry of Science and Technology
of China.
Baoding Liu
Tsinghua University
http://orsc.edu.cn/liu
January 17, 2009
Chapter 1
Probability Theory
Probability measure is essentially a set function (i.e., a function whose ar-

gument is a set) satisfying normality, nonnegativity and countable additivity
axioms. Probability theory is a branch of mathematics for studying the be-
havior of random phenomena. The emphasis in this chapter is mainly on
probability space, random variable, probability distribution, independence,
identical distribution, expected value, variance, moments, critical values, en-
tropy, distance, convergence almost surely, convergence in probability, con-
vergence in mean, convergence in distribution, and conditional probability.
The main results in this chapter are well-known. For this reason the credit
references are not provided.
1.1 Probability Space

Let Ω be a nonempty set, and A a σ-algebra over Ω. If Ω is countable, usually
A is the power set of Ω. If Ω is uncountable, for example Ω = [0, 1], usually
A is the Borel algebra of Ω. Each element in A is called an event.
In order to present an axiomatic definition of probability, it is necessary
to assign to each event A a number Pr{A} which indicates the probability
that A will occur. In order to ensure that the number Pr{A} has certain
mathematical properties which we intuitively expect a probability to have,
the following three axioms must be satisfied:
Axiom 1. (Normality) Pr{Ω} = 1.
Axiom 2. (Nonnegativity) Pr{A} ≥ 0 for any event A.
Axiom 3. (Countable Additivity) For every countable sequence of mutually
disjoint events {Ai }, we have
(∞ ) ∞
[ X
Pr Ai = Pr{Ai }. (1.1)
i=1 i=1
2 Chapter 1 - Probability Theory
Definition 1.1 The set function Pr is called a probability measure if it sat-

isfies the normality, nonnegativity, and countable additivity axioms.
Example 1.1: Let Ω = {ω1 , ω2 , · · · }, and let A be the power set of Ω.

Assume that p1 , p2 , · · · are nonnegative numbers such that p1 + p2 + · · · = 1.
Define a set function on A as
pi , A ∈ A.
X
Pr{A} = (1.2)
ωi ∈A
Then Pr is a probability measure.
Example 1.2: Let Ω = [0, 1] and let A be the Borel algebra over Ω. If Pr is
the Lebesgue measure, then Pr is a probability measure.
Example 1.3: Let φ be a nonnegative and integrable function on < (the set
of real numbers) such that
Z
φ(x)dx = 1. (1.3)
<
Then for any Borel set A, the set function

Z
Pr{A} = φ(x)dx (1.4)
A
is a probability measure on <.

Theorem 1.1 Let Ω be a nonempty set, A a σ-algebra over Ω, and Pr a
probability measure. Then we have
(a) Pr{∅} = 0;
(b) Pr is self-dual, i.e., Pr{A} + Pr{Ac } = 1 for any A ∈ A;
(c) Pr is increasing, i.e., Pr{A} ≤ Pr{B} whenever A ⊂ B.
Proof: (a) Since ∅ and Ω are disjoint events and ∅ ∪ Ω = Ω, we have
Pr{∅} + Pr{Ω} = Pr{Ω} which makes Pr{∅} = 0.
(b) Since A and Ac are disjoint events and A ∪ Ac = Ω, we have Pr{A} +
Pr{Ac } = Pr{Ω} = 1.
(c) Since A ⊂ B, we have B = A ∪ (B ∩ Ac ), where A and B ∩ Ac are
disjoint events. Therefore Pr{B} = Pr{A} + Pr{B ∩ Ac } ≥ Pr{A}.
Probability Continuity Theorem

Theorem 1.2 (Probability Continuity Theorem) Let Ω be a nonempty set,
A a σ-algebra over Ω, and Pr a probability measure. If A1 , A2 , · · · ∈ A and
limi→∞ Ai exists, then
n o
lim Pr{Ai } = Pr lim Ai . (1.5)
i→∞ i→∞
Section 1.1 - Probability Space 3
Proof: Step 1: Suppose {Ai } is an increasing sequence. Write Ai → A and

A0 = ∅. Then {Ai \Ai−1 } is a sequence of disjoint events and
∞
[ k
[
(Ai \Ai−1 ) = A, (Ai \Ai−1 ) = Ak
i=1 i=1
for k = 1, 2, · · · Thus we have

∞ ∞
S P
Pr{A} = Pr (Ai \Ai−1 ) = Pr {Ai \Ai−1 }
i=1 i=1
k
k

P S
= lim Pr {Ai \Ai−1 } = lim Pr (Ai \Ai−1 )
k→∞ i=1 k→∞ i=1
= lim Pr{Ak }.
k→∞
Step 2: If {Ai } is a decreasing sequence, then the sequence {A1 \Ai } is

clearly increasing. It follows that
n o
Pr{A1 } − Pr{A} = Pr lim (A1 \Ai ) = lim Pr {A1 \Ai }
i→∞ i→∞
= Pr{A1 } − lim Pr{Ai }

i→∞
which implies that Pr{Ai } → Pr{A}.

Step 3: If {Ai } is a sequence of events such that Ai → A, then for each
k, we have
∞
\ [∞
Ai ⊂ Ak ⊂ Ai .
i=k i=k
Since Pr is increasing, we have

(∞ ) (∞ )
\ [
Pr Ai ≤ Pr{Ak } ≤ Pr Ai .
i=k i=k
Note that
∞
\ ∞
[
Ai ↑ A, Ai ↓ A.
i=k i=k
It follows from Steps 1 and 2 that Pr{Ai } → Pr{A}.
Probability Space
Definition 1.2 Let Ω be a nonempty set, A a σ-algebra over Ω, and Pr a
probability measure. Then the triplet (Ω, A, Pr) is called a probability space.
Example 1.4: Let Ω = {ω1 , ω2 , · · · }, A the power set of Ω, and Pr a

probability measure defined by (1.2). Then (Ω, A, Pr) is a probability space.
Example 1.5: Let Ω = [0, 1], A the Borel algebra over Ω, and Pr the
Lebesgue measure. Then ([0, 1], A, Pr) is a probability space, and sometimes
is called Lebesgue unit interval. For many purposes it is sufficient to use it
as the basic probability space.
Product Probability Space

Let (Ωi , Ai , Pri ), i = 1, 2, · · · , n be probability spaces, and Ω = Ω1 × Ω2 ×
· · · × Ωn , A = A1 × A2 × · · · × An . Note that the probability measures
Pri , i = 1, 2, · · · , n are finite. It follows from the product measure theorem
that there is a unique measure Pr on A such that
Pr{A1 × A2 × · · · × An } = Pr1 {A1 } × Pr2 {A2 } × · · · × Prn {An }
for any Ai ∈ Ai , i = 1, 2, · · · , n. This conclusion is called the product proba-

bility theorem. The measure Pr is also a probability measure since
Pr{Ω} = Pr1 {Ω1 } × Pr2 {Ω2 } × · · · × Prn {Ωn } = 1.
Such a probability measure is called the product probability measure, denoted

by Pr = Pr1 × Pr2 × · · · × Prn .
Definition 1.3 Let (Ωi , Ai , Pri ), i = 1, 2, · · · , n be probability spaces, and

Ω = Ω1 × Ω2 × · · · × Ωn , A = A1 × A2 × · · · × An , Pr = Pr1 × Pr2 × · · · × Prn .
Then the triplet (Ω, A, Pr) is called the product probability space.
Infinite Product Probability Space

Let (Ωi , Ai , Pri ), i = 1, 2, · · · be an arbitrary sequence of probability spaces,
and
Ω = Ω1 × Ω2 × · · · , A = A1 × A2 × · · · (1.6)
It follows from the infinite product measure theorem that there is a unique
probability measure Pr on A such that
Pr {A1 × · · · × An × Ωn+1 × Ωn+2 × · · · } = Pr1 {A1 } × · · · × Prn {An }
for any measurable rectangle A1 × · · · × An × Ωn+1 × Ωn+2 × · · · and all

n = 1, 2, · · · The probability measure Pr is called the infinite product of
Pri , i = 1, 2, · · · and is denoted by
Pr = Pr1 × Pr2 × · · · (1.7)
Definition 1.4 Let (Ωi , Ai , Pri ), i = 1, 2, · · · be probability spaces, and Ω =

Ω1 × Ω2 × · · · , A = A1 × A2 × · · · , Pr = Pr1 × Pr2 × · · · Then the triplet
(Ω, A, Pr) is called the infinite product probability space.
Section 1.2 - Random Variables 5
1.2 Random Variables

Definition 1.5 A random variable is a measurable function from a proba-
bility space (Ω, A, Pr) to the set of real numbers, i.e., for any Borel set B of
real numbers, the set

{ξ ∈ B} = {ω ∈ Ω ξ(ω) ∈ B} (1.8)
is an event.
Example 1.6: Take (Ω, A, Pr) to be {ω1 , ω2 } with Pr{ω1 } = Pr{ω2 } = 0.5.
Then the function (
0, if ω = ω1
ξ(ω) =
1, if ω = ω2
is a random variable.
Example 1.7: Take (Ω, A, Pr) to be the interval [0, 1] with Borel algebra
and Lebesgue measure. We define ξ as an identity function from Ω to [0,1].
Since ξ is a measurable function, it is a random variable.
Example 1.8: A deterministic number c may be regarded as a special ran-

dom variable. In fact, it is the constant function ξ(ω) ≡ c on the probability
space (Ω, A, Pr).
Definition 1.6 A random variable ξ is said to be

(a) nonnegative if Pr{ξ < 0} = 0;
(b) positive if Pr{ξ ≤ 0} = 0;
(c) continuous if Pr{ξ = x} = 0 for each x ∈ <;
(d) simple if there exists a finite sequence {x1 , x2 , · · · , xm } such that
Pr {ξ 6= x1 , ξ 6= x2 , · · · , ξ 6= xm } = 0; (1.9)
(e) discrete if there exists a countable sequence {x1 , x2 , · · · } such that
Pr {ξ 6= x1 , ξ 6= x2 , · · · } = 0. (1.10)
Definition 1.7 Let ξ1 and ξ2 be random variables defined on the probability

space (Ω, A, Pr). We say ξ1 = ξ2 if ξ1 (ω) = ξ2 (ω) for almost all ω ∈ Ω.
Random Vector
Definition 1.8 An n-dimensional random vector is a measurable function
from a probability space (Ω, A, Pr) to the set of n-dimensional real vectors,
i.e., for any Borel set B of <n , the set

{ξ ∈ B} = ω ∈ Ω ξ(ω) ∈ B (1.11)
is an event.
Theorem 1.3 The vector (ξ1 , ξ2 , · · · , ξn ) is a random vector if and only if

ξ1 , ξ2 , · · · , ξn are random variables.
Proof: Write ξ = (ξ1 , ξ2 , · · · , ξn ). Suppose that ξ is a random vector on the

probability space (Ω, A, Pr). For any Borel set B of <, the set B × <n−1 is
also a Borel set of <n . Thus we have
{ξ1 ∈ B} = {ξ1 ∈ B, ξ2 ∈ <, · · · , ξn ∈ <} = {ξ ∈ B × <n−1 } ∈ A
which implies that ξ1 is a random variable. A similar process may prove that
ξ2 , ξ3 , · · · , ξn are random variables.
Conversely, suppose that all ξ1 , ξ2 , · · · , ξn are random variables on the
probability space (Ω, A, Pr). We define
B = B ⊂ <n {ξ ∈ B} ∈ A .

The vector ξ = (ξ1 , ξ2 , · · · , ξn ) is proved to be a random vector if we can

prove that B contains all Borel sets of <n . First, the class B contains all
open intervals of <n because
( n
) n
{ξi ∈ (ai , bi )} ∈ A.
Y \
ξ∈ (ai , bi ) =
i=1 i=1
Next, the class B is a σ-algebra of <n because (i) we have <n ∈ B since
{ξ ∈ <n } = Ω ∈ A; (ii) if B ∈ B, then {ξ ∈ B} ∈ A, and
{ξ ∈ B c } = {ξ ∈ B}c ∈ A
which implies that B c ∈ B; (iii) if Bi ∈ B for i = 1, 2, · · · , then {ξ ∈ Bi } ∈ A

and
∞ ∞
( )
{ξ ∈ Bi } ∈ A
[ [
ξ∈ Bi =
i=1 i=1
which implies that ∪i Bi ∈ B. Since the smallest σ-algebra containing all

open intervals of <n is just the Borel algebra of <n , the class B contains all
Borel sets of <n . The theorem is proved.
Random Arithmetic
In this subsections, we will suppose that all random variables are defined on a
common probability space. Otherwise, we may embed them into the product
probability space.
Definition 1.9 Let f : <n → < be a measurable function, and ξ1 , ξ2 , · · · , ξn

random variables defined on the probability space (Ω, A, Pr). Then ξ =
f (ξ1 , ξ2 , · · · , ξn ) is a random variable defined by
ξ(ω) = f (ξ1 (ω), ξ2 (ω), · · · , ξn (ω)), ∀ω ∈ Ω. (1.12)

Section 1.3 - Probability Distribution 7
Example 1.9: Let ξ1 and ξ2 be random variables on the probability space

(Ω, A, Pr). Then their sum is
(ξ1 + ξ2 )(ω) = ξ1 (ω) + ξ2 (ω), ∀ω ∈ Ω
and their product is
(ξ1 × ξ2 )(ω) = ξ1 (ω) × ξ2 (ω), ∀ω ∈ Ω.
The reader may wonder whether ξ(ω1 , ω2 , · · · , ωn ) defined by (1.9) is a

random variable. The following theorem answers this question.
Theorem 1.4 Let ξ be an n-dimensional random vector, and f : <n → < a

measurable function. Then f (ξ) is a random variable.
Proof: Assume that ξ is a random vector on the probability space (Ω, A, Pr).
For any Borel set B of <, since f is a measurable function, f −1 (B) is also a
Borel set of <n . Thus we have
{f (ξ) ∈ B} = ξ ∈ f −1 (B) ∈ A

which implies that f (ξ) is a random variable.
1.3 Probability Distribution

Definition 1.10 The probability distribution Φ: < → [0, 1] of a random
variable ξ is defined by
Φ(x) = Pr {ξ ≤ x} . (1.13)
That is, Φ(x) is the probability that the random variable ξ takes a value less
than or equal to x.
Example 1.10: Assume that the random variables ξ and η have the same
probability distribution. One question is whether ξ = η or not. Gener-
ally speaking, it is not true. Take (Ω, A, Pr) to be {ω1 , ω2 } with Pr{ω1 } =
Pr{ω2 } = 0.5. We now define two random variables as follows,

−1, if ω = ω1 1, if ω = ω1
ξ(ω) = η(ω) =
1, if ω = ω2 , −1, if ω = ω2 .
Then ξ and η have the same probability distribution,


 0, if x < −1

Φ(x) = 0.5, if − 1 ≤ x < 1

1, if x ≥ 1.

However, it is clear that ξ 6= η in the sense of Definition 1.7.

Theorem 1.5 (Sufficient and Necessary Condition for Probability Distribu-

tion) A function Φ : < → [0, 1] is a probability distribution if and only if it is
an increasing and right-continuous function with
lim Φ(x) = 0; lim Φ(x) = 1. (1.14)

x→−∞ x→+∞
Proof: For any x, y ∈ < with x < y, we have
Φ(y) − Φ(x) = Pr{x < ξ ≤ y} ≥ 0.
Thus the probability distribution Φ is increasing. Next, let {εi } be a sequence

of positive numbers such that εi → 0 as i → ∞. Then, for every i ≥ 1, we
have
Φ(x + εi ) − Φ(x) = Pr{x < ξ ≤ x + εi }.
It follows from the probability continuity theorem that
lim Φ(x + εi ) − Φ(x) = Pr{∅} = 0.

i→∞
Hence Φ is a right-continuous function. Finally,
lim Φ(x) = lim Pr{ξ ≤ x} = Pr{∅} = 0,

x→−∞ x→−∞
lim Φ(x) = lim Pr{ξ ≤ x} = Pr{Ω} = 1.

x→+∞ x→+∞
Conversely, it is known there is a unique probability measure Pr on the Borel

algebra over < such that Pr{(−∞, x]} = Φ(x) for all x ∈ <. Furthermore,
it is easy to verify that the random variable defined by ξ(x) = x from the
probability space (<, A, Pr) to < has the probability distribution Φ.
Remark 1.1: Theorem 1.5 states that the identity function is a universal
function for any probability distribution by defining an appropriate proba-
bility space. In fact, there is a universal probability space for any probability
distribution by defining an appropriate function.
Theorem 1.6 Let Φ be a probability distribution. Then there is a random

variable on Lebesgue unit interval ([0, 1], A, Pr) whose probability distribution
is just Φ.
Proof: Define ξ(ω) = sup{x|Φ(x) ≤ ω} on the Lebesgue unit interval. Then

ξ is a random variable whose probability distribution is just Φ because

Pr{ξ ≤ y} = Pr ω sup {x|Φ(x) ≤ ω} ≤ y = Pr ω ω ≤ Φ(y) = Φ(y)
for any y ∈ <.

Section 1.3 - Probability Distribution 9
Theorem 1.7 A random variable ξ with probability distribution Φ is

(a) nonnegative if and only if Φ(x) = 0 for all x < 0;
(b) positive if and only if Φ(x) = 0 for all x ≤ 0;
(c) simple if and only if Φ is a simple function;
(d) discrete if and only if Φ is a step function;
(e) continuous if and only if Φ is a continuous function.
Proof: The parts (a), (b), (c) and (d) follow immediately from the definition.
Next we prove the part (e). If ξ is a continuous random variable, then
Pr{ξ = x} = 0. It follows from the probability continuity theorem that
lim (Φ(x) − Φ(y)) = lim Pr{y < ξ ≤ x} = Pr{ξ = x} = 0

y↑x y↑x
which proves the left-continuity of Φ. Since a probability distribution is

always right-continuous, Φ is continuous. Conversely, if Φ is continuous,
then we immediately have Pr{ξ = x} = 0 for each x ∈ <.
Definition 1.11 A continuous random variable is said to be (a) singular if

its probability distribution is a singular function; (b) absolutely continuous if
its probability distribution is an absolutely continuous function.
Probability Density Function

Definition 1.12 The probability density function φ: < → [0, +∞) of a ran-
dom variable ξ is a function such that
Z x
Φ(x) = φ(y)dy (1.15)
−∞
holds for all x ∈ <, where Φ is the probability distribution of the random
variable ξ.
R +∞
Let φ : < → [0, +∞) be a measurable function such that −∞ φ(x)dx = 1.
Then φ Ris the probability density function of some random variable because
x
Φ(x) = −∞ φ(y)dy is an increasing and continuous function satisfying (1.14).
Theorem 1.8 (Probability Inversion Theorem) Let ξ be a random variable

whose probability density function φ exists. Then for any Borel set B of <,
we have Z
Pr{ξ ∈ B} = φ(y)dy. (1.16)
B
Proof: Let C be the class of all subsets C of < for which the relation
Z
Pr{ξ ∈ C} = φ(y)dy (1.17)
C
holds. We will show that C contains all Borel sets of <. It follows from the
probability continuity theorem and relation (1.17) that C is a monotone class.
It is also clear that C contains all intervals of the form (−∞, a], (a, b], (b, ∞)
and ∅ since Z a
Pr{ξ ∈ (−∞, a]} = Φ(a) = φ(y)dy,
−∞
Z +∞
Pr{ξ ∈ (b, +∞)} = Φ(+∞) − Φ(b) = φ(y)dy,
b
Z b
Pr{ξ ∈ (a, b]} = Φ(b) − Φ(a) = φ(y)dy,
a
Z
Pr{ξ ∈ ∅} = 0 = φ(y)dy
∅
where Φ is the probability distribution of ξ. Let F be the algebra consisting of
all finite unions of disjoint sets of the form (−∞, a], (a, b], (b, ∞) and ∅. Note
that for any disjoint sets C1 , C2 , · · · , Cm of F and C = C1 ∪ C2 ∪ · · · ∪ Cm ,
we have
Xm Xm Z Z
Pr{ξ ∈ C} = Pr{ξ ∈ Cj } = φ(y)dy = φ(y)dy.
j=1 j=1 Cj C
That is, C ∈ C. Hence we have F ⊂ C. Since the smallest σ-algebra containing

F is just the Borel algebra of <, the monotone class theorem implies that C
contains all Borel sets of <.
Some Special Distributions

Uniform Distribution: A random variable ξ has a uniform distribution if
its probability density function is defined by
 1 , if a ≤ x ≤ b

φ(x) = b−a (1.18)

0, otherwise

denoted by U(a, b), where a and b are given real numbers with a < b.
Exponential Distribution: A random variable ξ has an exponential dis-
tribution if its probability density function is defined by

 1 exp − x , if x ≥ 0

φ(x) = β β (1.19)

0, if x < 0

denoted by EX P(β), where β is a positive number.

Normal Distribution: A random variable ξ has a normal distribution if
its probability density function is defined by
(x − µ)2

1
φ(x) = √ exp − , x∈< (1.20)
σ 2π 2σ 2
denoted by N (µ, σ 2 ), where µ and σ are real numbers.
Section 1.4 - Independence 11
Joint Probability Distribution

Definition 1.13 The joint probability distribution Φ : <n → [0, 1] of a ran-
dom vector (ξ1 , ξ2 , · · · , ξn ) is defined by
Φ(x1 , x2 , · · · , xn ) = Pr {ξ1 ≤ x1 , ξ2 ≤ x2 , · · · , ξn ≤ xn } .
Definition 1.14 The joint probability density function φ: <n → [0, +∞) of
a random vector (ξ1 , ξ2 , · · · , ξn ) is a function such that
Z x1 Z x2 Z xn
Φ(x1 , x2 , · · · , xn ) = ··· φ(y1 , y2 , · · · , yn )dy1 dy2 · · · dyn
−∞ −∞ −∞
holds for all (x1 , x2 , · · · , xn ) ∈ <n , where Φ is the probability distribution of

the random vector (ξ1 , ξ2 , · · · , ξn ).
1.4 Independence
Definition 1.15 The random variables ξ1 , ξ2 , · · · , ξm are said to be indepen-
dent if (m )
\ Ym
Pr {ξi ∈ Bi } = Pr{ξi ∈ Bi } (1.21)
i=1 i=1
for any Borel sets B1 , B2 , · · · , Bm of <.
Theorem 1.9 Let ξ1 , ξ2 , · · · , ξm be independent random variables, and f1 ,

f2 , · · · , fn are measurable functions. Then f1 (ξ1 ), f2 (ξ2 ), · · · , fm (ξm ) are in-
dependent random variables.
Proof: For any Borel sets of B1 , B2 , · · · , Bm of <, we have

(m ) (m )
\ \
−1
Pr {fi (ξi ) ∈ Bi } = Pr {ξi ∈ fi (Bi )}
i=1 i=1
m
Y m
Y
= Pr{ξi ∈ fi−1 (Bi )} = Pr{fi (ξi ) ∈ Bi }.
i=1 i=1
Thus f1 (ξ1 ), f2 (ξ2 ), · · · , fm (ξm ) are independent random variables.
Theorem 1.10 Let ξi be random variables with probability distributions Φi ,

i = 1, 2, · · · , m, respectively, and Φ the probability distribution of the random
vector (ξ1 , ξ2 , · · · , ξm ). Then ξ1 , ξ2 , · · · , ξm are independent if and only if
Φ(x1 , x2 , · · · , xm ) = Φ1 (x1 )Φ2 (x2 ) · · · Φm (xm ) (1.22)
for all (x1 , x2 , · · · , xm ) ∈ <m .

Proof: If ξ1 , ξ2 , · · · , ξm are independent random variables, then we have

Φ(x1 , x2 , · · · , xm ) = Pr{ξ1 ≤ x1 , ξ2 ≤ x2 , · · · , ξm ≤ xm }
= Pr{ξ1 ≤ x1 } Pr{ξ2 ≤ x2 } · · · Pr{ξm ≤ xm }
= Φ1 (x1 )Φ2 (x2 ) · · · Φm (xm )
for all (x1 , x2 , · · · , xm ) ∈ <m .
Conversely, assume that (1.22) holds. Let x2 , x3 , · · · , xm be fixed real
numbers, and C the class of all subsets C of < for which the relation
m
Y
Pr{ξ1 ∈ C, ξ2 ≤ x2 , · · · , ξm ≤ xm } = Pr{ξ1 ∈ C} Pr{ξi ≤ xi } (1.23)
i=2
holds. We will show that C contains all Borel sets of <. It follows from
the probability continuity theorem and relation (1.23) that C is a monotone
class. It is also clear that C contains all intervals of the form (−∞, a], (a, b],
(b, ∞) and ∅. Let F be the algebra consisting of all finite unions of disjoint
sets of the form (−∞, a], (a, b], (b, ∞) and ∅. Note that for any disjoint sets
C1 , C2 , · · · , Ck of F and C = C1 ∪ C2 ∪ · · · ∪ Ck , we have
Pr{ξ1 ∈ C, ξ2 ≤ x2 , · · · , ξm ≤ xm }
m
P
= Pr{ξ1 ∈ Cj , ξ2 ≤ x2 , · · · , ξm ≤ xm }
j=1
= Pr{ξ1 ∈ C} Pr{ξ2 ≤ x2 } · · · Pr{ξm ≤ xm }.
Applying the same reasoning to each ξi in turn, we obtain the indepen-
dence of the random variables.
Theorem 1.11 Let ξi be random variables with probability density functions
φi , i = 1, 2, · · · , m, respectively, and φ the probability density function of the
random vector (ξ1 , ξ2 , · · · , ξm ). Then ξ1 , ξ2 , · · · , ξm are independent if and
only if
φ(x1 , x2 , · · · , xm ) = φ1 (x1 )φ2 (x2 ) · · · φm (xm ) (1.24)
for almost all (x1 , x2 , · · · , xm ) ∈ <m .
Proof: If φ(x1 , x2 , · · · , xm ) = φ1 (x1 )φ2 (x2 ) · · · φm (xm ) a.e., then we have
Z x1 Z x2 Z xm
Φ(x1 , x2 , · · · , xm ) = ··· φ(t1 , t2 , · · · , tm )dt1 dt2 · · · dtm
−∞ −∞ −∞
Z x1 Z x2 Z xm
= ··· φ1 (t1 )φ2 (t2 ) · · · φm (tm )dt1 dt2 · · · dtm
−∞ −∞ −∞
Z x1 Z x2 Z xm
= φ1 (t1 )dt1 φ2 (t2 )dt2 · · · φm (tm )dtm
−∞ −∞ −∞
= Φ1 (x1 )Φ2 (x2 ) · · · Φm (xm )

Section 1.5 - Identical Distribution 13
for all (x1 , x2 , · · · , xm ) ∈ <m . Thus ξ1 , ξ2 , · · · , ξm are independent. Con-

versely, if ξ1 , ξ2 , · · · , ξm are independent, then for any (x1 , x2 , · · · , xm ) ∈ <m ,
we have Φ(x1 , x2 , · · · , xm ) = Φ1 (x1 )Φ2 (x2 ) · · · Φm (xm ). Hence
Z x1 Z x2 Z xm
Φ(x1 , x2 , · · · , xm ) = ··· φ1 (t1 )φ2 (t2 ) · · · φm (tm )dt1 dt2 · · · dtm
−∞ −∞ −∞
which implies that φ(x1 , x2 , · · · , xm ) = φ1 (x1 )φ2 (x2 ) · · · φm (xm ) a.e.
Example 1.11: Let ξ1 , ξ2 , · · · , ξm be independent random variables with

probability density functions φ1 , φ2 , · · · , φm , respectively, and f : <m →
< a measurable function. Then for any Borel set B of real numbers, the
probability Pr{f (ξ1 , ξ2 , · · · , ξm ) ∈ B} is
ZZ Z
··· φ1 (x1 )φ2 (x2 ) · · · φm (xm )dx1 dx2 · · · dxm .
f (x1 ,x2 ,··· ,xm )∈B
1.5 Identical Distribution

Definition 1.16 The random variables ξ and η are said to be identically
distributed if
Pr{ξ ∈ B} = Pr{ξ ∈ B} (1.25)
for any Borel set B of <.
Theorem 1.12 The random variables ξ and η are identically distributed if

and only if they have the same probability distribution.
Proof: Let Φ and Ψ be the probability distributions of ξ and η, respectively.

If ξ and η are identically distributed random variables, then, for any x ∈ <,
we have
Φ(x) = Pr{ξ ∈ (−∞, x]} = Pr{η ∈ (−∞, x]} = Ψ(x).
Thus ξ and η have the same probability distribution.

Conversely, assume that ξ and η have the same probability distribution.
Let C be the class of all subsets C of < for which the relation
Pr{ξ ∈ C} = Pr{η ∈ C} (1.26)
holds. We will show that C contains all Borel sets of <. It follows from
the probability continuity theorem and relation (1.26) that C is a monotone
class. It is also clear that C contains all intervals of the form (−∞, a], (a, b],
(b, ∞) and ∅ since ξ and η have the same probability distribution. Let F be
the algebra consisting of all finite unions of disjoint sets of the form (−∞, a],
(a, b], (b, ∞) and ∅. Note that for any disjoint sets C1 , C2 , · · · , Ck of F and
C = C1 ∪ C2 ∪ · · · ∪ Ck , we have
k
X k
X
Pr{ξ ∈ C} = Pr{ξ ∈ Cj } = Pr{η ∈ Cj } = Pr{η ∈ C}.
j=1 j=1

Theorem 1.13 Let ξ and η be two random variables whose probability den-
sity functions exist. Then ξ and η are identically distributed if and only if
they have the same probability density function.
Proof: It follows from Theorem 1.12 that the random variables ξ and η are
identically distributed if and only if they have the same probability distribu-
tion, if and only if they have the same probability density function.
1.6 Expected Value

Definition 1.17 Let ξ be a random variable. Then the expected value of ξ
is defined by
Z +∞ Z 0
E[ξ] = Pr{ξ ≥ r}dr − Pr{ξ ≤ r}dr (1.27)
0 −∞
provided that at least one of the two integrals is finite.
Example 1.12: Let ξ be a uniformly distributed random variable U(a, b).

If a ≥ 0, then Pr{ξ ≤ r} ≡ 0 when r ≤ 0, and


 1, if r ≤ a
Pr{ξ ≥ r} = (b − r)/(b − a), if a ≤ r ≤ b

0, if r ≥ b,

!
a b +∞ 0
b−r
Z Z Z Z
a+b
E[ξ] = 1dr + dr + 0dr − 0dr = .
0 a b−a b −∞ 2
If b ≤ 0, then Pr{ξ ≥ r} ≡ 0 when r ≥ 0, and



 1, if r ≥ b
Pr{ξ ≤ r} = (r − a)/(b − a), if a ≤ r ≤ b

0, if r ≤ a,

Section 1.6 - Expected Value 15
!
+∞ a b 0
r−a
Z Z Z Z
a+b
E[ξ] = 0dr − 0dr + dr + 1dr = .
0 −∞ a b−a b 2
If a < 0 < b, then

(
(b − r)/(b − a), if 0 ≤ r ≤ b
Pr{ξ ≥ r} =
0, if r ≥ b,
(
0, if r ≤ a
Pr{ξ ≤ r} =
(r − a)/(b − a), if a ≤ r ≤ 0,
!
b +∞ Z a 0
b−r r−a
Z Z Z
a+b
E[ξ] = dr + 0dr − 0dr + dr = .
0 b−a b −∞ a b−a 2
Thus we always have the expected value (a + b)/2.
Example 1.13: Let ξ be an exponentially distributed random variable

EX P(β). Then its expected value is β.
Example 1.14: Let ξ be a normally distributed random variable N (µ, σ 2 ).

Then its expected value is µ.
Example 1.15: Assume that ξ is a discrete random variable taking values

xi with probabilities pi , i = 1, 2, · · · , m, respectively. It follows from the
definition of expected value operator that
m
X
E[ξ] = pi xi .
i=1
Theorem 1.14 Let ξ be a random variable whose probability density function

φ exists. If the Lebesgue integral
Z +∞
xφ(x)dx
−∞
is finite, then we have

Z +∞
E[ξ] = xφ(x)dx. (1.28)
−∞
Proof: It follows from Definition 1.17 and Fubini Theorem that

Z +∞ Z 0
E[ξ] = Pr{ξ ≥ r}dr − Pr{ξ ≤ r}dr
0 −∞
Z +∞ Z +∞ Z 0 Z r
= φ(x)dx dr − φ(x)dx dr
0 r −∞ −∞
Z +∞ Z x Z 0 Z 0
= φ(x)dr dx − φ(x)dr dx
0 0 −∞ x
Z +∞ Z 0
= xφ(x)dx + xφ(x)dx
0 −∞
Z +∞
= xφ(x)dx.
−∞
The theorem is proved.
Theorem 1.15 Let ξ be a random variable with probability distribution Φ.

If the Lebesgue-Stieltjes integral
Z +∞
xdΦ(x)
−∞
is finite, then we have Z +∞

E[ξ] = xdΦ(x). (1.29)
−∞
R +∞
Proof: Since the Lebesgue-Stieltjes integral −∞ xdΦ(x) is finite, we imme-
diately have
Z y Z +∞ Z 0 Z 0
lim xdΦ(x) = xdΦ(x), lim xdΦ(x) = xdΦ(x)
y→+∞ 0 0 y→−∞ y −∞
and Z +∞ Z y
lim xdΦ(x) = 0, lim xdΦ(x) = 0.
y→+∞ y y→−∞ −∞
It follows from
Z +∞
xdΦ(x) ≥ y lim Φ(z) − Φ(y) = y(1 − Φ(y)) ≥ 0, if y > 0,
y z→+∞
Z y
xdΦ(x) ≤ y Φ(y) − lim Φ(z) = yΦ(y) ≤ 0, if y < 0
−∞ z→−∞
that
lim y (1 − Φ(y)) = 0, lim yΦ(y) = 0.
y→+∞ y→−∞
Let 0 = x0 < x1 < x2 < · · · < xn = y be a partition of [0, y]. Then we have
n−1
X Z y
xi (Φ(xi+1 ) − Φ(xi )) → xdΦ(x)
i=0 0
and
n−1
X Z y
(1 − Φ(xi+1 ))(xi+1 − xi ) → Pr{ξ ≥ r}dr
i=0 0
as max{|xi+1 − xi | : i = 0, 1, · · · , n − 1} → 0. Since
n−1
X n−1
X
xi (Φ(xi+1 ) − Φ(xi )) − (1 − Φ(xi+1 ))(xi+1 − xi ) = y(Φ(y) − 1) → 0
i=0 i=0
as y → +∞. This fact implies that

Z +∞ Z +∞
Pr{ξ ≥ r}dr = xdΦ(x).
0 0
A similar way may prove that

Z 0 Z 0
− Pr{ξ ≤ r}dr = xdΦ(x).
−∞ −∞
Thus (1.29) is verified by the above two equations.
Linearity of Expected Value Operator

Theorem 1.16 Let ξ and η be random variables with finite expected values.
Then for any numbers a and b, we have
E[aξ + bη] = aE[ξ] + bE[η]. (1.30)
Proof: Step 1: We first prove that E[ξ + b] = E[ξ] + b for any real number
b. When b ≥ 0, we have
Z ∞ Z 0
E[ξ + b] = Pr{ξ + b ≥ r}dr − Pr{ξ + b ≤ r}dr
0 −∞
Z ∞ Z 0
= Pr{ξ ≥ r − b}dr − Pr{ξ ≤ r − b}dr
0 −∞
Z b
= E[ξ] + (Pr{ξ ≥ r − b} + Pr{ξ < r − b}) dr
0
= E[ξ] + b.
If b < 0, then we have
Z 0
E[ξ + b] = E[ξ] − (Pr{ξ ≥ r − b} + Pr{ξ < r − b}) dr = E[ξ] + b.
b
Step 2: We prove that E[aξ] = aE[ξ] for any real number a. If a = 0,

then the equation E[aξ] = aE[ξ] holds trivially. If a > 0, we have
Z ∞ Z 0
E[aξ] = Pr{aξ ≥ r}dr − Pr{aξ ≤ r}dr
0 −∞
Z ∞ Z 0
n ro n ro
= Pr ξ ≥ dr − Pr ξ ≤ dr
0 a −∞ a
Z ∞ n Z 0
ro r n ro r
=a Pr ξ ≥ d −a Pr ξ ≤ d
0 a a −∞ a a
= aE[ξ].
If a < 0, we have
Z ∞ Z 0
E[aξ] = Pr{aξ ≥ r}dr − Pr{aξ ≤ r}dr
0 −∞
Z ∞ Z 0
n ro n ro
= Pr ξ ≤ dr − Pr ξ ≥ dr
0 a −∞ a
Z ∞ n Z 0
ro r n ro r
=a Pr ξ ≥ d −a Pr ξ ≤ d
0 a a −∞ a a
= aE[ξ].
Step 3: We prove that E[ξ + η] = E[ξ] + E[η] when both ξ and η

are nonnegative simple random variables taking values a1 , a2 , · · · , am and
b1 , b2 , · · · , bn , respectively. Then ξ + η is also a nonnegative simple random
variable taking values ai + bj , i = 1, 2, · · · , m, j = 1, 2, · · · , n. Thus we have
m P
P n
E[ξ + η] = (ai + bj ) Pr{ξ = ai , η = bj }
i=1 j=1
Pm P n m P
P n
= ai Pr{ξ = ai , η = bj } + bj Pr{ξ = ai , η = bj }
i=1 j=1 i=1 j=1
Pm n
P
= ai Pr{ξ = ai } + bj Pr{η = bj }
i=1 j=1
= E[ξ] + E[η].
Step 4: We prove that E[ξ + η] = E[ξ] + E[η] when both ξ and η are
nonnegative random variables. For every i ≥ 1 and every ω ∈ Ω, we define

 k − 1 , if k − 1 ≤ ξ(ω) < k , k = 1, 2, · · · , i2i

ξi (ω) = 2i 2i 2i

 i, if i ≤ ξ(ω),

 k − 1 , if k − 1 ≤ η(ω) < k , k = 1, 2, · · · , i2i

ηi (ω) = 2i 2i 2i

 i, if i ≤ η(ω).
Then {ξi }, {ηi } and {ξi + ηi } are three sequences of nonnegative simple
random variables such that ξi ↑ ξ, ηi ↑ η and ξi + ηi ↑ ξ + η as i → ∞. Note
that the functions Pr{ξi > r}, Pr{ηi > r}, Pr{ξi + ηi > r}, i = 1, 2, · · · are
also simple. It follows from the probability continuity theorem that
Pr{ξi > r} ↑ Pr{ξ > r}, ∀r ≥ 0
as i → ∞. Since the expected value E[ξ] exists, we have

Z +∞ Z +∞
E[ξi ] = Pr{ξi > r}dr → Pr{ξ > r}dr = E[ξ]
0 0
as i → ∞. Similarly, we may prove that E[ηi ] → E[η] and E[ξi +ηi ] → E[ξ+η]
as i → ∞. It follows from Step 3 that E[ξ + η] = E[ξ] + E[η].
Step 5: We prove that E[ξ + η] = E[ξ] + E[η] when ξ and η are arbitrary
random variables. Define
( (
ξ(ω), if ξ(ω) ≥ −i η(ω), if η(ω) ≥ −i
ξi (ω) = ηi (ω) =
−i, otherwise, −i, otherwise.
Since the expected values E[ξ] and E[η] are finite, we have
lim E[ξi ] = E[ξ], lim E[ηi ] = E[η], lim E[ξi + ηi ] = E[ξ + η].
i→∞ i→∞ i→∞
Note that (ξi + i) and (ηi + i) are nonnegative random variables. It follows
from Steps 1 and 4 that
E[ξ + η] = lim E[ξi + ηi ]

i→∞
= lim (E[(ξi + i) + (ηi + i)] − 2i)
i→∞
= lim (E[ξi + i] + E[ηi + i] − 2i)
i→∞
= lim (E[ξi ] + i + E[ηi ] + i − 2i)
i→∞
= lim E[ξi ] + lim E[ηi ]
i→∞ i→∞
= E[ξ] + E[η].
Step 6: The linearity E[aξ + bη] = aE[ξ] + bE[η] follows immediately

from Steps 2 and 5. The theorem is proved.
Product of Independent Random Variables

Theorem 1.17 Let ξ and η be independent random variables with finite ex-
pected values. Then the expected value of ξη exists and
E[ξη] = E[ξ]E[η]. (1.31)
Proof: Step 1: We first prove the case where both ξ and η are nonnega-
tive simple random variables taking values a1 , a2 , · · · , am and b1 , b2 , · · · , bn ,
respectively. Then ξη is also a nonnegative simple random variable taking
values ai bj , i = 1, 2, · · · , m, j = 1, 2, · · · , n. It follows from the independence
of ξ and η that
m P
P n
E[ξη] = ai bj Pr{ξ = ai , η = bj }
i=1 j=1
Pm P n
= ai bj Pr{ξ = ai } Pr{η = bj }
i=1 j=1
m !
P n
P
= ai Pr{ξ = ai } bj Pr{η = bj }
i=1 j=1
= E[ξ]E[η].
Step 2: Next we prove the case where ξ and η are nonnegative random
variables. For every i ≥ 1 and every ω ∈ Ω, we define

 k − 1 , if k − 1 ≤ ξ(ω) < k , k = 1, 2, · · · , i2i

ξi (ω) = 2i 2i 2i

 i, if i ≤ ξ(ω),

 k − 1 , if k − 1 ≤ η(ω) < k , k = 1, 2, · · · , i2i

ηi (ω) = 2i 2i 2i

 i, if i ≤ η(ω).
Then {ξi }, {ηi } and {ξi ηi } are three sequences of nonnegative simple random
variables such that ξi ↑ ξ, ηi ↑ η and ξi ηi ↑ ξη as i → ∞. It follows from
the independence of ξ and η that ξi and ηi are independent. Hence we have
E[ξi ηi ] = E[ξi ]E[ηi ] for i = 1, 2, · · · It follows from the probability continuity
theorem that Pr{ξi > r}, i = 1, 2, · · · are simple functions such that
Pr{ξi > r} ↑ Pr{ξ > r}, for all r ≥ 0
as i → ∞. Since the expected value E[ξ] exists, we have
Z +∞ Z +∞
E[ξi ] = Pr{ξi > r}dr → Pr{ξ > r}dr = E[ξ]
0 0
as i → ∞. Similarly, we may prove that E[ηi ] → E[η] and E[ξi ηi ] → E[ξη]

as i → ∞. Therefore E[ξη] = E[ξ]E[η].
Step 3: Finally, if ξ and η are arbitrary independent random variables,

then the nonnegative random variables ξ + and η + are independent and so
are ξ + and η − , ξ − and η + , ξ − and η − . Thus we have
E[ξ + η + ] = E[ξ + ]E[η + ], E[ξ + η − ] = E[ξ + ]E[η − ],
E[ξ − η + ] = E[ξ − ]E[η + ], E[ξ − η − ] = E[ξ − ]E[η − ].
It follows that
E[ξη] = E[(ξ + − ξ − )(η + − η − )]
= E[ξ + η + ] − E[ξ + η − ] − E[ξ − η + ] + E[ξ − η − ]
= E[ξ + ]E[η + ] − E[ξ + ]E[η − ] − E[ξ − ]E[η + ] + E[ξ − ]E[η − ]
= (E[ξ + ] − E[ξ − ]) (E[η + ] − E[η − ])
= E[ξ + − ξ − ]E[η + − η − ]
= E[ξ]E[η]
which proves the theorem.
Expected Value of Function of Random Variable

Theorem 1.18 Let ξ be a random variable with probability distribution Φ,
and f : < → < a measurable function. If the Lebesgue-Stieltjes integral
Z +∞
f (x)dΦ(x)
−∞

Z +∞
E[f (ξ)] = f (x)dΦ(x). (1.32)
−∞
Proof: It follows from the definition of expected value operator that

Z +∞ Z 0
E[f (ξ)] = Pr{f (ξ) ≥ r}dr − Pr{f (ξ) ≤ r}dr. (1.33)
0 −∞
If f is a nonnegative simple measurable function, i.e.,

if x ∈ B1

 a1 ,

if x ∈ B2

f (x) = a2 ,

 ···
if x ∈ Bm

am ,
where B1 , B2 , · · · , Bm are mutually disjoint Borel sets, then we have
Z +∞ Xm
E[f (ξ)] = Pr{f (ξ) ≥ r}dr = ai Pr{ξ ∈ Bi }
0 i=1
m
X Z Z +∞
= ai dΦ(x) = f (x)dΦ(x).
i=1 Bi −∞
We next prove the case where f is a nonnegative measurable function. Let

f1 , f2 , · · · be a sequence of nonnegative simple functions such that fi ↑ f as
i → ∞. We have proved that
Z +∞ Z +∞
E[fi (ξ)] = Pr{fi (ξ) ≥ r}dr = fi (x)dΦ(x).
0 −∞
In addition, it is also known that Pr{fi (ξ) > r} ↑ Pr{f (ξ) > r} as i → ∞ for
r ≥ 0. It follows from the monotone convergence theorem that
Z +∞
E[f (ξ)] = Pr{f (ξ) > r}dr
0
Z +∞
= lim Pr{fi (ξ) > r}dr
i→∞ 0
Z +∞
= lim fi (x)dΦ(x)
i→∞ −∞
Z +∞
= f (x)dΦ(x).
−∞
Finally, if f is an arbitrary measurable function, then we have f = f + − f −

and
E[f (ξ)] = E[f + (ξ) − f − (ξ)] = E[f + (ξ)] − E[f − (ξ)]
Z +∞ Z +∞
= +
f (x)dΦ(x) − f − (x)dΦ(x)
−∞ −∞
Z +∞
= f (x)dΦ(x).
−∞
Sum of a Random Number of Random Variables

Theorem 1.19 (Wald Identity) Assume that {ξi } is a sequence of iid ran-
dom variables, and η is a positive random integer (i.e., a random variable
taking “positive integer” values) that is independent of the sequence {ξi }.
Then we have " η #
X
E ξi = E[η]E[ξ1 ]. (1.34)
i=1
Proof: Since η is independent of the sequence {ξi }, we have

( η ) ∞
X X
Pr ξi ≥ r = Pr{η = k} Pr {ξ1 + ξ2 + · · · + ξk ≥ r} .
i=1 k=1
Section 1.7 - Variance 23
If ξi are nonnegative random variables, then we have

" η # Z ( η )
X +∞ X
E ξi = Pr ξi ≥ r dr
i=1 0 i=1
Z ∞
+∞ X
= Pr{η = k} Pr {ξ1 + ξ2 + · · · + ξk ≥ r} dr
0 k=1
∞
X Z +∞
= Pr{η = k} Pr {ξ1 + ξ2 + · · · + ξk ≥ r} dr
k=1 0
X∞
= Pr{η = k} (E[ξ1 ] + E[ξ2 ] + · · · + E[ξk ])
k=1
X∞
= Pr{η = k}kE[ξ1 ] (by iid hypothesis)
k=1
= E[η]E[ξ1 ].
If ξi are arbitrary random variables, then ξi = ξi+ − ξi− , and

" η # " η # " η η
#
X X X X
+ − + −
E ξi = E (ξi − ξi ) = E ξi − ξi
i=1 i=1 i=1 i=1
η η
" # " #
X X
=E ξi+ − E ξi− = E[η]E[ξ1+ ] − E[η]E[ξ1− ]
i=1 i=1
= E[η](E[ξ1+ ] − E[ξ1− ]) = E[η]E[ξ1+ − ξ1− ] = E[η]E[ξ1 ].
The theorem is thus proved.
1.7 Variance
Definition 1.18 Let ξ be a random variable with finite expected value e.
Then the variance of ξ is defined by V [ξ] = E[(ξ − e)2 ].
The variance of a random variable provides a measure of the spread of the

distribution around its expected value. A small value of variance indicates
that the random variable is tightly concentrated around its expected value;
and a large value of variance indicates that the random variable has a wide
spread around its expected value.
Example 1.16: Let ξ be a uniformly distributed random variable U(a, b).

Then its expected value is (a + b)/2 and variance is (b − a)2 /12.
Example 1.17: Let ξ be an exponentially distributed random variable

EX P(β). Then its expected value is β and variance is β 2 .
Example 1.18: Let ξ be a normally distributed random variable N (µ, σ 2 ).

Then its expected value is µ and variance is σ 2 .
Theorem 1.20 If ξ is a random variable whose variance exists, a and b are

real numbers, then V [aξ + b] = a2 V [ξ].
Proof: It follows from the definition of variance that
V [aξ + b] = E (aξ + b − aE[ξ] − b)2 = a2 E[(ξ − E[ξ])2 ] = a2 V [ξ].

Theorem 1.21 Let ξ be a random variable with expected value e. Then

V [ξ] = 0 if and only if Pr{ξ = e} = 1.
Proof: If V [ξ] = 0, then E[(ξ − e)2 ] = 0. Thus we have

Z +∞
Pr{(ξ − e)2 ≥ r}dr = 0
0
which implies Pr{(ξ−e) ≥ r} = 0 for any r > 0. Hence we have Pr{(ξ−e)2 =

2
0} = 1, i.e., Pr{ξ = e} = 1.
Conversely, if Pr{ξ = e} = 1, then we have Pr{(ξ − e)2 = 0} = 1 and
Pr{(ξ − e)2 ≥ r} = 0 for any r > 0. Thus
Z +∞
V [ξ] = Pr{(ξ − e)2 ≥ r}dr = 0.
0
Theorem 1.22 If ξ1 , ξ2 , · · · , ξn are independent random variables with finite

expected values, then
V [ξ1 + ξ2 + · · · + ξn ] = V [ξ1 ] + V [ξ2 ] + · · · + V [ξn ]. (1.35)

n

ξi = E (ξ1 + ξ2 + · · · + ξn − E[ξ1 ] − E[ξ2 ] − · · · − E[ξn ])2
P
V
i=1
n n−1 n
E (ξi − E[ξi ])2 + 2
P P P
= E [(ξi − E[ξi ])(ξj − E[ξj ])] .
i=1 i=1 j=i+1
Since ξ1 , ξ2 , · · · , ξn are independent, E [(ξi − E[ξi ])(ξj − E[ξj ])] = 0 for all
i, j with i 6= j. Thus (1.35) holds.
Maximum Variance Theorem

Let ξ be a random variable that takes values in [a, b], but whose probability
distribution is otherwise arbitrary. If its expected value is given, what is the
possible maximum variance? The maximum variance theorem will answer
this question, thus playing an important role in treating games against nature.
Section 1.8 - Moments 25
Theorem 1.23 (Edmundson-Madansky Inequality) Let f be a convex func-

tion on [a, b], and ξ a random variable that takes values in [a, b] and has
expected value e. Then
b−e e−a
E[f (ξ)] ≤ f (a) + f (b). (1.36)
b−a b−a
Proof: For each ω ∈ Ω, we have a ≤ ξ(ω) ≤ b and
b − ξ(ω) ξ(ω) − a
ξ(ω) = a+ b.
b−a b−a
It follows from the convexity of f that
b − ξ(ω) ξ(ω) − a
f (ξ(ω)) ≤ f (a) + f (b).
b−a b−a
Taking expected values on both sides, we obtain (1.36).
Theorem 1.24 (Maximum Variance Theorem) Let ξ be a random variable
that takes values in [a, b] and has expected value e. Then
V [ξ] ≤ (e − a)(b − e) (1.37)
and equality holds if the random variable ξ is determined by
b−e


 , if x = a
b−a

Pr{ξ = x} = (1.38)
 e − a , if x = b.


b−a
Proof: It follows from Theorem 1.23 immediately by defining f (x) = (x−e)2 .
It is also easy to verify that the random variable determined by (1.38) has
variance (e − a)(b − e). The theorem is proved.
1.8 Moments
Definition 1.19 Let ξ be a random variable, and k a positive number. Then
(a) the expected value E[ξ k ] is called the kth moment;
(b) the expected value E[|ξ|k ] is called the kth absolute moment;
(c) the expected value E[(ξ − E[ξ])k ] is called the kth central moment;
(d) the expected value E[|ξ −E[ξ]|k ] is called the kth absolute central moment.
Note that the first central moment is always 0, the first moment is just
the expected value, and the second central moment is just the variance.
Theorem 1.25 Let ξ be a nonnegative random variable, and k a positive
number. Then the k-th moment
Z +∞
k
E[ξ ] = k rk−1 Pr{ξ ≥ r}dr. (1.39)
0
Proof: It follows from the nonnegativity of ξ that

Z ∞ Z ∞ Z ∞
E[ξ k ] = Pr{ξ k ≥ x}dx = Pr{ξ ≥ r}drk = k rk−1 Pr{ξ ≥ r}dr.
0 0 0
Theorem 1.26 Let ξ be a random variable that takes values in [a, b] and has
expected value e. Then for any positive integer k, the kth absolute moment
and kth absolute central moment satisfy the following inequalities,
b−e k e−a k
E[|ξ|k ] ≤ |a| + |b| , (1.40)
b−a b−a
b−e e−a
E[|ξ − e|k ] ≤ (e − a)k + (b − e)k . (1.41)
b−a b−a
Proof: It follows from Theorem 1.23 immediately by defining f (x) = |x|k

and f (x) = |x − e|k .
1.9 Critical Values

Let ξ be a random variable. In order to measure it, we may use its expected
value. Alternately, we may employ α-optimistic value and α-pessimistic value
as a ranking measure.
Definition 1.20 Let ξ be a random variable, and α ∈ (0, 1]. Then

ξsup (α) = sup r Pr {ξ ≥ r} ≥ α (1.42)
is called the α-optimistic value of ξ, and

ξinf (α) = inf r Pr {ξ ≤ r} ≥ α (1.43)
is called the α-pessimistic value of ξ.
This means that the random variable ξ will reach upwards of the α-
optimistic value ξsup (α) at least α of time, and will be below the α-pessimistic
value ξinf (α) at least α of time. The optimistic value is also called percentile.
Theorem 1.27 Let ξ be a random variable. Then we have
Pr{ξ ≥ ξsup (α)} ≥ α, Pr{ξ ≤ ξinf (α)} ≥ α (1.44)
where ξsup (α) and ξinf (α) are the α-optimistic and α-pessimistic values of the
random variable ξ, respectively.
Section 1.9 - Critical Values 27
Proof: It follows from the definition of the optimistic value that there exists
an increasing sequence {ri } such that Pr{ξ ≥ ri } ≥ α and ri ↑ ξsup (α) as
i → ∞. Since {ω|ξ(ω) ≥ ri } ↓ {ω|ξ(ω) ≥ ξsup (α)}, it follows from the
probability continuity theorem that
Pr{ξ ≥ ξsup (α)} = lim Pr{ξ ≥ ri } ≥ α.
i→∞
The inequality Pr{ξ ≤ ξinf (α)} ≥ α may be proved similarly.
Example 1.19: Note that Pr{ξ ≥ ξsup (α)} > α and Pr{ξ ≤ ξinf (α)} > α
may hold. For example,
(
0 with probability 0.4
ξ=
1 with probability 0.6.
If α = 0.8, then ξsup (0.8) = 0 which makes Pr{ξ ≥ ξsup (0.8)} = 1 > 0.8. In
addition, ξinf (0.8) = 1 and Pr{ξ ≤ ξinf (0.8)} = 1 > 0.8.
(a) ξinf (α) is an increasing and left-continuous function of α;
(b) ξsup (α) is a decreasing and left-continuous function of α.
Proof: (a) It is easy to prove that ξinf (α) is an increasing function of α.
Next, we prove the left-continuity of ξinf (α) with respect to α. Let {αi } be
an arbitrary sequence of positive numbers such that αi ↑ α. Then {ξinf (αi )}
is an increasing sequence. If the limitation is equal to ξinf (α), then the left-
continuity is proved. Otherwise, there exists a number z ∗ such that
lim ξinf (αi ) < z ∗ < ξinf (α).
i→∞
Thus Pr{ξ ≤ z ∗ } ≥ αi for each i. Letting i → ∞, we get Pr{ξ ≤ z ∗ } ≥ α.

Hence z ∗ ≥ ξinf (α). A contradiction proves the left-continuity of ξinf (α) with
respect to α. The part (b) may be proved similarly.
(a) if α > 0.5, then ξinf (α) ≥ ξsup (α);
(b) if α ≤ 0.5, then ξinf (α) ≤ ξsup (α).
Proof: Part (a): Write ξ(α) = (ξinf (α) + ξsup (α))/2. If ξinf (α) < ξsup (α),
then we have
1 ≥ Pr{ξ < ξ(α)} + Pr{ξ > ξ(α)} ≥ α + α > 1.
A contradiction proves ξinf (α) ≥ ξsup (α). Part (b): Assume that ξinf (α) >
ξsup (α). It follows from the definition of ξinf (α) that Pr{ξ ≤ ξ(α)} < α.
Similarly, it follows from the definition of ξsup (α) that Pr{ξ ≥ ξ(α)} < α.
Thus
1 ≤ Pr{ξ ≤ ξ(α)} + Pr{ξ ≥ ξ(α)} < α + α ≤ 1.
A contradiction proves ξinf (α) ≤ ξsup (α). The theorem is proved.

(a) if c ≥ 0, then (cξ)sup (α) = cξsup (α) and (cξ)inf (α) = cξinf (α);
(b) if c < 0, then (cξ)sup (α) = cξinf (α) and (cξ)inf (α) = cξsup (α).
Proof: (a) If c = 0, then it is obviously valid. When c > 0, we have
(cξ)sup (α) = sup {r | Pr{cξ ≥ r} ≥ α}

= c sup {r/c | Pr {ξ ≥ r/c} ≥ α}
= cξsup (α).
A similar way may prove that (cξ)inf (α) = cξinf (α).

(b) In order to prove this part, it suffices to verify that (−ξ)sup (α) =
−ξinf (α) and (−ξ)inf (α) = −ξsup (α). In fact, for any α ∈ (0, 1], we have
(−ξ)sup (α) = sup{r | Pr{−ξ ≥ r} ≥ α}

= − inf{−r | Pr{ξ ≤ −r} ≥ α}
= −ξinf (α).
Similarly, we may prove that (−ξ)inf (α) = −ξsup (α). The theorem is proved.
1.10 Entropy
Given a random variable, what is the degree of difficulty of predicting the
specified value that the random variable will take? In order to answer this
question, Shannon [219] defined a concept of entropy as a measure of uncer-
tainty.
Entropy of Discrete Random Variables

Definition 1.21 Let ξ be a discrete random variable taking values xi with
probabilities pi , i = 1, 2, · · · , respectively. Then its entropy is defined by
∞
X
H[ξ] = − pi ln pi . (1.45)
i=1
It should be noticed that the entropy depends only on the number of

values and their probabilities and does not depend on the actual values that
the random variable takes.
Theorem 1.31 Let ξ be a discrete random variable taking values xi with

probabilities pi , i = 1, 2, · · · , respectively. Then
H[ξ] ≥ 0 (1.46)
and equality holds if and only if there exists an index k such that pk = 1, i.e.,
ξ is essentially a deterministic number.
Section 1.10 - Entropy 29
S(t)
....
.........
....
...
......................
1/e ... .....
.........
.......... ... ..................
.......
... ....... .. ......
... ... ... ......
... ... .. .....
... .....
...... .. .....
...
...
.
........................................................................................................................................................................
....
..... t
0 ...1/e
...
1 .....
.....
....
....
... ....
....
... ....
... ....
....
... ....
... ...
...
... ...
.. ...
Figure 1.1: Function S(t) = −t ln t is concave
Proof: The nonnegativity is clear. In addition, H[ξ] = 0 if and only if

pi = 0 or 1 for each i. That is, there exists one and only one index k such
that pk = 1. The theorem is proved.
This theorem states that the entropy of a discrete random variable reaches
its minimum 0 when the random variable degenerates to a deterministic num-
ber. In this case, there is no uncertainty.
Theorem 1.32 Let ξ be a simple random variable taking values xi with prob-
abilities pi , i = 1, 2, · · · , n, respectively. Then
H[ξ] ≤ ln n (1.47)
and equality holds if and only if pi ≡ 1/n for all i = 1, 2, · · · , n.
Proof: Since the function S(t) is a concave function of t and p1 + p2 + · · · +

pn = 1, we have
n n
! n
!
X 1X 1X
− pi ln pi ≤ −n pi ln pi = ln n
i=1
n i=1 n i=1
which implies that H[ξ] ≤ ln n and equality holds if and only if p1 = p2 =

· · · = pn , i.e., pi ≡ 1/n for all i = 1, 2, · · · , n.
This theorem states that the entropy of a simple random variable reaches
its maximum ln n when all outcomes are equiprobable. In this case, there is
no preference among all the values that the random variable will take.
Entropy of Absolutely Continuous Random Variables

Definition 1.22 Let ξ be a random variable with probability density function
φ. Then its entropy is defined by
Z +∞
H[ξ] = − φ(x) ln φ(x)dx. (1.48)
−∞
Example 1.20: Let ξ be a uniformly distributed random variable on [a, b].

Then its entropy is H[ξ] = ln(b − a). This example shows that the entropy of
absolutely continuous random variable may assume both positive and nega-
tive values since ln(b − a) < 0 if b − a < 1; and ln(b − a) > 0 if b − a > 1.
Example 1.21: Let ξ be an exponentially distributed random variable with

expected value β. Then its entropy is H[ξ] = 1 + ln β.
Example 1.22: Let ξ be a normally distributed random variable with √ ex-

pected value e and variance σ 2 . Then its entropy is H[ξ] = 1/2 + ln 2πσ.
Maximum Entropy Principle

Given some constraints, for example, expected value and variance, there are
usually multiple compatible probability distributions. For this case, we would
like to select the distribution that maximizes the value of entropy and satisfies
the prescribed constraints. This method is often referred to as the maximum
entropy principle (Jaynes [68]).
Example 1.23: Let ξ be an absolutely continuous random variable on [a, b].

The maximum entropy principle attempts to find the probability density
function φ(x) that maximizes the entropy
Z b
− φ(x) ln φ(x)dx
a
Rb
subject to the natural constraint a φ(x)dx = 1. The Lagrangian is
Z b Z b !
L=− φ(x) ln φ(x)dx − λ φ(x)dx − 1 .
a a
It follows from the Euler-Lagrange equation that the maximum entropy prob-
ability density function meets
ln φ(x) + 1 + λ = 0
and has the form φ(x) = exp(−1 − λ). Substituting it into the natural
constraint, we get
1
φ∗ (x) = , a≤x≤b
b−a
which is just the uniformly distributed random variable, and the maximum
entropy is H[ξ ∗ ] = ln(b − a).
Example 1.24: Let ξ be an absolutely continuous random variable on [0, ∞).

Assume that the expected value of ξ is prescribed to be β. The maximum
entropy probability density function φ(x) should maximize the entropy
Z +∞
0
subject to the constraints

Z +∞ Z +∞
φ(x)dx = 1, xφ(x)dx = β.
0 0
The Lagrangian is
Z ∞ Z ∞ Z ∞
L=− φ(x) ln φ(x)dx − λ1 φ(x)dx − 1 − λ2 xφ(x)dx − β .
0 0 0
The maximum entropy probability density function meets Euler-Lagrange

equation
ln φ(x) + 1 + λ1 + λ2 x = 0
and has the form φ(x) = exp(−1 − λ1 − λ2 x). Substituting it into the
constraints, we get

1 x
φ∗ (x) = exp − , x≥0
β β
which is just the exponentially distributed random variable, and the maxi-
mum entropy is H[ξ ∗ ] = 1 + ln β.
Example 1.25: Let ξ be an absolutely continuous random variable on

(−∞, +∞). Assume that the expected value and variance of ξ are prescribed
to be µ and σ 2 , respectively. The maximum entropy probability density
function φ(x) should maximize the entropy
Z +∞
−∞
subject to the constraints

Z +∞ Z +∞ Z +∞
φ(x)dx = 1, xφ(x)dx = µ, (x − µ)2 φ(x)dx = σ 2 .
−∞ −∞ −∞
The Lagrangian is
Z +∞ Z +∞
L=− φ(x) ln φ(x)dx − λ1 φ(x)dx − 1
−∞ −∞
Z +∞ Z +∞
2 2
−λ2 xφ(x)dx − µ − λ3 (x − µ) φ(x)dx − σ .
−∞ −∞
The maximum entropy probability density function meets Euler-Lagrange

equation
ln φ(x) + 1 + λ1 + λ2 x + λ3 (x − µ)2 = 0
and has the form φ(x) = exp(−1 − λ1 − λ2 x − λ3 (x − µ)2 ). Substituting it

into the constraints, we get
(x − µ)2

∗ 1
φ (x) = √ exp − , x∈<
σ 2π 2σ 2
which is just the normally distributed

√ random variable, and the maximum
entropy is H[ξ ∗ ] = 1/2 + ln 2πσ.
1.11 Distance
Distance is a powerful concept in many disciplines of science and engineering.
This section introduces the distance between random variables.
Definition 1.23 The distance between random variables ξ and η is defined

as
d(ξ, η) = E[|ξ − η|]. (1.49)
Theorem 1.33 Let ξ, η, τ be random variables, and let d(·, ·) be the distance.
Then we have
(a) (Nonnegativity) d(ξ, η) ≥ 0;
(b) (Identification) d(ξ, η) = 0 if and only if ξ = η;
(c) (Symmetry) d(ξ, η) = d(η, ξ);
(d) (Triangle Inequality) d(ξ, η) ≤ d(ξ, τ ) + d(η, τ ).
Proof: The parts (a), (b) and (c) follow immediately from the definition.
The part (d) is proved by the following relation,
E[|ξ − η|] ≤ E[|ξ − τ | + |η − τ |] = E[|ξ − τ |] + E[|η − τ |].
1.12 Inequalities
It is well-known that there are several inequalities in probability theory, such
as Markov inequality, Chebyshev inequality, Hölder’s inequality, Minkowski
inequality, and Jensen’s inequality. They play an important role in both
theory and applications.
Theorem 1.34 Let ξ be a random variable, and f a nonnegative measurable

function. If f is even (i.e., f (x) = f (−x) for any x ∈ <) and increasing on
[0, ∞), then for any given number t > 0, we have
E[f (ξ)]
Pr{|ξ| ≥ t} ≤ . (1.50)
f (t)
Section 1.12 - Inequalities 33
Proof: It is clear that Pr{|ξ| ≥ f −1 (r)} is a monotone decreasing function

of r on [0, ∞). It follows from the nonnegativity of f (ξ) that
Z +∞
E[f (ξ)] = Pr{f (ξ) ≥ r}dr
0
Z +∞
= Pr{|ξ| ≥ f −1 (r)}dr
0
Z f (t)
≥ Pr{|ξ| ≥ f −1 (r)}dr
0
Z f (t)
≥ dr · Pr{|ξ| ≥ f −1 (f (t))}
0
= f (t) · Pr{|ξ| ≥ t}
which proves the inequality.
Theorem 1.35 (Markov Inequality) Let ξ be a random variable. Then for

any given numbers t > 0 and p > 0, we have
E[|ξ|p ]
Pr{|ξ| ≥ t} ≤ . (1.51)
tp
Proof: It is a special case of Theorem 1.34 when f (x) = |x|p .
Example 1.26: For any given positive number t, we define a random variable
as follows, (
0 with probability 1/2
ξ=
t with probability 1/2.
Then Pr{ξ ≥ t} = 1/2 = E[|ξ|p ]/tp .
Theorem 1.36 (Chebyshev Inequality) Let ξ be a random variable whose

variance V [ξ] exists. Then for any given number t > 0, we have
V [ξ]
Pr {|ξ − E[ξ]| ≥ t} ≤ . (1.52)
t2
Proof: It is a special case of Theorem 1.34 when the random variable ξ is
replaced with ξ − E[ξ] and f (x) = x2 .
Example 1.27: For any given positive number t, we define a random variable
as follows, (
−t with probability 1/2
ξ=
t with probability 1/2.
Then Pr{|ξ − E[ξ]| ≥ t} = 1 = V [ξ]/t2 .

Theorem 1.37 (Hölder’s Inequality) Let p and q be two positive real num-
bers with 1/p+1/q = 1, and let ξ and η be random variables with E[|ξ|p ] < ∞
and E[|η|q ] < ∞. Then we have
p
p
p
E[|ξη|] ≤ E[|ξ|p ] q E[|η|q ]. (1.53)
Proof: The inequality holds trivially if at least one of ξ and η is zero a.s. Now
we assume√E[|ξ|p ] > 0 and E[|η|q ] > 0. It is easy to prove that the function
√
f (x, y) = p x q y is a concave function on D = {(x, y) : x ≥ 0, y ≥ 0}. Thus
for any point (x0 , y0 ) with x0 > 0 and y0 > 0, there exist two real numbers
a and b such that
f (x, y) − f (x0 , y0 ) ≤ a(x − x0 ) + b(y − y0 ), ∀(x, y) ∈ D.
Letting x0 = E[|ξ|p ], y0 = E[|η|q ], x = |ξ|p and y = |η|q , we have
f (|ξ|p , |η|q ) − f (E[|ξ|p ], E[|η|q ]) ≤ a(|ξ|p − E[|ξ|p ]) + b(|η|q − E[|η|q ]).
Taking the expected values on both sides, we obtain
E[f (|ξ|p , |η|q )] ≤ f (E[|ξ|p ], E[|η|q ]).
Hence the inequality (1.53) holds.
Theorem 1.38 (Minkowski Inequality) Let p be a real number with p ≥ 1,

and let ξ and η be random variables with E[|ξ|p ] < ∞ and E[|η|p ] < ∞. Then
we have p p p
p
E[|ξ + η|p ] ≤ p E[|ξ|p ] + p E[|η|p ]. (1.54)
we assume √ E[|ξ|p ] > 0 and E[|η|p ] > 0. It is easy to prove that the function
√
f (x, y) = ( p x + p y)p is a concave function on D = {(x, y) : x ≥ 0, y ≥ 0}.
Thus for any point (x0 , y0 ) with x0 > 0 and y0 > 0, there exist two real
numbers a and b such that
Letting x0 = E[|ξ|p ], y0 = E[|η|p ], x = |ξ|p and y = |η|p , we have
f (|ξ|p , |η|p ) − f (E[|ξ|p ], E[|η|p ]) ≤ a(|ξ|p − E[|ξ|p ]) + b(|η|p − E[|η|p ]).
E[f (|ξ|p , |η|p )] ≤ f (E[|ξ|p ], E[|η|p ]).

Section 1.13 - Convergence Concepts 35
Theorem 1.39 (Jensen’s Inequality) Let ξ be a random variable, and f :

< → < a convex function. If E[ξ] and E[f (ξ)] are finite, then
f (E[ξ]) ≤ E[f (ξ)]. (1.55)
Especially, when f (x) = |x|p and p ≥ 1, we have |E[ξ]|p ≤ E[|ξ|p ].
Proof: Since f is a convex function, for each y, there exists a number k such
that f (x) − f (y) ≥ k · (x − y). Replacing x with ξ and y with E[ξ], we obtain
f (ξ) − f (E[ξ]) ≥ k · (ξ − E[ξ]).
Taking the expected values on both sides, we have
E[f (ξ)] − f (E[ξ]) ≥ k · (E[ξ] − E[ξ]) = 0
1.13 Convergence Concepts

There are four main types of convergence concepts of random sequence:
convergence almost surely (a.s.), convergence in probability, convergence in
mean, and convergence in distribution.
Table 1.1: Relations among Convergence Concepts
Convergence Almost Surely

& Convergence Convergence
→
% in Probability in Distribution
Convergence in Mean
Definition 1.24 Suppose that ξ, ξ1 , ξ2 , · · · are random variables defined on

the probability space (Ω, A, Pr). The sequence {ξi } is said to be convergent
a.s. to ξ if and only if there exists a set A ∈ A with Pr{A} = 1 such that
lim |ξi (ω) − ξ(ω)| = 0 (1.56)

i→∞
for every ω ∈ A. In that case we write ξi → ξ, a.s.
Definition 1.25 Suppose that ξ, ξ1 , ξ2 , · · · are random variables defined on

the probability space (Ω, A, Pr). We say that the sequence {ξi } converges in
probability to ξ if
lim Pr {|ξi − ξ| ≥ ε} = 0 (1.57)
i→∞
for every ε > 0.

Definition 1.26 Suppose that ξ, ξ1 , ξ2 , · · · are random variables with finite

expected values on the probability space (Ω, A, Pr). We say that the sequence
{ξi } converges in mean to ξ if
lim E[|ξi − ξ|] = 0. (1.58)
i→∞
In addition, the sequence {ξi } is said to converge in mean square to ξ if

lim E[|ξi − ξ|2 ] = 0. (1.59)
i→∞
Definition 1.27 Suppose that Φ, Φ1 , Φ2 , · · · are the probability distributions

of random variables ξ, ξ1 , ξ2 , · · · , respectively. We say that {ξi } converges in
distribution to ξ if Φi → Φ at any continuity point of Φ.
Convergence Almost Surely vs. Convergence in Probability

Theorem 1.40 Suppose that ξ, ξ1 , ξ2 , · · · are random variables defined on
the probability space (Ω, A, Pr). Then {ξi } converges a.s. to ξ if and only if,
for every ε > 0, we have
(∞ )
[
lim Pr {|ξi − ξ| ≥ ε} = 0. (1.60)
n→∞
i=n
Proof: For every i ≥ 1 and ε > 0, we define

n o
X = ω ∈ Ω lim ξi (ω) 6= ξ(ω) ,
i→∞

Xi (ε) = ω ∈ Ω |ξi (ω) − ξ(ω)| ≥ ε .
It is clear that
∞
∞ [
!
[ \
X= Xi (ε) .
ε>0 n=1 i=n
Note that ξi → ξ, a.s. if and only if Pr{X} = 0. That is, ξi → ξ, a.s. if and
only if (∞ ∞ )
\ [
Pr Xi (ε) = 0
n=1 i=n
for every ε > 0. Since
∞
[ ∞ [
\ ∞
Xi (ε) ↓ Xi (ε),
i=n n=1 i=n
it follows from the probability continuity theorem that

(∞ ) (∞ ∞ )
[ \ [
lim Pr Xi (ε) = Pr Xi (ε) = 0.
n→∞
i=n n=1 i=n


the probability space (Ω, A, Pr). If {ξi } converges a.s. to ξ, then {ξi } converges
in probability to ξ.
Proof: It follows from the convergence a.s. and Theorem 1.40 that
(∞ )
[
lim Pr {|ξi − ξ| ≥ ε} = 0
n→∞
i=n
for each ε > 0. For every n ≥ 1, since

∞
[
{|ξn − ξ| ≥ ε} ⊂ {|ξi − ξ| ≥ ε},
i=n
we have Pr{|ξn − ξ| ≥ ε} → 0 as n → ∞. Hence the theorem holds.
Example 1.28: Convergence in probability does not imply convergence a.s.

For example, take (Ω, A, Pr) to be the interval [0, 1] with Borel algebra and
Lebesgue measure. For any positive integer i, there is an integer j such that
i = 2j + k, where k is an integer between 0 and 2j − 1. We define a random
variable on Ω by
(
1, if k/2j ≤ ω ≤ (k + 1)/2j
ξi (ω) =
0, otherwise
for i = 1, 2, · · · and ξ = 0. For any small number ε > 0, we have

1
Pr {|ξi − ξ| ≥ ε} = →0
2j
as i → ∞. That is, the sequence {ξi } converges in probability to ξ. However,
for any ω ∈ [0, 1], there is an infinite number of intervals of the form [k/2j , (k+
1)/2j ] containing ω. Thus ξi (ω) 6→ 0 as i → ∞. In other words, the sequence
{ξi } does not converge a.s. to ξ.
Convergence in Probability vs. Convergence in Mean

the probability space (Ω, A, Pr). If the sequence {ξi } converges in mean to ξ,
then {ξi } converges in probability to ξ.
Proof: It follows from the Markov inequality that, for any given number
ε > 0,
E[|ξi − ξ|]
Pr {|ξi − ξ| ≥ ε} ≤ →0
ε
as i → ∞. Thus {ξi } converges in probability to ξ.
Example 1.29: Convergence in probability does not imply convergence in

mean. For example, take (Ω, A, Pr) to be {ω1 , ω2 , · · · } with Pr{ωj } = 1/2j
for j = 1, 2, · · · The random variables are defined by
i
2 , if j = i
ξi {ωj } =
0, otherwise

1
Pr {|ξi − ξ| ≥ ε} = → 0.
2i
That is, the sequence {ξi } converges in probability to ξ. However, we have
1
E [|ξi − ξ|] = 2i · = 1.
2i
That is, the sequence {ξi } does not converge in mean to ξ.
Convergence Almost Surely vs. Convergence in Mean
Example 1.30: Convergence a.s. does not imply convergence in mean. For
example, take (Ω, A, Pr) to be {ω1 , ω2 , · · · } with Pr{ωj } = 1/2j for j =
1, 2, · · · The random variables are defined by
i
2 , if j = i
ξi {ωj } =
0, otherwise
for i = 1, 2, · · · and ξ = 0. Then {ξi } converges a.s. to ξ. However, the

sequence {ξi } does not converge in mean to ξ.
Example 1.31: Convergence in mean does not imply convergence a.s. For
example, take (Ω, A, Pr) to be the interval [0, 1] with Borel algebra and
Lebesgue measure. For any positive integer i, there is an integer j such
that i = 2j + k, where k is an integer between 0 and 2j − 1. We define a
random variable on Ω by
(
1, if k/2j ≤ ω ≤ (k + 1)/2j
ξi (ω) =
0, otherwise
for i = 1, 2, · · · and ξ = 0. Then

1
E [|ξi − ξ|] = → 0.
2j
That is, the sequence {ξi } converges in mean to ξ. However, {ξi } does not
converge a.s. to ξ.
Section 1.14 - Conditional Probability 39
Convergence in Probability vs. Convergence in Distribution

the probability space (Ω, A, Pr). If the sequence {ξi } converges in probability
to ξ, then {ξi } converges in distribution to ξ.
Proof: Let x be any given continuity point of the distribution Φ. On the

one hand, for any y > x, we have
{ξi ≤ x} = {ξi ≤ x, ξ ≤ y} ∪ {ξi ≤ x, ξ > y} ⊂ {ξ ≤ y} ∪ {|ξi − ξ| ≥ y − x}
which implies that
Φi (x) ≤ Φ(y) + Pr{|ξi − ξ| ≥ y − x}.
Since {ξi } converges in probability to ξ, we have Pr{|ξi − ξ| ≥ y − x} → 0.

Thus we obtain lim supi→∞ Φi (x) ≤ Φ(y) for any y > x. Letting y → x, we
get
lim sup Φi (x) ≤ Φ(x). (1.61)
i→∞
On the other hand, for any z < x, we have
{ξ ≤ z} = {ξ ≤ z, ξi ≤ x} ∪ {ξ ≤ z, ξi > x} ⊂ {ξi ≤ x} ∪ {|ξi − ξ| ≥ x − z}
which implies that
Φ(z) ≤ Φi (x) + Pr{|ξi − ξ| ≥ x − z}.
Since Pr{|ξi − ξ| ≥ x − z} → 0, we obtain Φ(z) ≤ lim inf i→∞ Φi (x) for any
z < x. Letting z → x, we get
Φ(x) ≤ lim inf Φi (x). (1.62)

i→∞
It follows from (1.61) and (1.62) that Φi (x) → Φ(x). The theorem is proved.
Example 1.32: Convergence in distribution does not imply convergence

in probability. For example, take (Ω, A, Pr) to be {ω1 , ω2 } with Pr{ω1 } =
Pr{ω2 } = 0.5, and

−1, if ω = ω1
ξ(ω) =
1, if ω = ω2 .
We also define ξi = −ξ for all i. Then ξi and ξ are identically distributed.

Thus {ξi } converges in distribution to ξ. But, for any small number ε > 0,
we have Pr{|ξi − ξ| > ε} = Pr{Ω} = 1. That is, the sequence {ξi } does not
converge in probability to ξ.
1.14 Conditional Probability

We consider the probability of an event A after it has been learned that
some other event B has occurred. This new probability of A is called the
conditional probability of A given B.
Definition 1.28 Let (Ω, A, Pr) be a probability space, and A, B ∈ A. Then

the conditional probability of A given B is defined by
Pr{A ∩ B}
Pr{A|B} = (1.63)
Pr{B}
provided that Pr{B} > 0.
Theorem 1.44 Let (Ω, A, Pr) be a probability space, and B an event with
Pr{B} > 0. Then Pr{·|B} defined by (1.63) is a probability measure, and
(Ω, A, Pr{·|B}) is a probability space.
Proof: It is sufficient to prove that Pr{·|B} satisfies the normality, nonneg-

ativity and countable additivity axioms. At first, we have
Pr{Ω ∩ B} Pr{B}
Pr{Ω|B} = = = 1.
Pr{B} Pr{B}
Secondly, for any A ∈ A, the set function Pr{A|B} is nonnegative. Finally,

for any countable sequence {Ai } of mutually disjoint events, we have
∞ ∞
S P
(∞
[
) Pr Ai ∩ B Pr{Ai ∩ B} X ∞
i=1
Pr Ai |B = = i=1 = Pr{Ai |B}.
i=1
Pr{B} Pr{B} i=1
Thus Pr{·|B} is a probability measure. Furthermore, (Ω, A, Pr{·|B}) is a

probability space.
Theorem 1.45 (Bayes Formula) Let the events A1 , A2 , · · · , An form a par-

tition of the space Ω such that Pr{Ai } > 0 for i = 1, 2, · · · , n, and let B be
an event with Pr{B} > 0. Then we have
Pr{Ak } Pr{B|Ak }
Pr{Ak |B} = P
n (1.64)
Pr{Ai } Pr{B|Ai }
i=1
for k = 1, 2, · · · , n.
Proof: Since A1 , A2 , · · · , An form a partition of the space Ω, we have

n
X n
X
Pr{B} = Pr{Ai ∩ B} = Pr{Ai } Pr{B|Ai }
i=1 i=1
which is also called the formula for total probability. Thus, for any k, we have
Pr{Ak ∩ B} Pr{Ak } Pr{B|Ak }
Pr{Ak |B} = = P
n .
Pr{B}
Pr{Ai } Pr{B|Ai }
i=1
Remark 1.2: Especially, let A and B be two events with Pr{A} > 0 and
Pr{B} > 0. Then A and Ac form a partition of the space Ω, and the Bayes
formula is
Pr{A} Pr{B|A}
Pr{A|B} = . (1.65)
Pr{B}
Remark 1.3: In statistical applications, the events A1 , A2 , · · · , An are often

called hypotheses. Furthermore, for each i, the Pr{Ai } is called the prior
probability of Ai , and Pr{Ai |B} is called the posterior probability of Ai after
the occurrence of event B.

expected value β. Then for any real numbers a > 0 and x > 0, the conditional
probability of ξ ≥ a + x given ξ ≥ a is
Pr{ξ ≥ a + x|ξ ≥ a} = exp(−x/β) = Pr{ξ ≥ x}
which means that the conditional probability is identical to the original prob-
ability. This is the so-called memoryless property of exponential distribution.
In other words, it is as good as new if it is functioning on inspection.
Definition 1.29 Let ξ be a random variable on (Ω, A, Pr). A conditional
random variable of ξ given B is a measurable function ξ|B from the condi-
tional probability space (Ω, A, Pr{·|B}) to the set of real numbers such that
ξ|B (ω) ≡ ξ(ω), ∀ω ∈ Ω. (1.66)
Definition 1.30 The conditional probability distribution Φ: < → [0, 1] of a
random variable ξ given B is defined by
Φ(x|B) = Pr {ξ ≤ x|B} (1.67)
provided that Pr{B} > 0.
Example 1.34: Let ξ and η be random variables. Then the conditional

probability distribution of ξ given η = y is
Pr{ξ ≤ x, η = y}
Φ(x|η = y) = Pr {ξ ≤ x|η = y} =
Pr{η = y}
provided that Pr{η = y} > 0.
Definition 1.31 The conditional probability density function φ of a random

variable ξ given B is a nonnegative function such that
Z x
Φ(x|B) = φ(y|B)dy, ∀x ∈ < (1.68)
−∞
where Φ(x|B) is the conditional probability distribution of ξ given B.
Example 1.35: Let (ξ, η) be a random vector with joint probability density
function ψ. Then the marginal probability density functions of ξ and η are
Z +∞ Z +∞
f (x) = ψ(x, y)dy, g(y) = ψ(x, y)dx,
−∞ −∞
respectively. Furthermore, we have

Z x Z y Z y Z x
ψ(r, t)
Pr{ξ ≤ x, η ≤ y} = ψ(r, t)drdt = dr g(t)dt
−∞ −∞ −∞ −∞ g(t)
which implies that the conditional probability distribution of ξ given η = y

is Z x
ψ(r, y)
Φ(x|η = y) = dr, a.s. (1.69)
−∞ g(y)
and the conditional probability density function of ξ given η = y is
ψ(x, y) ψ(x, y)
φ(x|η = y) = =Z +∞ , a.s. (1.70)
g(y)
ψ(x, y)dx
−∞
Note that (1.69) and (1.70) are defined only for g(y) 6= 0. In fact, the set
{y|g(y) = 0} has probability 0. Especially, if ξ and η are independent random
variables, then ψ(x, y) = f (x)g(y) and φ(x|η = y) = f (x).
Definition 1.32 Let ξ be a random variable. Then the conditional expected

value of ξ given B is defined by
Z +∞ Z 0
E[ξ|B] = Pr{ξ ≥ r|B}dr − Pr{ξ ≤ r|B}dr (1.71)
0 −∞
Following conditional probability and conditional expected value, we also

have conditional variance, conditional moments, conditional critical values,
conditional entropy as well as conditional convergence.
Hazard Rate
Definition 1.33 Let ξ be a nonnegative random variable representing life-
time. Then the hazard rate (or failure rate) is

Pr{ξ ≤ x + ∆ ξ > x}
h(x) = lim . (1.72)
∆↓0 ∆
The hazard rate tells us the probability of a failure just after time x when
it is functioning at time x. If ξ has probability distribution Φ and probability
density function φ, then the hazard rate
φ(x)
h(x) = .
1 − Φ(x)

expected value β. Then its hazard rate h(x) ≡ 1/β.
Chapter 2
Credibility Theory
The concept of fuzzy set was initiated by Zadeh [259] via membership function
in 1965. In order to measure a fuzzy event, Zadeh [262] proposed the concept
of possibility measure. Although possibility measure has been widely used,
it has no self-duality property. However, a self-dual measure is absolutely
needed in both theory and practice. In order to define a self-dual measure,
Liu and Liu [132] presented the concept of credibility measure. In addition,
a sufficient and necessary condition for credibility measure was given by Li
and Liu [102]. Credibility theory, founded by Liu [135] in 2004 and refined
by Liu [138] in 2007, is a branch of mathematics for studying the behavior of
fuzzy phenomena.
The emphasis in this chapter is mainly on credibility measure, credibility
space, fuzzy variable, membership function, credibility distribution, indepen-
dence, identical distribution, expected value, variance, moments, critical val-
ues, entropy, distance, convergence almost surely, convergence in credibility,
convergence in mean, convergence in distribution, and conditional credibility.
2.1 Credibility Space
Let Θ be a nonempty set, and P the power set of Θ (i.e., the larggest σ-
algebra over Θ). Each element in P is called an event. In order to present an
axiomatic definition of credibility, it is necessary to assign to each event A a
number Cr{A} which indicates the credibility that A will occur. In order to
ensure that the number Cr{A} has certain mathematical properties which we
intuitively expect a credibility to have, we accept the following four axioms:
Axiom 1. (Normality) Cr{Θ} = 1.
Axiom 2. (Monotonicity) Cr{A} ≤ Cr{B} whenever A ⊂ B.
Axiom 3. (Self-Duality) Cr{A} + Cr{Ac } = 1 for any event A.
46 Chapter 2 - Credibility Theory
Axiom 4. (Maximality) Cr {∪i Ai } = supi Cr{Ai } for any events {Ai } with
supi Cr{Ai } < 0.5.
Definition 2.1 (Liu and Liu [132]) The set function Cr is called a cred-
ibility measure if it satisfies the normality, monotonicity, self-duality, and
maximality axioms.
Example 2.1: Let Θ = {θ1 , θ2 }. For this case, there are only four events:
∅, {θ1 }, {θ2 }, Θ. Define Cr{∅} = 0, Cr{θ1 } = 0.7, Cr{θ2 } = 0.3, and Cr{Θ} =
1. Then the set function Cr is a credibility measure because it satisfies the
four axioms.
Example 2.2: Let Θ be a nonempty set. Define Cr{∅} = 0, Cr{Θ} = 1 and

Cr{A} = 1/2 for any subset A (excluding ∅ and Θ). Then the set function
Cr is a credibility measure.
Example 2.3: Let µ be a nonnegative function on Θ (for example, the set

of real numbers) such that
sup µ(x) = 1. (2.1)
x∈Θ
Then the set function

1
Cr{A} = sup µ(x) + 1 − sup µ(x) (2.2)
2 x∈A x∈Ac
is a credibility measure on Θ.
Theorem 2.1 Let Θ be a nonempty set, P the power set of Θ, and Cr the
credibility measure. Then Cr{∅} = 0 and 0 ≤ Cr{A} ≤ 1 for any A ∈ P.
Proof: It follows from Axioms 1 and 3 that Cr{∅} = 1 − Cr{Θ} = 1 − 1 = 0.

Since ∅ ⊂ A ⊂ Θ, we have 0 ≤ Cr{A} ≤ 1 by using Axiom 2.
credibility measure. Then for any A, B ∈ P, we have
Cr{A ∪ B} = Cr{A} ∨ Cr{B} if Cr{A ∪ B} ≤ 0.5, (2.3)
Cr{A ∩ B} = Cr{A} ∧ Cr{B} if Cr{A ∩ B} ≥ 0.5. (2.4)

The above equations hold for not only finite number of events but also infinite
number of events.
Proof: If Cr{A ∪ B} < 0.5, then Cr{A} ∨ Cr{B} < 0.5 by using Axiom 2.
Thus the equation (2.3) follows immediately from Axiom 4. If Cr{A ∪ B} =
0.5 and (2.3) does not hold, then we have Cr{A} ∨ Cr{B} < 0.5. It follows
from Axiom 4 that
Cr{A ∪ B} = Cr{A} ∨ Cr{B} < 0.5.

Section 2.1 - Credibility Space 47
A contradiction proves (2.3). Next we prove (2.4). Since Cr{A ∩ B} ≥ 0.5,

we have Cr{Ac ∪ B c } ≤ 0.5 by the self-duality. Thus
Cr{A ∩ B} = 1 − Cr{Ac ∪ B c } = 1 − Cr{Ac } ∨ Cr{B c }

= (1 − Cr{Ac }) ∧ (1 − Cr{B c }) = Cr{A} ∧ Cr{B}.
credibility measure. Then for any A, B ∈ P, we have
Cr{A ∪ B} = Cr{A} ∨ Cr{B} if Cr{A} + Cr{B} < 1, (2.5)
Cr{A ∩ B} = Cr{A} ∧ Cr{B} if Cr{A} + Cr{B} > 1. (2.6)
Proof: Suppose Cr{A} + Cr{B} < 1. Then there exists at least one term
less than 0.5, say Cr{B} < 0.5. If Cr{A} < 0.5 also holds, then the equation
(2.3) follows immediately from Axiom 4. If Cr{A} ≥ 0.5, then by using
Theorem 2.2, we obtain
Cr{A} = Cr{A∪(B∩B c )} = Cr{(A∪B)∩(A∪B c )} = Cr{A∪B}∧Cr{A∪B c }.
On the other hand, we have
Cr{A} < 1 − Cr{B} = Cr{B c } ≤ Cr{A ∪ B c }.
Hence we must have Cr{A ∪ B} = Cr{A} = Cr{A} ∨ Cr{B}. The equation

(2.5) is proved. Next we suppose Cr{A} + Cr{B} > 1. Then Cr{Ac } +
Cr{B c } < 1. It follows from (2.5) that
Cr{A ∩ B} = 1 − Cr{Ac ∪ B c } = 1 − Cr{Ac } ∨ Cr{B c }

= (1 − Cr{Ac }) ∧ (1 − Cr{B c }) = Cr{A} ∧ Cr{B}.
Credibility Subadditivity Theorem

Theorem 2.4 (Liu [135], Credibility Subadditivity Theorem) The credibility
measure is subadditive. That is,
Cr{A ∪ B} ≤ Cr{A} + Cr{B} (2.7)
for any events A and B. In fact, credibility measure is not only finitely
subadditive but also countably subadditive.
Proof: The argument breaks down into three cases.

Case 1: Cr{A} < 0.5 and Cr{B} < 0.5. It follows from Axiom 4 that
Cr{A ∪ B} = Cr{A} ∨ Cr{B} ≤ Cr{A} + Cr{B}.
Case 2: Cr{A} ≥ 0.5. For this case, by using Axioms 2 and 3, we have
Cr{Ac } ≤ 0.5 and Cr{A ∪ B} ≥ Cr{A} ≥ 0.5. Then
Cr{Ac } = Cr{Ac ∩ B} ∨ Cr{Ac ∩ B c }

≤ Cr{Ac ∩ B} + Cr{Ac ∩ B c }
≤ Cr{B} + Cr{Ac ∩ B c }.
Applying this inequality, we obtain
Cr{A} + Cr{B} = 1 − Cr{Ac } + Cr{B}

≥ 1 − Cr{B} − Cr{Ac ∩ B c } + Cr{B}
= 1 − Cr{Ac ∩ B c }
= Cr{A ∪ B}.
Case 3: Cr{B} ≥ 0.5. This case may be proved by a similar process of

Case 2. The theorem is proved.
Remark 2.1: For any events A and B, it follows from the credibility subaddi-
tivity theorem that the credibility measure is null-additive, i.e., Cr{A ∪ B} =
Cr{A} + Cr{B} if either Cr{A} = 0 or Cr{B} = 0.
Theorem 2.5 Let {Bi } be a decreasing sequence of events with Cr{Bi } → 0

as i → ∞. Then for any event A, we have
lim Cr{A ∪ Bi } = lim Cr{A\Bi } = Cr{A}. (2.8)

i→∞ i→∞
Proof: It follows from the monotonicity axiom and credibility subadditivity

theorem that
Cr{A} ≤ Cr{A ∪ Bi } ≤ Cr{A} + Cr{Bi }
for each i. Thus we get Cr{A ∪ Bi } → Cr{A} by using Cr{Bi } → 0. Since

(A\Bi ) ⊂ A ⊂ ((A\Bi ) ∪ Bi ), we have
Cr{A\Bi } ≤ Cr{A} ≤ Cr{A\Bi } + Cr{Bi }.
Hence Cr{A\Bi } → Cr{A} by using Cr{Bi } → 0.
Theorem 2.6 A credibility measure on Θ is additive if and only if there are

at most two elements in Θ taking nonzero credibility values.
Proof: Suppose that the credibility measure is additive. If there are more
than two elements taking nonzero credibility values, then we may choose three
elements θ1 , θ2 , θ3 such that Cr{θ1 } ≥ Cr{θ2 } ≥ Cr{θ3 } > 0. If Cr{θ1 } ≥ 0.5,
it follows from Axioms 2 and 3 that
Cr{θ2 , θ3 } ≤ Cr{Θ \ {θ1 }} = 1 − Cr{θ1 } ≤ 0.5.
By using Axiom 4, we obtain
Cr{θ2 , θ3 } = Cr{θ2 } ∨ Cr{θ3 } < Cr{θ2 } + Cr{θ3 }.
This is in contradiction with the additivity assumption. If Cr{θ1 } < 0.5,
then Cr{θ3 } ≤ Cr{θ2 } < 0.5. It follows from Axiom 4 that
Cr{θ2 , θ3 } ∧ 0.5 = Cr{θ2 } ∨ Cr{θ3 } < 0.5
which implies that
Cr{θ2 , θ3 } = Cr{θ2 } ∨ Cr{θ3 } < Cr{θ2 } + Cr{θ3 }.
This is also in contradiction with the additivity assumption. Hence there are
at most two elements taking nonzero credibility values.
Conversely, suppose that there are at most two elements, say θ1 and θ2 ,
taking nonzero credibility values. Let A and B be two disjoint events. The
argument breaks down into two cases.
Case 1: Cr{A} = 0 or Cr{B} = 0. Then we have Cr{A ∪ B} = Cr{A} +
Cr{B} by using the credibility subadditivity theorem.
Case 2: Cr{A} > 0 or Cr{B} > 0. For this case, without loss of generality,
we suppose that θ1 ∈ A and θ2 ∈ B. Note that Cr{(A ∪ B)c } = 0. It follows
from Axiom 3 and the credibility subadditivity theorem that
Cr{A ∪ B} = Cr{A ∪ B ∪ (A ∪ B)c } = Cr{Θ} = 1,
Cr{A} + Cr{B} = Cr{A ∪ (A ∪ B)c } + Cr{B} = 1.
Hence Cr{A ∪ B} = Cr{A} + Cr{B}. The additivity is proved.
Remark 2.2: Theorem 2.6 states that a credibility measure is identical with
probability measure if there are effectively two elements in the universal set.
Credibility Semicontinuity Law

Generally speaking, the credibility measure is neither lower semicontinuous
nor upper semicontinuous. However, we have the following credibility semi-
continuity law.
Theorem 2.7 (Liu [135], Credibility Semicontinuity Law) For any events
A1 , A2 , · · · , we have
n o
lim Cr{Ai } = Cr lim Ai (2.9)
i→∞ i→∞
if one of the following conditions is satisfied:

(a) Cr {A} ≤ 0.5 and Ai ↑ A; (b) lim Cr{Ai } < 0.5 and Ai ↑ A;
i→∞
(c) Cr {A} ≥ 0.5 and Ai ↓ A; (d) lim Cr{Ai } > 0.5 and Ai ↓ A.
i→∞
Proof: (a) Since Cr{A} ≤ 0.5, we have Cr{Ai } ≤ 0.5 for each i. It follows
from Axiom 4 that
Cr{A} = Cr {∪i Ai } = sup Cr{Ai } = lim Cr{Ai }.

i i→∞
(b) Since limi→∞ Cr{Ai } < 0.5, we have supi Cr{Ai } < 0.5. It follows
from Axiom 4 that
Cr{A} = Cr {∪i Ai } = sup Cr{Ai } = lim Cr{Ai }.

i i→∞
(c) Since Cr{A} ≥ 0.5 and Ai ↓ A, it follows from the self-duality of

credibility measure that Cr{Ac } ≤ 0.5 and Aci ↑ Ac . Thus Cr{Ai } = 1 −
Cr{Aci } → 1 − Cr{Ac } = Cr{A} as i → ∞.
(d) Since limi→∞ Cr{Ai } > 0.5 and Ai ↓ A, it follows from the self-duality
of credibility measure that
lim Cr{Aci } = lim (1 − Cr{Ai }) < 0.5

i→∞ i→∞
and Aci ↑ Ac . Thus Cr{Ai } = 1 − Cr{Aci } → 1 − Cr{Ac } = Cr{A} as i → ∞.

Credibility Asymptotic Theorem
Theorem 2.8 (Credibility Asymptotic Theorem) For any events A1 , A2 , · · · ,

we have
lim Cr{Ai } ≥ 0.5, if Ai ↑ Θ, (2.10)
i→∞
lim Cr{Ai } ≤ 0.5, if Ai ↓ ∅. (2.11)

i→∞
Proof: Assume Ai ↑ Θ. If limi→∞ Cr{Ai } < 0.5, it follows from the credi-
bility semicontinuity law that
Cr{Θ} = lim Cr{Ai } < 0.5

i→∞
which is in contradiction with Cr{Θ} = 1. The first inequality is proved.

The second one may be verified similarly.
Credibility Extension Theorem

Suppose that the credibility of each singleton is given. Is the credibility
measure fully and uniquely determined? This subsection will answer the
question.
Theorem 2.9 Suppose that Θ is a nonempty set. If Cr is a credibility mea-

sure, then we have
sup Cr{θ} ≥ 0.5,
θ∈Θ
(2.12)
Cr{θ∗ } + sup Cr{θ} = 1 if Cr{θ∗ } ≥ 0.5.
θ6=θ ∗
We will call (2.12) the credibility extension condition.
Proof: If sup Cr{θ} < 0.5, then by using Axiom 4, we have
1 = Cr{Θ} = sup Cr{θ} < 0.5.

θ∈Θ
This contradiction proves sup Cr{θ} ≥ 0.5. We suppose that θ∗ ∈ Θ is a point

with Cr{θ∗ } ≥ 0.5. It follows from Axioms 3 and 4 that Cr{Θ \ {θ∗ }} ≤ 0.5,
and
Cr{Θ \ {θ∗ }} = sup Cr{θ}.
θ6=θ ∗
Hence the second formula of (2.12) is true by the self-duality of credibility

measure.
Theorem 2.10 (Li and Liu [102], Credibility Extension Theorem) Suppose
that Θ is a nonempty set, and Cr{θ} is a nonnegative function on Θ satisfying
the credibility extension condition (2.12). Then Cr{θ} has a unique extension
to a credibility measure as follows,

 sup Cr{θ}, if sup Cr{θ} < 0.5
θ∈A θ∈A
Cr{A} = (2.13)
 1 − sup Cr{θ}, if sup Cr{θ} ≥ 0.5.
θ∈Ac θ∈A
Proof: We first prove that the set function Cr{A} defined by (2.13) is a
credibility measure.
Step 1: By the credibility extension condition sup Cr{θ} ≥ 0.5, we have
θ∈Θ
Cr{Θ} = 1 − sup Cr{θ} = 1 − 0 = 1.

θ∈∅
Step 2: If A ⊂ B, then B ⊂ Ac . The proof breaks down into two cases.

c
Case 1: sup Cr{θ} < 0.5. For this case, we have

θ∈A
Cr{A} = sup Cr{θ} ≤ sup Cr{θ} ≤ Cr{B}.

θ∈A θ∈B
Case 2: sup Cr{θ} ≥ 0.5. For this case, we have sup Cr{θ} ≥ 0.5, and
θ∈A θ∈B
Cr{A} = 1 − sup Cr{θ} ≤ 1 − sup Cr{θ} = Cr{B}.

θ∈Ac θ∈B c
Step 3: In order to prove Cr{A} + Cr{Ac } = 1, the argument breaks

down into two cases.
Case 1: sup Cr{θ} < 0.5. For this case, we have sup Cr{θ} ≥ 0.5. Thus,
θ∈A θ∈Ac
Cr{A} + Cr{Ac } = sup Cr{θ} + 1 − sup Cr{θ} = 1.

θ∈A θ∈A
Case 2: sup Cr{θ} ≥ 0.5. For this case, we have sup Cr{θ} ≤ 0.5, and
θ∈A θ∈Ac
Cr{A} + Cr{Ac } = 1 − sup Cr{θ} + sup Cr{θ} = 1.

θ∈Ac θ∈Ac
Step 4: For any collection {Ai } with supi Cr{Ai } < 0.5, we have
Cr{∪i Ai } = sup Cr{θ} = sup sup Cr{θ} = sup Cr{Ai }.

θ∈∪i Ai i θ∈Ai i
Thus Cr is a credibility measure because it satisfies the four axioms.

Finally, let us prove the uniqueness. Assume that Cr1 and Cr2 are two
credibility measures such that Cr1 {θ} = Cr2 {θ} for each θ ∈ Θ. Let us prove
that Cr1 {A} = Cr2 {A} for any event A. The argument breaks down into
three cases.
Case 1: Cr1 {A} < 0.5. For this case, it follows from Axiom 4 that
Cr1 {A} = sup Cr1 {θ} = sup Cr2 {θ} = Cr2 {A}.
θ∈A θ∈A
Case 2: Cr1 {A} > 0.5. For this case, we have Cr1 {Ac } < 0.5. It follows
from the first case that Cr1 {Ac } = Cr2 {Ac } which implies Cr1 {A} = Cr2 {A}.
Case 3: Cr1 {A} = 0.5. For this case, we have Cr1 {Ac } = 0.5, and
Cr2 {A} ≥ sup Cr2 {θ} = sup Cr1 {θ} = Cr1 {A} = 0.5,
θ∈A θ∈A
Cr2 {Ac } ≥ sup Cr2 {θ} = sup Cr1 {θ} = Cr1 {Ac } = 0.5.
θ∈Ac θ∈Ac
Hence Cr2 {A} = 0.5 = Cr1 {A}. The uniqueness is proved.
Credibility Space
Definition 2.2 Let Θ be a nonempty set, P the power set of Θ, and Cr a
credibility measure. Then the triplet (Θ, P, Cr) is called a credibility space.
Example 2.4: The triplet (Θ, P, Cr) is a credibility space if
Θ = {θ1 , θ2 , · · · }, Cr{θi } ≡ 1/2 for i = 1, 2, · · · (2.14)
Note that the credibility measure is produced by the credibility extension

theorem as follows,

 0, if A = ∅

Cr{A} = 1, if A = Θ

1/2, otherwise.

Θ = {θ1 , θ2 , · · · }, Cr{θi } = i/(2i + 1) for i = 1, 2, · · · (2.15)
By using the credibility extension theorem, we obtain the following credibility

measure,
 i
 sup , if A is finite
2i + 1

θi ∈A

Cr{A} =
i
 1 − sup

 , if A is infinite.
θi ∈A c 2i + 1
Θ = {θ1 , θ2 , · · · }, Cr{θ1 } = 1/2, Cr{θi } = 1/i for i = 2, 3, · · · (2.16)
For this case, the credibility measure is


 sup 1/i, if A contains neither θ1 nor θ2


 θi ∈A
Cr{A} = 1/2, if A contains only one of θ1 and θ2


 1 − sup 1/i, if A contains both θ1 and θ2 .


θi ∈Ac
Θ = [0, 1], Cr{θ} = θ/2 for θ ∈ Θ. (2.17)
For this case, the credibility measure is

 1
 sup θ, if sup θ < 1
2 θ∈A

 θ∈A
Cr{A} =
1
 1 − sup θ, if sup θ = 1.


2 θ∈Ac θ∈A
Product Credibility Measure

Product credibility measure may be defined in multiple ways. This book
accepts the following axiom.
Axiom 5. (Product Credibility Axiom) Let Θk be nonempty sets on which
Crk are credibility measures, k = 1, 2, · · · , n, respectively, and Θ = Θ1 ×Θ2 ×
· · · × Θn . Then
Cr{(θ1 , θ2 , · · · , θn )} = Cr1 {θ1 } ∧ Cr2 {θ2 } ∧ · · · ∧ Crn {θn } (2.18)
for each (θ1 , θ2 , · · · , θn ) ∈ Θ.
Theorem 2.11 (Product Credibility Theorem) Let Θk be nonempty sets on

which Crk are the credibility measures, k = 1, 2, · · · , n, respectively, and Θ =
Θ1 × Θ2 × · · · × Θn . Then Cr = Cr1 ∧ Cr2 ∧ · · · ∧ Crn defined by Axiom 5 has
a unique extension to a credibility measure on Θ as follows,

 sup min Crk {θk },
(θ1 ,θ2 ··· ,θn )∈A 1≤k≤n







 if sup min Crk {θk } < 0.5
(θ1 ,θ2 ,··· ,θn )∈A 1≤k≤n


Cr{A} = (2.19)


 1− sup min Crk {θk },
(θ1 ,θ2 ,··· ,θn )∈Ac 1≤k≤n




if sup min Crk {θk } ≥ 0.5.



(θ1 ,θ2 ,··· ,θn )∈A 1≤k≤n

Proof: For each θ = (θ1 , θ2 , · · · , θn ) ∈ Θ, we have Cr{θ} = Cr1 {θ1 } ∧

Cr2 {θ2 } ∧ · · · ∧ Crn {θn }. Let us prove that Cr{θ} satisfies the credibility
extension condition. Since sup Cr{θk } ≥ 0.5 for each k, we have
θk ∈Θk
sup Cr{θ} = sup min Crk {θk } ≥ 0.5.

θ∈Θ (θ1 ,θ2 ,··· ,θn )∈Θ 1≤k≤n
Now we suppose that θ ∗ = (θ1∗ , θ2∗ , · · · , θn∗ ) is a point with Cr{θ ∗ } ≥ 0.5.
Without loss of generality, let i be the index such that
Cr{θ ∗ } = min Crk {θk∗ } = Cri {θi∗ }. (2.20)

1≤k≤n
We also immediately have
Crk {θk∗ } ≥ 0.5, k = 1, 2, · · · , n; (2.21)
Crk {θk∗ } + sup Crk {θk } = 1, k = 1, 2, · · · , n; (2.22)

∗
θk 6=θk
sup Cri {θi } ≥ sup Crk {θk }, k = 1, 2, · · · , n; (2.23)

θi 6=θi∗ ∗
θk 6=θk
sup Crk {θk } ≤ 0.5, k = 1, · · · , n. (2.24)

∗
θk 6=θk
It follows from (2.21) and (2.24) that
sup Cr{θ} = sup min Crk {θk }

θ6=θ ∗ ∗ ) 1≤k≤n
(θ1 ,θ2 ,··· ,θn )6=(θ1∗ ,θ2∗ ,··· ,θn
∗
≥ sup min Crk {θk } ∧ Cri {θi } ∧ min Crk {θk∗ }
θi 6=θi∗ 1≤k≤i−1 i+1≤k≤n
= sup Cri {θi }.

θi 6=θi∗
We next suppose that
sup Cr{θ} > sup Cri {θi }.

θ6=θ ∗ θi 6=θi∗
Then there is a point (θ10 , θ20 , · · · , θn0 ) 6= (θ1∗ , θ2∗ , · · · , θn∗ ) such that
min Crk {θk0 } > sup Cri {θi }.

1≤k≤n θi 6=θi∗
Let j be one of the index such that θj0 6= θj∗ . Then
Crj {θj0 } > sup Cri {θi }.

θi 6=θi∗
That is,
sup Crj {θj } > sup Cri {θi }
θj 6=θj∗ θi 6=θi∗
which is in contradiction with (2.23). Thus
sup Cr{θ} = sup Cri {θi }. (2.25)

It follows from (2.20), (2.22) and (2.25) that
Cr{θ ∗ } + sup Cr{θ} = Cri {θi∗ } + sup Cri {θi } = 1.

Thus Cr satisfies the credibility extension condition. It follows from the cred-
ibility extension theorem that Cr{A} is just the unique extension of Cr{θ}.
Definition 2.3 Let (Θk , Pk , Crk ), k = 1, 2, · · · , n be credibility spaces, Θ =

Θ1 × Θ2 × · · · × Θn and Cr = Cr1 ∧ Cr2 ∧ · · · ∧ Crn . Then (Θ, P, Cr) is called
the product credibility space of (Θk , Pk , Crk ), k = 1, 2, · · · , n.
Theorem 2.12 Let (Θ, P, Cr) be the product credibility space of (Θk , Pk , Crk ),
k = 1, 2, · · · , n. Then for any Ak ∈ Pk , k = 1, 2, · · · , n, we have
Cr{A1 × A2 × · · · × Ak } = Cr1 {A1 } ∧ Cr2 {A2 } ∧ · · · ∧ Crn {An }.

Proof: We only prove the case of n = 2. If Cr1 {A1 } < 0.5 or Cr2 {A2 } < 0.5,
then we have
sup Cr1 {θ1 } < 0.5 or sup Cr2 {θ2 } < 0.5.
θ1 ∈A1 θ2 ∈A2
It follows from
sup Cr1 {θ1 } ∧ Cr2 {θ2 } = sup Cr1 {θ1 } ∧ sup Cr2 {θ2 } < 0.5
(θ1 ,θ2 )∈A1 ×A2 θ1 ∈A1 θ2 ∈A2
that
Cr{A1 × A2 } = sup Cr1 {θ1 } ∧ sup Cr2 {θ2 } = Cr1 {A1 } ∧ Cr2 {A2 }.
θ1 ∈A1 θ2 ∈A2
If Cr1 {A1 } ≥ 0.5 and Cr2 {A2 } ≥ 0.5, then we have

sup Cr1 {θ1 } ≥ 0.5 and sup Cr2 {θ2 } ≥ 0.5.
θ1 ∈A1 θ2 ∈A2
It follows from
sup Cr1 {θ1 } ∧ Cr2 {θ2 } = sup Cr1 {θ1 } ∧ sup Cr2 {θ2 } ≥ 0.5
(θ1 ,θ2 )∈A1 ×A2 θ1 ∈A1 θ2 ∈A2
that
Cr{A1 × A2 } = 1 − sup Cr1 {θ1 } ∧ Cr2 {θ2 }
/ A1 ×A2
(θ1 ,θ2 )∈
! !
= 1 − sup Cr1 {θ1 } ∧ 1 − sup Cr2 {θ2 }
θ1 ∈Ac1 θ2 ∈Ac2
= Cr1 {A1 } ∧ Cr2 {A2 }.

Theorem 2.13 (Infinite Product Credibility Theorem) Suppose that Θk are
nonempty sets, Crk the credibility measures on Pk , k = 1, 2, · · · , respectively.
Let Θ = Θ1 × Θ2 × · · · Then

 sup inf Crk {θk },
(θ1 ,θ2 ,··· )∈A 1≤k<∞







 if sup inf Crk {θk } < 0.5
(θ1 ,θ2 ,··· )∈A 1≤k<∞


Cr{A} =


 1− sup inf Crk {θk },
(θ1 ,θ2 ,··· )∈Ac 1≤k<∞




if sup inf Crk {θk } ≥ 0.5



(θ1 ,θ2 ,··· )∈A 1≤k<∞

is a credibility measure on P.
Proof: Like Theorem 2.11 except Cr{(θ1 , θ2 , · · · )} = inf 1≤k<∞ Crk {θk }.
Definition 2.4 Let (Θk , Pk , Crk ), k = 1, 2, · · · be credibility spaces. Define
Θ = Θ1 × Θ2 × · · · and Cr = Cr1 ∧ Cr2 ∧ · · · Then (Θ, P, Cr) is called the
infinite product credibility space of (Θk , Pk , Crk ), k = 1, 2, · · ·
Section 2.2 - Fuzzy Variables 57
2.2 Fuzzy Variables

Definition 2.5 A fuzzy variable is a (measurable) function from a credibility
space (Θ, P, Cr) to the set of real numbers.
Example 2.8: Take (Θ, P, Cr) to be {θ1 , θ2 } with Cr{θ1 } = Cr{θ2 } = 0.5.
Then the function (
0, if θ = θ1
ξ(θ) =
1, if θ = θ2
is a fuzzy variable.
Example 2.9: Take (Θ, P, Cr) to be the interval [0, 1] with Cr{θ} = θ/2 for
each θ ∈ [0, 1]. Then the identity function ξ(θ) = θ is a fuzzy variable.
Example 2.10: A crisp number c may be regarded as a special fuzzy vari-

able. In fact, it is the constant function ξ(θ) ≡ c on the credibility space
(Θ, P, Cr).
Remark 2.3: Since a fuzzy variable ξ is a function on a credibility space,

for any set B of real numbers, the set

{ξ ∈ B} = θ ∈ Θ ξ(θ) ∈ B (2.26)
is always an element in P. In other words, the fuzzy variable ξ is always a

measurable function and {ξ ∈ B} is always an event.
Definition 2.6 A fuzzy variable ξ is said to be

(a) nonnegative if Cr{ξ < 0} = 0;
(b) positive if Cr{ξ ≤ 0} = 0;
(c) continuous if Cr{ξ = x} is a continuous function of x;
Cr {ξ 6= x1 , ξ 6= x2 , · · · , ξ 6= xm } = 0; (2.27)
Cr {ξ 6= x1 , ξ 6= x2 , · · · } = 0. (2.28)
Definition 2.7 Let ξ1 and ξ2 be fuzzy variables defined on the credibility

space (Θ, P, Cr). We say ξ1 = ξ2 if ξ1 (θ) = ξ2 (θ) for almost all θ ∈ Θ.
Fuzzy Vector
Definition 2.8 An n-dimensional fuzzy vector is defined as a function from
a credibility space (Θ, P, Cr) to the set of n-dimensional real vectors.
Theorem 2.14 The vector (ξ1 , ξ2 , · · · , ξn ) is a fuzzy vector if and only if

ξ1 , ξ2 , · · · , ξn are fuzzy variables.
Proof: Write ξ = (ξ1 , ξ2 , · · · , ξn ). Suppose that ξ is a fuzzy vector. Then

ξ1 , ξ2 , · · · , ξn are functions from Θ to <. Thus ξ1 , ξ2 , · · · , ξn are fuzzy vari-
ables. Conversely, suppose that ξi are fuzzy variables defined on the cred-
ibility spaces (Θi , Pi , Cri ), i = 1, 2, · · · , n, respectively. It is clear that
(ξ1 , ξ2 , · · · , ξn ) is a function from the product credibility space (Θ, P, Cr)
to <n , i.e.,
ξ(θ1 , θ2 , · · · , θn ) = (ξ1 (θ1 ), ξ2 (θ2 ), · · · , ξn (θn ))
for all (θ1 , θ2 , · · · , θn ) ∈ Θ. Hence ξ = (ξ1 , ξ2 , · · · , ξn ) is a fuzzy vector.
Fuzzy Arithmetic
In this subsection, we will suppose that all fuzzy variables are defined on a
common credibility space. Otherwise, we may embed them into the product
credibility space.
Definition 2.9 Let f : <n → < be a function, and ξ1 , ξ2 , · · · , ξn fuzzy vari-

ables on the credibility space (Θ, P, Cr). Then ξ = f (ξ1 , ξ2 , · · · , ξn ) is a fuzzy
variable defined as
ξ(θ) = f (ξ1 (θ), ξ2 (θ), · · · , ξn (θ)) (2.29)
for any θ ∈ Θ.
Example 2.11: Let ξ1 and ξ2 be fuzzy variables on the credibility space

(Θ, P, Cr). Then their sum is
(ξ1 + ξ2 )(θ) = ξ1 (θ) + ξ2 (θ), ∀θ ∈ Θ
and their product is
(ξ1 × ξ2 )(θ) = ξ1 (θ) × ξ2 (θ), ∀θ ∈ Θ.
The reader may wonder whether ξ(θ1 , θ2 , · · · , θn ) defined by (2.29) is a

fuzzy variable. The following theorem answers this question.
Theorem 2.15 Let ξ be an n-dimensional fuzzy vector, and f : <n → < a

function. Then f (ξ) is a fuzzy variable.
Proof: Since f (ξ) is a function from a credibility space to the set of real
numbers, it is a fuzzy variable.
2.3 Membership Function

Definition 2.10 Let ξ be a fuzzy variable defined on the credibility space
(Θ, P, Cr). Then its membership function is derived from the credibility mea-
sure by
µ(x) = (2Cr{ξ = x}) ∧ 1, x ∈ <. (2.30)
Section 2.3 - Membership Function 59
Membership function represents the degree that the fuzzy variable ξ takes
some prescribed value. How do we determine membership functions? There
are several methods reported in the past literature. Anyway, the membership
degree µ(x) = 0 if x is an impossible point, and µ(x) = 1 if x is the most
possible point that ξ takes.
Example 2.12: It is clear that a fuzzy variable has a unique membership

function. However, a membership function may produce multiple fuzzy vari-
ables. For example, let Θ = {θ1 , θ2 } and Cr{θ1 } = Cr{θ2 } = 0.5. Then
(Θ, P, Cr) is a credibility space. We define

0, if θ = θ1 1, if θ = θ1
ξ1 (θ) = ξ2 (θ) =
1, if θ = θ2 , 0, if θ = θ2 .
It is clear that both of them are fuzzy variables and have the same member-
ship function, µ(x) ≡ 1 on x = 0 or 1.
Theorem 2.16 (Credibility Inversion Theorem) Let ξ be a fuzzy variable
with membership function µ. Then for any set B of real numbers, we have

1
Cr{ξ ∈ B} = sup µ(x) + 1 − sup µ(x) . (2.31)
2 x∈B x∈B c
Proof: If Cr{ξ ∈ B} ≤ 0.5, then by Axiom 2, we have Cr{ξ = x} ≤ 0.5 for

each x ∈ B. It follows from Axiom 4 that

1 1
Cr{ξ ∈ B} = sup (2Cr{ξ = x} ∧ 1) = sup µ(x). (2.32)
2 x∈B 2 x∈B
The self-duality of credibility measure implies that Cr{ξ ∈ B c } ≥ 0.5 and
supx∈B c Cr{ξ = x} ≥ 0.5, i.e.,
sup µ(x) = sup (2Cr{ξ = x} ∧ 1) = 1. (2.33)
x∈B c x∈B c
It follows from (2.32) and (2.33) that (2.31) holds.

If Cr{ξ ∈ B} ≥ 0.5, then Cr{ξ ∈ B c } ≤ 0.5. It follows from the first case
that

c 1
Cr{ξ ∈ B} = 1 − Cr{ξ ∈ B } = 1 − sup µ(x) + 1 − sup µ(x)
2 x∈B c x∈B

1
= sup µ(x) + 1 − sup µ(x) .
2 x∈B x∈B c
Example 2.13: Let ξ be a fuzzy variable with membership function µ. Then

the following equations follow immediately from Theorem 2.16:
!
1
Cr{ξ = x} = µ(x) + 1 − sup µ(y) , ∀x ∈ <; (2.34)
2 y6=x

1
Cr{ξ ≤ x} = sup µ(y) + 1 − sup µ(y) , ∀x ∈ <; (2.35)
2 y≤x y>x

1
Cr{ξ ≥ x} = sup µ(y) + 1 − sup µ(y) , ∀x ∈ <. (2.36)
2 y≥x y<x
Especially, if µ is a continuous function, then
µ(x)
Cr{ξ = x} = , ∀x ∈ <. (2.37)
2
Theorem 2.17 (Sufficient and Necessary Condition for Membership Func-
tion) A function µ : < → [0, 1] is a membership function if and only if
sup µ(x) = 1.
Proof: If µ is a membership function, then there exists a fuzzy variable ξ

whose membership function is just µ, and
sup µ(x) = sup (2Cr{ξ = x}) ∧ 1.

x∈< x∈<
If there is some point x ∈ < such that Cr{ξ = x} ≥ 0.5, then sup µ(x) = 1.
Otherwise, we have Cr{ξ = x} < 0.5 for each x ∈ <. It follows from Axiom 4
that
sup µ(x) = sup (2Cr{ξ = x}) ∧ 1 = 2 sup Cr{ξ = x} = 2 (Cr{Θ} ∧ 0.5) = 1.

x∈< x∈< x∈<
Conversely, suppose that sup µ(x) = 1. For each x ∈ <, we define

!
1
Cr{x} = µ(x) + 1 − sup µ(y) .
2 y6=x
It is clear that
1
sup Cr{x} ≥ (1 + 1 − 1) = 0.5.
x∈< 2
For any x ∈ < with Cr{x∗ } ≥ 0.5, we have µ(x∗ ) = 1 and
∗
Cr{x∗ } + sup Cr{y}

y6=x∗
! !
1 ∗ 1
= µ(x ) + 1 − sup µ(y) + sup µ(y) + 1 − sup µ(z)
2 y6=x∗ y6=x∗ 2 z6=y
1 1
= 1− sup µ(y) + sup µ(y) = 1.
2 y6=x∗ 2 y6=x∗
Thus Cr{x} satisfies the credibility extension condition, and has a unique
extension to credibility measure on P(<) by using the credibility extension
Section 2.3 - Membership Function 61
theorem. Now we define a fuzzy variable ξ as an identity function from the

credibility space (<, P(<), Cr) to <. Then the membership function of the
fuzzy variable ξ is
!
(2Cr{ξ = x}) ∧ 1 = µ(x) + 1 − sup µ(y) ∧ 1 = µ(x)
y6=x
for each x. The theorem is proved.
Remark 2.4: Theorem 2.17 states that the identity function is a universal
function for any fuzzy variable by defining an appropriate credibility space.
Theorem 2.18 A fuzzy variable ξ with membership function µ is

(a) nonnegative if and only if µ(x) = 0 for all x < 0;
(b) positive if and only if µ(x) = 0 for all x ≤ 0;
(c) simple if and only if µ takes nonzero values at a finite number of points;
(d) discrete if and only if µ takes nonzero values at a countable set of points;
(e) continuous if and only if µ is a continuous function.
Proof: The theorem is obvious since the membership function µ(x) =

(2Cr{ξ = x}) ∧ 1 for each x ∈ <.
Some Special Membership Functions
By an equipossible fuzzy variable we mean the fuzzy variable fully determined

by the pair (a, b) of crisp numbers with a < b, whose membership function is
given by
(
1, if a ≤ x ≤ b
µ1 (x) =
0, otherwise.
By a triangular fuzzy variable we mean the fuzzy variable fully determined

by the triplet (a, b, c) of crisp numbers with a < b < c, whose membership
function is given by
 x−a
 , if a ≤ x ≤ b
 b−a



µ2 (x) = x−c
, if b ≤ x ≤ c
 b−c




0, otherwise.
By a trapezoidal fuzzy variable we mean the fuzzy variable fully determined

by the quadruplet (a, b, c, d) of crisp numbers with a < b < c < d, whose
membership function is given by
x−a


 , if a ≤ x ≤ b



 b−a


 1, if b ≤ x ≤ c
µ3 (x) =
x−d
if c ≤ x ≤ d


 ,



 c−d

 0, otherwise.
µ1 (x) µ2 (x) µ3 (x)

... ... ...
.......... ......... .........
... . . . . ......................................................... . . . . . . . . . . . ...... . . . . . . . . . . ... . . . . . . . . . . . . . . . . . . .... . . . . . . . . .....................................
1 ...
...
.
.
.
.
.
.
....
..
..
..
......
...
...
...
....
...
.....
... . . ... .. . ... ... ... . ...
... . . .
. .
... . .... .
. .. .
.
.
. ..
. . . . . ..
... . . .... ... . .... .... ... . . ....
... . . ... ... . .... ... ... . . ...
... .
.
.
. ... ... .. .... ... ... . . ..
. .
... . . ... ... . ... .
. .. ..
. . .....
... . . .
. ..
. . ... ..
. ... . . ..
... . . .
.
. .. . ... .
.
. . .
.
. . ...
. . . . ... . . ...
... . . .... ... . ...
.
.
.
. .
.
.
. . . ...
... . . .
. ..
. . .
... .
. .
. . . ...
... . . .
.
. .. . .
.
. .
.
. . . ...
. . . . ... . .
... . . ... .... ..
. . . .
.
. .
. . . .
x x
............................................................................................................. .................................................................................................................................
.
x
0 a
..............................................................................................
... a .... c ....
a b c d
... b ... b ...
Figure 2.1: Membership Functions µ1 , µ2 and µ3
Joint Membership Function

Definition 2.11 If ξ = (ξ1 , ξ2 , · · · , ξn ) is a fuzzy vector on the credibility
space (Θ, P, Cr). Then its joint membership function is derived from the
credibility measure by
µ(x) = (2Cr{ξ = x}) ∧ 1, ∀x ∈ <n . (2.38)
Theorem 2.19 (Sufficient and Necessary Condition for Joint Membership

Function) A function µ : <n → [0, 1] is a joint membership function if and
only if sup µ(x) = 1.
Proof: Like Theorem 2.17.
2.4 Credibility Distribution

Definition 2.12 (Liu [130]) The credibility distribution Φ : < → [0, 1] of a
fuzzy variable ξ is defined by

Φ(x) = Cr θ ∈ Θ ξ(θ) ≤ x . (2.39)
Section 2.4 - Credibility Distribution 63
That is, Φ(x) is the credibility that the fuzzy variable ξ takes a value less than
or equal to x. Generally speaking, the credibility distribution Φ is neither
left-continuous nor right-continuous.
Example 2.14: The credibility distribution of an equipossible fuzzy variable

(a, b) is 
 0, if x < a

Φ1 (x) = 1/2, if a ≤ x < b

1, if x ≥ b.

Especially, if ξ is an equipossible fuzzy variable on <, then Φ1 (x) ≡ 1/2.
Example 2.15: The credibility distribution of a triangular fuzzy variable

(a, b, c) is 

 0, if x≤a
x − a


a≤x≤b

 2(b − a) , if


Φ2 (x) =
x + c − 2b
, if b≤x≤c


2(c − b)





1, if x ≥ c.

Example 2.16: The credibility distribution of a trapezoidal fuzzy variable

(a, b, c, d) is 

 0, if x ≤ a
x − a


, if a ≤ x ≤ b


 2(b − a)




 1
Φ3 (x) = , if b ≤ x ≤ c

 2
x + d − 2c


, if c ≤ x ≤ d






 2(d − c)
1, if x ≥ d.

Φ1 (x) Φ2 (x) Φ3 (x)

... ... ...
.......... ......... .........
... . . . . . . . . . . . . . ............................ . . . . . . . . . ...... . . . . . . . . . . . . . . . . . ............. . . . . . . . . . .... . . . . . . . . . . . . . . . . . . . . . . .............
1 ...
...
.
.
.
.
....
..
....
. ...
...
...
....
... . ... .. . ... .. .
...
. .
. .
... . .
. .
... .
. . . . .
... . .... ... .. .... ... ..
... .
. .. ... . .. ... .
... . . . . ........................................ . . . . . . . . . . . . . . . . ...... . . . . . . . . . . . . ........ . . ... . . . . . . . . . . . .... . . . . . . . . ......................................... .
0.5 ...
...
.
.
.
.
.
.
.... ..... .
..
..
.... .
.
.
.
.
....
.
.
...
. .
.
.
.
...
. . .
. ...
.. . .
. .
... . . .
. . ... . . . .. . .
... . . ..
. ...
. . . .
.
. .. . .
... . . .
.
. ...
.... . . .
.
. .
... . . .
...
. . .
.
. ...
.... . . .
.
. ... . . .
. . ... . . . . .
................................................................................................. ..
. ...
. . . .
.
. .
.. . . .
x ...............................................................................................................
x ............................................................................................................................... x
0 a
.. ...
....
a ....
...
... b ... b c ... a b c d
Figure 2.2: Credibility Distributions Φ1 , Φ2 and Φ3

Theorem 2.20 Let ξ be a fuzzy variable with membership function µ. Then

its credibility distribution is

1
Φ(x) = sup µ(y) + 1 − sup µ(y) , ∀x ∈ <. (2.40)
2 y≤x y>x
Proof: It follows from the credibility inversion theorem immediately.

Theorem 2.21 (Liu [135], Sufficient and Necessary Condition for Credibil-
ity Distribution) A function Φ : < → [0, 1] is a credibility distribution if and
only if it is an increasing function with
lim Φ(x) ≤ 0.5 ≤ lim Φ(x), (2.41)
x→−∞ x→∞
lim Φ(y) = Φ(x) if lim Φ(y) > 0.5 or Φ(x) ≥ 0.5. (2.42)
y↓x y↓x
Proof: It is obvious that a credibility distribution Φ is an increasing func-

tion. The inequalities (2.41) follow from the credibility asymptotic theorem
immediately. Assume that x is a point at which limy↓x Φ(y) > 0.5. That is,
lim Cr{ξ ≤ y} > 0.5.
y↓x
Since {ξ ≤ y} ↓ {ξ ≤ x} as y ↓ x, it follows from the credibility semicontinuity

law that
Φ(y) = Cr{ξ ≤ y} ↓ Cr{ξ ≤ x} = Φ(x)
as y ↓ x. When x is a point at which Φ(x) ≥ 0.5, if limy↓x Φ(y) 6= Φ(x), then
we have
lim Φ(y) > Φ(x) ≥ 0.5.
y↓x
For this case, we have proved that limy↓x Φ(y) = Φ(x). Thus (2.41) and
(2.42) are proved.
Conversely, if Φ : < → [0, 1] is an increasing function satisfying (2.41) and
(2.42), then


 2Φ(x), if Φ(x) < 0.5


µ(x) = 1, if lim Φ(y) < 0.5 ≤ Φ(x) (2.43)
y↑x

 2 − 2Φ(x), if 0.5 ≤ lim Φ(y)


y↑x
takes values in [0, 1] and sup µ(x) = 1. It follows from Theorem 2.17 that
there is a fuzzy variable ξ whose membership function is just µ. Let us verify
that Φ is the credibility distribution of ξ, i.e., Cr{ξ ≤ x} = Φ(x) for each x.
The argument breaks down into two cases. (i) If Φ(x) < 0.5, then we have
supy>x µ(y) = 1, and µ(y) = 2Φ(y) for each y with y ≤ x. Thus

1
Cr{ξ ≤ x} = sup µ(y) + 1 − sup µ(y) = sup Φ(y) = Φ(x).
2 y≤x y>x y≤x
Section 2.4 - Credibility Distribution 65
(ii) If Φ(x) ≥ 0.5, then we have supy≤x µ(y) = 1 and Φ(y) ≥ Φ(x) ≥ 0.5 for
each y with y > x. Thus µ(y) = 2 − 2Φ(y) and

1
Cr{ξ ≤ x} = sup µ(y) + 1 − sup µ(y)
2 y≤x y>x

1
= 1 + 1 − sup(2 − 2Φ(y))
2 y>x
= inf Φ(y) = lim Φ(y) = Φ(x).

y>x y↓x
Example 2.17: Let a and b be two numbers with 0 ≤ a ≤ 0.5 ≤ b ≤ 1. We

define a fuzzy variable by the following membership function,

 2a,
 if x < 0
µ(x) = 1, if x = 0

2 − 2b, if x > 0.

Then its credibility distribution is

(
a, if x < 0
Φ(x) =
b, if x ≥ 0.
Thus we have
lim Φ(x) = a, lim Φ(x) = b.
x→−∞ x→+∞
Theorem 2.22 A fuzzy variable with credibility distribution Φ is

(a) nonnegative if and only if Φ(x) = 0 for all x < 0;
(b) positive if and only if Φ(x) = 0 for all x ≤ 0.
Proof: It follows immediately from the definition.
Theorem 2.23 Let ξ be a fuzzy variable. Then we have
(a) if ξ is simple, then its credibility distribution is a simple function;
(b) if ξ is discrete, then its credibility distribution is a step function;
(c) if ξ is continuous on the real line <, then its credibility distribution is a
continuous function.
Proof: The parts (a) and (b) follow immediately from the definition. The
part (c) follows from Theorem 2.20 and the continuity of the membership
function.
Example 2.18: However, the inverse of Theorem 2.23 is not true. For
example, let ξ be a fuzzy variable whose membership function is
(
x, if 0 ≤ x ≤ 1
µ(x) =
1, otherwise.
Then its credibility distribution is Φ(x) ≡ 0.5. It is clear that Φ(x) is simple
and continuous. But the fuzzy variable ξ is neither simple nor continuous.
Definition 2.13 A continuous fuzzy variable is said to be (a) singular if its

credibility distribution is a singular function; (b) absolutely continuous if its
credibility distribution is absolutely continuous.
Definition 2.14 (Liu [130]) The credibility density function φ: < → [0, +∞)
of a fuzzy variable ξ is a function such that
Z x
Φ(x) = φ(y)dy, ∀x ∈ <, (2.44)
−∞
Z +∞
φ(y)dy = 1 (2.45)
−∞
where Φ is the credibility distribution of the fuzzy variable ξ.
Example 2.19: The credibility density function of a triangular fuzzy vari-

able (a, b, c) is
1

 , if a ≤ x ≤ b
− a)



 2(b
φ(x) = 1
, if b ≤ x ≤ c
2(c − b)





0, otherwise.
Example 2.20: The credibility density function of a trapezoidal fuzzy vari-

able (a, b, c, d) is
1

 , if a ≤ x ≤ b
− a)



 2(b
φ(x) = 1
, if c ≤ x ≤ d
2(d − c)





0, otherwise.
Example 2.21: The credibility density function of an equipossible fuzzy

variable (a, b) does not exist.
Example 2.22: The credibility density function does not necessarily exist
even if the membership function is continuous and unimodal with a finite
support. Let f be the Cantor function, and set

 f (x),
 if 0 ≤ x ≤ 1
µ(x) = f (2 − x), if 1 < x ≤ 2 (2.46)

0, otherwise.

Then µ is a continuous and unimodal function with µ(1) = 1. Hence µ

is a membership function. However, its credibility distribution is not an
absolutely continuous function. Thus the credibility density function does
not exist.
Theorem 2.24 Let ξ be a fuzzy variable whose credibility density function

φ exists. Then we have
Z x Z +∞
Cr{ξ ≤ x} = φ(y)dy, Cr{ξ ≥ x} = φ(y)dy. (2.47)
−∞ x
Proof: The first part follows immediately from the definition. In addition,
by the self-duality of credibility measure, we have
Z +∞ Z x Z +∞
Cr{ξ ≥ x} = 1 − Cr{ξ < x} = φ(y)dy − φ(y)dy = φ(y)dy.
−∞ −∞ x
Example 2.23: Different from the random case, generally speaking,

Z b
Cr{a ≤ ξ ≤ b} =
6 φ(y)dy.
a
Consider the trapezoidal fuzzy variable ξ = (1, 2, 3, 4). Then Cr{2 ≤ ξ ≤

3} = 0.5. However, it is obvious that φ(x) = 0 when 2 ≤ x ≤ 3 and
Z 3
φ(y)dy = 0 6= 0.5 = Cr{2 ≤ ξ ≤ 3}.
2
Joint Credibility Distribution

Definition 2.15 Let (ξ1 , ξ2 , · · · , ξn ) be a fuzzy vector. Then the joint credi-
bility distribution Φ : <n → [0, 1] is defined by

Φ(x1 , x2 , · · · , xn ) = Cr θ ∈ Θ ξ1 (θ) ≤ x1 , ξ2 (θ) ≤ x2 , · · · , ξn (θ) ≤ xn .
Definition 2.16 The joint credibility density function φ : <n → [0, +∞) of
a fuzzy vector (ξ1 , ξ2 , · · · , ξn ) is a function such that
Z x1 Z x2 Z xn
−∞ −∞ −∞
holds for all (x1 , x2 , · · · , xn ) ∈ <n , and

Z +∞ Z +∞ Z +∞
··· φ(y1 , y2 , · · · , yn )dy1 dy2 · · · dyn = 1
−∞ −∞ −∞
where Φ is the joint credibility distribution of the fuzzy vector (ξ1 , ξ2 , · · · , ξn ).

2.5 Independence
The independence of fuzzy variables has been discussed by many authors
from different angles. Here we use the following condition.
Definition 2.17 (Liu and Gao [158]) The fuzzy variables ξ1 , ξ2 , · · · , ξm are
said to be independent if
(m )
\
Cr {ξi ∈ Bi } = min Cr {ξi ∈ Bi } (2.48)
1≤i≤m
i=1
for any sets B1 , B2 , · · · , Bm of <.

Theorem 2.25 The fuzzy variables ξ1 , ξ2 , · · · , ξm are independent if and
only if (m )
[
Cr {ξi ∈ Bi } = max Cr {ξi ∈ Bi } (2.49)
1≤i≤m
i=1
for any sets B1 , B2 , · · · , Bm of <.
Proof: It follows from the self-duality of credibility measure that ξ1 , ξ2 , · · · , ξm
are independent if and only if
(m ) (m )
[ \
c
Cr {ξi ∈ Bi } = 1 − Cr {ξi ∈ Bi }
i=1 i=1
= 1 − min Cr{ξi ∈ Bic } = max Cr {ξi ∈ Bi } .

1≤i≤m 1≤i≤m
Thus (2.49) is verified. The proof is complete.

Theorem 2.26 The fuzzy variables ξ1 , ξ2 , · · · , ξm are independent if and
only if (m )
\
Cr {ξi = xi } = min Cr {ξi = xi } (2.50)
1≤i≤m
i=1
for any real numbers x1 , x2 , · · · , xm .
Proof: If ξ1 , ξ2 , · · · , ξm are independent, then we have (2.50) immediately
by taking Bi = {xi } for each i. Conversely, if Cr{∩m i=1 (ξi ∈ Bi )} ≥ 0.5, it
follows from Theorem 2.2 that (2.48) holds. Otherwise, we have Cr{∩m i=1 (ξi =
xi )} < 0.5 for any real numbers xi ∈ Bi , i = 1, 2, · · · , m, and
(m )  
\  [ \m 
Cr {ξi ∈ Bi } = Cr {ξi = xi }
 
i=1 xi ∈Bi ,1≤i≤m i=1
(m )
\
= sup Cr {ξi = xi } = sup min Cr{ξi = xi }
xi ∈Bi ,1≤i≤m i=1 xi ∈Bi ,1≤i≤m 1≤i≤m
= min sup Cr {ξi = xi } = min Cr {ξi ∈ Bi } .

1≤i≤m xi ∈Bi 1≤i≤m
Hence (2.48) is true, and ξ1 , ξ2 , · · · , ξm are independent. The theorem is thus

proved.
Theorem 2.27 Let µi be membership functions of fuzzy variables ξi , i =

1, 2, · · · , m, respectively, and µ the joint membership function of fuzzy vector
(ξ1 , ξ2 , · · · , ξm ). Then the fuzzy variables ξ1 , ξ2 , · · · , ξm are independent if
and only if
µ(x1 , x2 , · · · , xm ) = min µi (xi ) (2.51)
1≤i≤m
Proof: Suppose that ξ1 , ξ2 , · · · , ξm are independent. It follows from Theo-

rem 2.26 that
(m )!
\
µ(x1 , x2 , · · · , xm ) = 2Cr {ξi = xi } ∧1
i=1

= 2 min Cr{ξi = xi } ∧ 1
1≤i≤m
= min (2Cr{ξi = xi }) ∧ 1 = min µi (xi ).

1≤i≤m 1≤i≤m
Conversely, for any real numbers x1 , x2 , · · · , xm with Cr{∩mi=1 {ξi = xi }} <

0.5, we have
(m ) (m )!
\ 1 \
Cr {ξi = xi } = 2Cr {ξi = xi } ∧1
i=1
2 i=1
1 1
= µ(x1 , x2 , · · · , xm ) = min µi (xi )
2 2 1≤i≤m

1
= min (2Cr {ξi = xi }) ∧ 1
2 1≤i≤m
= min Cr {ξi = xi } .
1≤i≤m
It follows from Theorem 2.26 that ξ1 , ξ2 , · · · , ξm are independent. The theo-

rem is proved.
Theorem 2.28 Let Φi be credibility distributions of fuzzy variables ξi , i =

1, 2, · · · , m, respectively, and Φ the joint credibility distribution of fuzzy vector
(ξ1 , ξ2 , · · · , ξm ). If ξ1 , ξ2 , · · · , ξm are independent, then we have
Φ(x1 , x2 , · · · , xm ) = min Φi (xi ) (2.52)

1≤i≤m

Proof: Since ξ1 , ξ2 , · · · , ξm are independent fuzzy variables, we have

(m )
\
Φ(x1 , x2 , · · · , xm ) = Cr {ξi ≤ xi } = min Cr{ξi ≤ xi } = min Φi (xi )
1≤i≤m 1≤i≤m
i=1
for any real numbers x1 , x2 , · · · , xm . The theorem is proved.
Example 2.24: However, the equation (2.52) does not imply that the fuzzy
variables are independent. For example, let ξ be a fuzzy variable with credi-
bility distribution Φ. Then the joint credibility distribution Ψ of fuzzy vector
(ξ, ξ) is
Ψ(x1 , x2 ) = Cr{ξ ≤ x1 , ξ ≤ x2 } = Cr{ξ ≤ x1 } ∧ Cr{ξ ≤ x2 } = Φ(x1 ) ∧ Φ(x2 )
for any real numbers x1 and x2 . But, generally speaking, a fuzzy variable is
not independent with itself.
Theorem 2.29 Let ξ1 , ξ2 , · · · , ξm be independent fuzzy variables, and f1 , f2 ,

· · · , fn are real-valued functions. Then f1 (ξ1 ), f2 (ξ2 ), · · · , fm (ξm ) are inde-
pendent fuzzy variables.
Proof: For any sets B1 , B2 , · · · , Bm of <, we have

(m ) (m )
\ \
Cr {fi (ξi ) ∈ Bi } = Cr {ξi ∈ fi−1 (Bi )}
i=1 i=1
= min Cr{ξi ∈ fi−1 (Bi )} = min Cr{fi (ξi ) ∈ Bi }.

1≤i≤m 1≤i≤m
Thus f1 (ξ1 ), f2 (ξ2 ), · · · , fm (ξm ) are independent fuzzy variables.

Definition 2.18 (Liu [135]) The fuzzy variables ξ and η are said to be iden-
tically distributed if
Cr{ξ ∈ B} = Cr{η ∈ B} (2.53)
for any set B of <.
Theorem 2.30 The fuzzy variables ξ and η are identically distributed if and
only if ξ and η have the same membership function.
Proof: Let µ and ν be the membership functions of ξ and η, respectively. If

ξ and η are identically distributed fuzzy variables, then, for any x ∈ <, we
have
µ(x) = (2Cr{ξ = x}) ∧ 1 = (2Cr{η = x}) ∧ 1 = ν(x).
Thus ξ and η have the same membership function.
Section 2.7 - Extension Principle of Zadeh 71
Conversely, if ξ and η have the same membership function, i.e., µ(x) ≡

ν(x), then, by using the credibility inversion theorem, we have

1
Cr{ξ ∈ B} = sup µ(x) + 1 − sup µ(x)
2 x∈B x∈B c

1
= sup ν(x) + 1 − sup ν(x) = Cr{η ∈ B}
2 x∈B x∈B c
for any set B of <. Thus ξ and η are identically distributed fuzzy variables.
Theorem 2.31 The fuzzy variables ξ and η are identically distributed if and
only if Cr{ξ = x} = Cr{η = x} for each x ∈ <.
Proof: If ξ and η are identically distributed fuzzy variables, then we im-
mediately have Cr{ξ = x} = Cr{η = x} for each x. Conversely, it follows
from
µ(x) = (2Cr{ξ = x}) ∧ 1 = (2Cr{η = x}) ∧ 1 = ν(x)
that ξ and η have the same membership function. Thus ξ and η are identically
distributed fuzzy variables.
Theorem 2.32 If ξ and η are identically distributed fuzzy variables, then ξ
and η have the same credibility distribution.
Proof: If ξ and η are identically distributed fuzzy variables, then, for any
x ∈ <, we have Cr{ξ ∈ (−∞, x]} = Cr{η ∈ (−∞, x]}. Thus ξ and η have the
same credibility distribution.
Example 2.25: The inverse of Theorem 2.32 is not true. We consider two
fuzzy variables with the following membership functions,
 
 1.0, if x = 0
  1.0, if x = 0

µ(x) = 0.6, if x = 1 ν(x) = 0.7, if x = 1
 
0.8, if x = 2, 0.8, if x = 2.
 
It is easy to verify that ξ and η have the same credibility distribution,


 0, if x < 0

Φ(x) = 0.6, if 0 ≤ x < 2

1, if x ≥ 2.

However, they are not identically distributed fuzzy variables.

Theorem 2.33 Let ξ and η be two fuzzy variables whose credibility density
functions exist. If ξ and η are identically distributed, then they have the same
credibility density function.
Proof: It follows from Theorem 2.32 that the fuzzy variables ξ and η have the
same credibility distribution. Hence they have the same credibility density
function.
2.7 Extension Principle of Zadeh

Theorem 2.34 (Extension Principle of Zadeh) Let ξ1 , ξ2 , · · · , ξn be inde-
pendent fuzzy variables with membership functions µ1 , µ2 , · · · , µn , respec-
tively, and f : <n → < a function. Then the membership function µ of
ξ = f (ξ1 , ξ2 , · · · , ξn ) is derived from the membership functions µ1 , µ2 , · · · , µn
by
µ(x) = sup min µi (xi ) (2.54)
x=f (x1 ,x2 ,··· ,xn ) 1≤i≤n
for any x ∈ <. Here we set µ(x) = 0 if there are not real numbers x1 , x2 , · · · , xn
such that x = f (x1 , x2 , · · · , xn ).
Proof: It follows from Definition 2.10 that the membership function of ξ =

f (ξ1 , ξ2 , · · · , ξn ) is
µ(x) = (2Cr {f (ξ1 , ξ2 , · · · , ξn ) = x}) ∧ 1

  
 [ 
= 2Cr {ξ1 = x1 , ξ2 = x2 , · · · , ξn = xn }  ∧ 1
 
x=f (x1 ,x2 ,··· ,xn )
!
= 2 sup Cr{ξ1 = x1 , ξ2 = x2 , · · · , ξn = xn } ∧1
x=f (x1 ,x2 ,··· ,xn )
!
= 2 sup min Cr{ξi = xi } ∧1 (by independence)
x=f (x1 ,x2 ,··· ,xn ) 1≤k≤n
= sup min (2Cr{ξi = xi }) ∧ 1

x=f (x1 ,x2 ,··· ,xn ) 1≤k≤n
= sup min µi (xi ).

x=f (x1 ,x2 ,··· ,xn ) 1≤i≤n
Remark 2.5: The extension principle of Zadeh is only applicable to the

operations on independent fuzzy variables. In the past literature, the exten-
sion principle is used as a postulate. However, it is treated as a theorem in
credibility theory.
Example 2.26: The sum of independent equipossible fuzzy variables ξ =

(a1 , a2 ) and η = (b1 , b2 ) is also an equipossible fuzzy variable, and
ξ + η = (a1 + b1 , a2 + b2 ).
Their product is also an equipossible fuzzy variable, and

ξ·η = min xy, max xy .
a1 ≤x≤a2 ,b1 ≤y≤b2 a1 ≤x≤a2 ,b1 ≤y≤b2
Example 2.27: The sum of independent triangular fuzzy variables ξ =

(a1 , a2 , a3 ) and η = (b1 , b2 , b3 ) is also a triangular fuzzy variable, and
ξ + η = (a1 + b1 , a2 + b2 , a3 + b3 ).
The product of a triangular fuzzy variable ξ = (a1 , a2 , a3 ) and a scalar number
λ is (
(λa1 , λa2 , λa3 ), if λ ≥ 0
λ·ξ =
(λa3 , λa2 , λa1 ), if λ < 0.
That is, the product of a triangular fuzzy variable and a scalar number is
also a triangular fuzzy variable. However, the product of two triangular fuzzy
variables is not a triangular one.
Example 2.28: The sum of independent trapezoidal fuzzy variables ξ =

(a1 , a2 , a3 , a4 ) and η = (b1 , b2 , b3 , b4 ) is also a trapezoidal fuzzy variable, and
ξ + η = (a1 + b1 , a2 + b2 , a3 + b3 , a4 + b4 ).
The product of a trapezoidal fuzzy variable ξ = (a1 , a2 , a3 , a4 ) and a scalar
number λ is (
(λa1 , λa2 , λa3 , λa4 ), if λ ≥ 0
λ·ξ =
(λa4 , λa3 , λa2 , λa1 ), if λ < 0.
That is, the product of a trapezoidal fuzzy variable and a scalar number is
also a trapezoidal fuzzy variable. However, the product of two trapezoidal
fuzzy variables is not a trapezoidal one.
Example 2.29: Let ξ1 , ξ2 , · · · , ξn be independent fuzzy variables with mem-

bership functions µ1 , µ2 , · · · , µn , respectively, and f : <n → < a function.
Then for any set B of real numbers, the credibility Cr{f (ξ1 , ξ2 , · · · , ξn ) ∈ B}
is
!
1
sup min µi (xi ) + 1 − sup min µi (xi ) .
2 f (x1 ,x2 ,··· ,xn )∈B 1≤i≤n f (x1 ,x2 ,··· ,xn )∈B c 1≤i≤n
2.8 Expected Value

There are many ways to define an expected value operator for fuzzy variables.
The most general definition of expected value operator of fuzzy variable was
given by Liu and Liu [132]. This definition is applicable to not only continuous
fuzzy variables but also discrete ones.
Definition 2.19 (Liu and Liu [132]) Let ξ be a fuzzy variable. Then the
expected value of ξ is defined by
Z +∞ Z 0
E[ξ] = Cr{ξ ≥ r}dr − Cr{ξ ≤ r}dr (2.55)
0 −∞

Example 2.30: Let ξ be the equipossible fuzzy variable (a, b). If a ≥ 0,

then Cr{ξ ≤ r} ≡ 0 when r < 0, and

 1, if
 r≤a
Cr{ξ ≥ r} = 0.5, if a<r≤b

0, if r > b,

!
Z a Z b Z +∞ Z 0
a+b
E[ξ] = 1dr + 0.5dr + 0dr − 0dr = .
0 a b −∞ 2
If b ≤ 0, then Cr{ξ ≥ r} ≡ 0 when r > 0, and


 1,
 if r ≥ b
Cr{ξ ≤ r} = 0.5, if a ≤ r < b

0, if r < a,

!
Z +∞ Z a Z b Z 0
a+b
E[ξ] = 0dr − 0dr + 0.5dr + 1dr = .
0 −∞ a b 2
If a < 0 < b, then
(
0.5, if 0 ≤ r ≤ b
Cr{ξ ≥ r} =
0, if r > b,
(
0, if r < a
Cr{ξ ≤ r} =
0.5, if a ≤ r ≤ 0,
!
Z b Z +∞ Z a Z 0
a+b
E[ξ] = 0.5dr + 0dr − 0dr + 0.5dr = .
0 b −∞ a 2
Thus we always have the expected value (a + b)/2.
Example 2.31: The triangular fuzzy variable ξ = (a, b, c) has an expected

value E[ξ] = (a + 2b + c)/4.
Example 2.32: The trapezoidal fuzzy variable ξ = (a, b, c, d) has an ex-

pected value E[ξ] = (a + b + c + d)/4.
Example 2.33: Let ξ be a continuous fuzzy variable with membership func-

tion µ. If its expected value exists, and there is a point x0 such that µ(x) is in-
creasing on (−∞, x0 ) and decreasing on (x0 , +∞), then Cr{ξ ≥ x} = µ(x)/2
for any x > x0 and Cr{ξ ≤ x} = µ(x)/2 for any x < x0 . Thus
Z +∞ Z x0
1 1
E[ξ] = x0 + µ(x)dx − µ(x)dx.
2 x0 2 −∞
Example 2.34: Let ξ be a fuzzy variable with membership function


 0, if x < 0

µ(x) = x, if 0 ≤ x ≤ 1

1, if x > 1.

Then its expected value is +∞. If ξ is a fuzzy variable with membership

function

 1,
 if x < 0
µ(x) = 1 − x, if 0 ≤ x ≤ 1

0, if x > 1.

Then its expected value is −∞.
Example 2.35: The definition of expected value operator is also applicable

to discrete case. Assume that ξ is a simple fuzzy variable whose membership
function is given by


 µ1 , if x = x1
µ2 , if x = x2

µ(x) = (2.56)

 ···
µm , if x = xm

where x1 , x2 , · · · , xm are distinct numbers. Note that µ1 ∨ µ2 ∨ · · · ∨ µm = 1.

Definition 2.19 implies that the expected value of ξ is
m
X
E[ξ] = wi xi (2.57)
i=1
where the weights are given by

1
wi = max {µj |xj ≤ xi } − max {µj |xj < xi }
2 1≤j≤m 1≤j≤m

+ max {µj |xj ≥ xi } − max {µj |xj > xi }
1≤j≤m 1≤j≤m
for i = 1, 2, · · · , m. It is easy to verify that all wi ≥ 0 and the sum of all

weights is just 1.
Example 2.36: Consider the fuzzy variable ξ defined by (2.56). Suppose

x1 < x2 < · · · < xm and there exists an index k with 1 < k < m such that
µ1 ≤ µ2 ≤ · · · ≤ µk and µk ≥ µk+1 ≥ · · · ≥ µm .
Note that µk ≡ 1. Then the expected value is determined by (2.57) and the
weights are given by
 µ1
 , if i = 1
2





 µi − µi−1
, if i = 2, 3, · · · , k − 1


2




µk−1 + µk+1

wi = 1− , if i = k

 2
µi − µi+1


if i = k + 1, k + 2, · · · , m − 1


 ,


 2
µm


, if i = m.


2
Theorem 2.35 (Liu [130]) Let ξ be a fuzzy variable whose credibility density
function φ exists. If the Lebesgue integral
Z +∞
xφ(x)dx
−∞

E[ξ] = xφ(x)dx. (2.58)
−∞
Proof: It follows from the definition of expected value operator and Fubini
Theorem that
Z +∞ Z 0
E[ξ] = Cr{ξ ≥ r}dr − Cr{ξ ≤ r}dr
0 −∞
Z +∞ Z +∞ Z 0 Z r
0 r −∞ −∞
Z +∞ Z x Z 0 Z 0
0 0 −∞ x
Z +∞ Z 0
0 −∞
Z +∞
= xφ(x)dx.
−∞
Example 2.37: Let ξ be a fuzzy variable with credibility distribution Φ.

Generally speaking, Z +∞
E[ξ] 6= xdΦ(x).
−∞
For example, let ξ be a fuzzy variable with membership function


 0,
 if x < 0
µ(x) = x, if 0 ≤ x ≤ 1

1, if x > 1.

Then E[ξ] = +∞. However

Z +∞
1
xdΦ(x) = 6= +∞.
−∞ 4
Theorem 2.36 (Liu [135]) Let ξ be a fuzzy variable with credibility distri-
bution Φ. If
lim Φ(x) = 0, lim Φ(x) = 1
x→−∞ x→∞
and the Lebesgue-Stieltjes integral

Z +∞
xdΦ(x)
−∞

E[ξ] = xdΦ(x). (2.59)
−∞
R +∞
Proof: Since the Lebesgue-Stieltjes integral −∞ xdΦ(x) is finite, we imme-
diately have
Z y Z +∞ Z 0 Z 0
y→+∞ 0 0 y→−∞ y −∞
and Z +∞ Z y
y→+∞ y y→−∞ −∞
It follows from
Z +∞
xdΦ(x) ≥ y lim Φ(z) − Φ(y) = y (1 − Φ(y)) ≥ 0, for y > 0,
y z→+∞
Z y
xdΦ(x) ≤ y Φ(y) − lim Φ(z) = yΦ(y) ≤ 0, for y < 0
−∞ z→−∞
that
lim y (1 − Φ(y)) = 0, lim yΦ(y) = 0.
y→+∞ y→−∞
n−1
X Z y
xi (Φ(xi+1 ) − Φ(xi )) → xdΦ(x)
i=0 0
and
n−1
X Z y
(1 − Φ(xi+1 ))(xi+1 − xi ) → Cr{ξ ≥ r}dr
i=0 0
n−1
X n−1
X
xi (Φ(xi+1 ) − Φ(xi )) − (1 − Φ(xi+1 )(xi+1 − xi ) = y(Φ(y) − 1) → 0
i=0 i=0

Z +∞ Z +∞
Cr{ξ ≥ r}dr = xdΦ(x).
0 0

Z 0 Z 0
− Cr{ξ ≤ r}dr = xdΦ(x).
−∞ −∞
It follows that the equation (2.59) holds.
Linearity of Expected Value Operator

Theorem 2.37 (Liu and Liu [151]) Let ξ and η be independent fuzzy vari-
ables with finite expected values. Then for any numbers a and b, we have
E[aξ + bη] = aE[ξ] + bE[η]. (2.60)
b. If b ≥ 0, we have
Z ∞ Z 0
E[ξ + b] = Cr{ξ + b ≥ r}dr − Cr{ξ + b ≤ r}dr
0 −∞
Z ∞ Z 0
= Cr{ξ ≥ r − b}dr − Cr{ξ ≤ r − b}dr
0 −∞
Z b
= E[ξ] + (Cr{ξ ≥ r − b} + Cr{ξ < r − b}) dr
0
= E[ξ] + b.

Z 0
E[ξ + b] = E[ξ] − (Cr{ξ ≥ r − b} + Cr{ξ < r − b}) dr = E[ξ] + b.
b
Step 2: We prove that E[aξ] = aE[ξ] for any real number a. If a = 0,

then the equation E[aξ] = aE[ξ] holds trivially. If a > 0, we have
Z ∞ Z 0
E[aξ] = Cr{aξ ≥ r}dr − Cr{aξ ≤ r}dr
0 −∞
Z ∞ Z 0
n ro n ro
= Cr ξ ≥ dr − Cr ξ ≤ dr
0 a −∞ a
Z ∞ n Z 0
ro r n ro r
=a Cr ξ ≥ d −a Cr ξ ≤ d = aE[ξ].
0 a a −∞ a a
If a < 0, we have
Z ∞ Z 0
E[aξ] = Cr{aξ ≥ r}dr − Cr{aξ ≤ r}dr
0 −∞
Z ∞ Z 0
n ro n ro
= Cr ξ ≤ dr − Cr ξ ≥ dr
0 a −∞ a
Z ∞ n Z 0
ro r n ro r
=a Cr ξ ≥ d −a Cr ξ ≤ d = aE[ξ].
0 a a −∞ a a
Step 3: We prove that E[ξ + η] = E[ξ] + E[η] when both ξ and η are
simple fuzzy variables with the following membership functions,
 

 µ1 , if x = a1 
 ν1 , if x = b1
 
µ2 , if x = a2 ν2 , if x = b2
 
µ(x) = ν(x) =


 ··· 

 ···
µm , if x = am , νn , if x = bn .
 
Then ξ+η is also a simple fuzzy variable taking values ai +bj with membership
degrees µi ∧ νj , i = 1, 2, · · · , m, j = 1, 2, · · · , n, respectively. Now we define

1
wi0 = max {µk |ak ≤ ai } − max {µk |ak < ai }
2 1≤k≤m 1≤k≤m

+ max {µk |ak ≥ ai } − max {µk |ak > ai } ,
1≤k≤m 1≤k≤m

1
wj00 = max {νl |bl ≤ bi } − max {νl |bl < bi }
2 1≤l≤n 1≤l≤n

+ max {νl |bl ≥ bi } − max {νl |bl > bi } ,
1≤l≤n 1≤l≤n

1
wij = max {µk ∧ νl |ak + bl ≤ ai + bj }
2 1≤k≤m,1≤l≤n
− max {µk ∧ νl |ak + bl < ai + bj }

1≤k≤m,1≤l≤n
+ max {µk ∧ νl |ak + bl ≥ ai + bj }

1≤k≤m,1≤l≤n

− max {µk ∧ νl |ak + bl > ai + bj }
1≤k≤m,1≤l≤n
for i = 1, 2, · · · , m and j = 1, 2, · · · , n. It is also easy to verify that

n
X m
X
wi0 = wij , wj00 = wij
j=1 i=1
for i = 1, 2, · · · , m and j = 1, 2, · · · , n. If {ai }, {bj } and {ai + bj } are

sequences consisting of distinct elements, then
m
X n
X m X
X n
E[ξ] = ai wi0 , E[η] = bj wj00 , E[ξ + η] = (ai + bj )wij .
i=1 j=1 i=1 j=1
Thus E[ξ + η] = E[ξ] + E[η]. If not, we may give them a small perturbation
such that they are distinct, and prove the linearity by letting the perturbation
tend to zero.
Step 4: We prove that E[ξ + η] = E[ξ] + E[η] when ξ and η are fuzzy
variables such that
1
lim Cr{ξ ≤ y} ≤ ≤ Cr{ξ ≤ 0},
y↑0 2
(2.61)
1
lim Cr{η ≤ y} ≤ ≤ Cr{η ≤ 0}.
y↑0 2
We define simple fuzzy variables ξi via credibility distributions as follows,
k−1 k−1 k


i
, if ≤ Cr{ξ ≤ x} < , k = 1, 2, · · · , 2i−1
2i 2i



 2

Φi (x) = k k−1 k
i
, if ≤ Cr{ξ ≤ x} < , k = 2i−1 + 1, · · · , 2i
 2 2i 2i





1, if Cr{ξ ≤ x} = 1
for i = 1, 2, · · · Thus {ξi } is a sequence of simple fuzzy variables satisfying
Cr{ξi ≤ r} ↑ Cr{ξ ≤ r}, if r ≤ 0

Cr{ξi ≥ r} ↑ Cr{ξ ≥ r}, if r ≥ 0
as i → ∞. Similarly, we define simple fuzzy variables ηi via credibility

distributions as follows,
k−1 k−1 k


i
, if ≤ Cr{η ≤ x} < , k = 1, 2, · · · , 2i−1
2i 2i



 2

Ψi (x) = k k−1 k
i
, if ≤ Cr{η ≤ x} < , k = 2i−1 + 1, · · · , 2i
2 2i 2i





if Cr{η ≤ x} = 1

1,
for i = 1, 2, · · · Thus {ηi } is a sequence of simple fuzzy variables satisfying
Cr{ηi ≤ r} ↑ Cr{η ≤ r}, if r ≤ 0

Cr{ηi ≥ r} ↑ Cr{η ≥ r}, if r ≥ 0
as i → ∞. It is also clear that {ξi +ηi } is a sequence of simple fuzzy variables.

Furthermore, when r ≤ 0, it follows from (2.61) that
lim Cr{ξi + ηi ≤ r} = lim sup Cr{ξi ≤ x} ∧ Cr{ηi ≤ y}

i→∞ i→∞ x≤0,y≤0,x+y≤r
= sup lim Cr{ξi ≤ x} ∧ Cr{ηi ≤ y}

x≤0,y≤0,x+y≤r i→∞
= sup Cr{ξ ≤ x} ∧ Cr{η ≤ y}

x≤0,y≤0,x+y≤r
= Cr{ξ + η ≤ r}.
That is,
Cr{ξi + ηi ≤ r} ↑ Cr{ξ + η ≤ r}, if r ≤ 0.
Cr{ξi + ηi ≥ r} ↑ Cr{ξ + η ≥ r}, if r ≥ 0.
Since the expected values E[ξ] and E[η] exist, we have

Z +∞ Z 0
E[ξi ] = Cr{ξi ≥ r}dr − Cr{ξi ≤ r}dr
0 −∞
Z +∞ Z 0
→ Cr{ξ ≥ r}dr − Cr{ξ ≤ r}dr = E[ξ],
0 −∞
Z +∞ Z 0
E[ηi ] = Cr{ηi ≥ r}dr − Cr{ηi ≤ r}dr
0 −∞
Z +∞ Z 0
→ Cr{η ≥ r}dr − Cr{η ≤ r}dr = E[η],
0 −∞
Z +∞ Z 0
E[ξi + ηi ] = Cr{ξi + ηi ≥ r}dr − Cr{ξi + ηi ≤ r}dr
0 −∞
Z +∞ Z 0
→ Cr{ξ + η ≥ r}dr − Cr{ξ + η ≤ r}dr = E[ξ + η]
0 −∞
as i → ∞. It follows from Step 3 that E[ξ + η] = E[ξ] + E[η].

Step 5: We prove that E[ξ + η] = E[ξ] + E[η] when ξ and η are arbi-
trary fuzzy variables. Since they have finite expected values, there exist two
numbers c and d such that
1
lim Cr{ξ + c ≤ y} ≤ ≤ Cr{ξ + c ≤ 0},
y↑0 2
1
lim Cr{η + d ≤ y} ≤ ≤ Cr{η + d ≤ 0}.
y↑0 2
It follows from Steps 1 and 4 that
E[ξ + η] = E[(ξ + c) + (η + d) − c − d]
= E[(ξ + c) + (η + d)] − c − d
= E[ξ + c] + E[η + d] − c − d
= E[ξ] + c + E[η] + d − c − d
= E[ξ] + E[η].
Step 6: We prove that E[aξ + bη] = aE[ξ] + bE[η] for any real numbers
a and b. In fact, the equation follows immediately from Steps 2 and 5. The
theorem is proved.
Example 2.38: Theorem 2.37 does not hold if ξ and η are not independent.
For example, take (Θ, P, Cr) to be {θ1 , θ2 , θ3 } with Cr{θ1 } = 0.7, Cr{θ2 } =
0.3 and Cr{θ3 } = 0.2. The fuzzy variables are defined by
 
 1, if θ = θ1
  0, if θ = θ1

ξ1 (θ) = 0, if θ = θ2 ξ2 (θ) = 2, if θ = θ2
 
2, if θ = θ3 , 3, if θ = θ3 .
 
Then we have 
 1, if θ = θ1

(ξ1 + ξ2 )(θ) = 2, if θ = θ2

5, if θ = θ3 .

Thus E[ξ1 ] = 0.9, E[ξ2 ] = 0.8, and E[ξ1 + ξ2 ] = 1.9. This fact implies that
E[ξ1 + ξ2 ] > E[ξ1 ] + E[ξ2 ].

If the fuzzy variables are defined by

 
 0,
 if θ = θ1  0, if θ = θ1

η1 (θ) = 1, if θ = θ2 η2 (θ) = 3, if θ = θ2
 
2, if θ = θ3 , 1, if θ = θ3 .
 
Then we have 
 0, if θ = θ1

(η1 + η2 )(θ) = 4, if θ = θ2

3, if θ = θ3 .

Thus E[η1 ] = 0.5, E[η2 ] = 0.9, and E[η1 + η2 ] = 1.2. This fact implies that
E[η1 + η2 ] < E[η1 ] + E[η2 ].
Expected Value of Function of Fuzzy Variable

Let ξ be a fuzzy variable, and f : < → < a function. Then the expected
value of f (ξ) is
Z +∞ Z 0
E[f (ξ)] = Cr{f (ξ) ≥ r}dr − Cr{f (ξ) ≤ r}dr.
0 −∞
For random case, it has been proved that the expected value E[f (ξ)] is the
Lebesgue-Stieltjes integral of f (x) with respect to the probability distribution
Φ of ξ if the integral exists. However, generally speaking, it is not true for
fuzzy case.
Example 2.39: We consider a fuzzy variable ξ whose membership function

is given by 
 0.6, if − 1 ≤ x < 0

µ(x) = 1, if 0 ≤ x ≤ 1

0, otherwise.

Then the expected value E[ξ 2 ] = 0.5. However, the credibility distribution
of ξ is 

 0, if x < −1
0.3, if − 1 ≤ x < 0


Φ(x) =

 0.5, if 0 ≤ x < 1

1, if x ≥ 1


Z +∞
x2 dΦ(x) = (−1)2 × 0.3 + 02 × 0.2 + 12 × 0.5 = 0.8 6= E[ξ 2 ].
−∞
Theorem 2.38 (Zhu and Ji [276]) Let ξ be a fuzzy variable whose credibility
distribution Φ satisfies
lim Φ(x) = 0, lim Φ(x) = 1. (2.62)

x→−∞ x→∞
If f (x) is a monotone function such that the Lebesgue-Stieltjes integral

Z +∞
f (x)dΦ(x) (2.63)
−∞

Z +∞
E[f (ξ)] = f (x)dΦ(x). (2.64)
−∞
Proof: We first suppose that f (x) is a monotone increasing function. Since

the Lebesgue-Stieltjes integral (2.63) is finite, we immediately have
lim Cr{ξ ≥ y}f (y) = lim (1 − Φ(y))f (y) = 0, (2.65)

y→+∞ y→+∞
lim Cr{ξ ≤ y}f (y) = lim Φ(y)f (y) = 0. (2.66)

y→−∞ y→−∞
Assume that a and b are two real numbers such that a < 0 < b. The
integration by parts produces
Z b Z b Z f −1 (b)
−1
Cr{f (ξ) ≥ r}dr = Cr{ξ ≥ f (r)}dr = Cr{ξ ≥ y}df (y)
0 0 f −1 (0)
−1
Z f (b)
= Cr{ξ ≥ f −1 (b)}f (f −1 (b)) − f (y)dCr{ξ ≥ y}
f −1 (0)
Z f −1 (b)
−1 −1
= Cr{ξ ≥ f (b)}f (f (b)) + f (y)dΦ(y).
f −1 (0)
Using (2.65) and letting b → +∞, we obtain

Z +∞ Z +∞
Cr{f (ξ) ≥ r}dr = f (y)dΦ(y). (2.67)
0 f −1 (0)
In addition,
Z 0 Z 0 Z f −1 (0)
Cr{f (ξ) ≤ r}dr = Cr{ξ ≤ f −1 (r)}dr = Cr{ξ ≤ y}df (y)
a a f −1 (a)
Z f −1 (0)
= −Cr{ξ ≤ f −1 (a)}f (f −1 (a)) − f (y)dCr{ξ ≤ y}
f −1 (a)
Z f −1 (0)
= −Cr{ξ ≤ f −1 (a)}f (f −1 (a)) − f (y)dΦ(y).
f −1 (a)
Using (2.66) and letting a → −∞, we obtain

Z 0 Z f −1 (0)
Cr{f (ξ) ≤ r}dr = − f (y)dΦ(y). (2.68)
−∞ −∞

Z +∞ Z 0 Z +∞
E[f (ξ)] = Cr{f (ξ) ≥ r} − Cr{f (ξ) ≤ r}dr = f (y)dΦ(y).
0 −∞ −∞
If f (x) is a monotone decreasing function, then −f (x) is a monotone increas-

ing function. Hence
Z +∞ Z +∞
E[f (ξ)] = −E[−f (ξ)] = − −f (x)dΦ(x) = f (y)dΦ(y).
−∞ −∞
The theorem is verified.
Sum of a Fuzzy Number of Fuzzy Variables

Theorem 2.39 (Zhao and Liu [265]) Assume that {ξi } is a sequence of iid
fuzzy variables, and ñ is a positive fuzzy integer (i.e., a fuzzy variable taking
“positive integer” values) that is independent of the sequence {ξi }. Then we
have " ñ #
X
E ξi = E [ñξ1 ] . (2.69)
i=1
Proof: Since {ξi } is a sequence of iid fuzzy variables and ñ is independent

of {ξi }, we have
ñ

P
Cr ξi ≥ r = sup Cr{ñ = n} ∧ min Cr {ξi = xi }
i=1 n,x1 +x2 +···+xn ≥r 1≤i≤n

≥ sup Cr{ñ = n} ∧ min Cr {ξi = x}
nx≥r 1≤i≤n
= sup Cr{ñ = n} ∧ Cr{ξ1 = x}

nx≥r
= Cr {ñξ1 ≥ r} .
On the other hand, for any given ε > 0, there exists an integer n and real
numbers x1 , x2 , · · · , xn with x1 + x2 + · · · + xn ≥ r such that
( ñ )
X
Cr ξi ≥ r − ε ≤ Cr{ñ = n} ∧ Cr{ξi = xi }
i=1
for each i with 1 ≤ i ≤ n. Without loss of generality, we assume that nx1 ≥ r.

Then we have
( ñ )
X
Cr ξi ≥ r − ε ≤ Cr{ñ = n} ∧ Cr{ξ1 = x1 } ≤ Cr {ñξ1 ≥ r} .
i=1
Letting ε → 0, we get
( ñ )
X
Cr ξi ≥ r ≤ Cr {ñξ1 ≥ r} .
i=1
It follows that ( ñ )
X
Cr ξi ≥ r = Cr {ñξ1 ≥ r} .
i=1
Similarly, the above identity still holds if the symbol “≥” is replaced with
“≤”. Finally, by the definition of expected value operator, we have
" ñ
# ( ñ ) ( ñ )
X Z +∞ X Z 0 X
E ξi = Cr ξi ≥ r dr − Cr ξi ≤ r dr
i=1 0 i=1 −∞ i=1
Z +∞ Z 0
= Cr {ñξ1 ≥ r} dr − Cr {ñξ1 ≤ r} dr = E [ñξ1 ] .
0 −∞
2.9 Variance
Definition 2.20 (Liu and Liu [132]) Let ξ be a fuzzy variable with finite
expected value e. Then the variance of ξ is defined by V [ξ] = E[(ξ − e)2 ].
The variance of a fuzzy variable provides a measure of the spread of the

distribution around its expected value.
Example 2.40: Let ξ be an equipossible fuzzy variable (a, b). Then its
expected value is e = (a + b)/2, and for any positive number r, we have
(
1/2, if r ≤ (b − a)2 /4
Cr{(ξ − e)2 ≥ r} =
0, if r > (b − a)2 /4.
Thus the variance is

+∞ (b−a)2 /4
(b − a)2
Z Z
2 1
V [ξ] = Cr{(ξ − e) ≥ r}dr = dr = .
0 0 2 8
Example 2.41: Let ξ = (a, b, c) be a triangular fuzzy variable. Then its

variance is
33α3 + 21α2 β + 11αβ 2 − β 3
V [ξ] =
384α
where α = (b − a) ∨ (c − b) and β = (b − a) ∧ (c − b). Especially, if ξ is
symmetric, i.e., b − a = c − b, then its variance is V [ξ] = (c − a)2 /24.
Example 2.42: A fuzzy variable ξ is called normally distributed if it has a

normal membership function
−1
π|x − e|
µ(x) = 2 1 + exp √ , x ∈ <, σ > 0. (2.70)
6σ
The expected value is e and variance is σ 2 . Let ξ1 and ξ2 be independently

and normally distributed fuzzy variables with expected values e1 and e2 ,
variances σ12 and σ22 , respectively. Then for any real numbers a1 and a2 , the
fuzzy variable a1 ξ1 + a2 ξ2 is also normally distributed with expected value
a1 e1 + a2 e2 and variance (|a1 |σ1 + |a2 |σ2 )2 .
µ(x)
...
..........
... . . . . . . . . . . . . . . . . . . ..
1 ...
...
....
.........
.... . ....
... .... . ....
... ..
..... . ........
... . .....
... .... . ....
... .... . .....
... .... . .....
..... . ..
... . . . . . . . . ............ . . . . . . . . . . . . . . . ...........
0.434 ...
... ..
.. ..
.
.... ..
.
. ......
. ...........
... .
...
..... . .
. . .......
... .......
................. .
. .
.
.
. .........
... ............
.............................. .... .
. .
.
.
. ....................
.............................................................................................................................................................................................................................................................................
...
x
...
0 e−σ .... e e+σ
Figure 2.3: Normal Membership Function
Theorem 2.40 If ξ is a fuzzy variable whose variance exists, a and b are

real numbers, then V [aξ + b] = a2 V [ξ].

Theorem 2.41 Let ξ be a fuzzy variable with expected value e. Then V [ξ] =
0 if and only if Cr{ξ = e} = 1.
Proof: If V [ξ] = 0, then E[(ξ − e)2 ] = 0. Note that

Z +∞
E[(ξ − e)2 ] = Cr{(ξ − e)2 ≥ r}dr
0
which implies Cr{(ξ−e)2 ≥ r} = 0 for any r > 0. Hence we have Cr{(ξ−e)2 =

0} = 1, i.e., Cr{ξ = e} = 1. Conversely, if Cr{ξ = e} = 1, then we have
Cr{(ξ − e)2 = 0} = 1 and Cr{(ξ − e)2 ≥ r} = 0 for any r > 0. Thus
Z +∞
V [ξ] = Cr{(ξ − e)2 ≥ r}dr = 0.
0

Let ξ be a fuzzy variable that takes values in [a, b], but whose membership
function is otherwise arbitrary. When its expected value is given, the maxi-
mum variance theorem will provide the maximum variance of ξ, thus playing
an important role in treating games against nature.
Theorem 2.42 (Li and Liu [111]) Let f be a convex function on [a, b], and
ξ a fuzzy variable that takes values in [a, b] and has expected value e. Then
b−e e−a
E[f (ξ)] ≤ f (a) + f (b). (2.71)
b−a b−a
Proof: For each θ ∈ Θ, we have a ≤ ξ(θ) ≤ b and
b − ξ(θ) ξ(θ) − a
ξ(θ) = a+ b.
b−a b−a
b − ξ(θ) ξ(θ) − a
f (ξ(θ)) ≤ f (a) + f (b).
b−a b−a
Theorem 2.43 (Li and Liu [111], Maximum Variance Theorem) Let ξ be a
fuzzy variable that takes values in [a, b] and has expected value e. Then
V [ξ] ≤ (e − a)(b − e) (2.72)
and equality holds if the fuzzy variable ξ has membership function
2(b − e)

 b − a ∧ 1, if x = a


µ(x) = (2.73)
 2(e − a) ∧ 1, if x = b.


b−a

It is also easy to verify that the fuzzy variable determined by (2.73) has
variance (e − a)(b − e). The theorem is proved.
2.10 Moments
Definition 2.21 (Liu [134]) Let ξ be a fuzzy variable, and k a positive num-
ber. Then
Example 2.43: A fuzzy variable ξ is called exponentially distributed if it

has an exponential membership function
−1
πx
µ(x) = 2 1 + exp √ , x ≥ 0, m > 0. (2.74)
6m
√
The expected value is ( 6m ln 2)/π and the second moment is m2 . Let ξ1
and ξ2 be independently and exponentially distributed fuzzy variables with
second moments m21 and m22 , respectively. Then for any positive real numbers
a1 and a2 , the fuzzy variable a1 ξ1 +a2 ξ2 is also exponentially distributed with
second moment (a1 m1 + a2 m2 )2 .
µ(x)
..
.........
...
1 .....
.... ........
.
... ......
... ....
.....
... ....
... .....
... .....
..
... . . . . . . . ......... .....
0.434 ...
...
. ........
. ..
... . ...........
... . .......
. .........
... . ............
...................
... . ..
................................................................................................................................................................................................................... x
..
0 ..... m
Figure 2.4: Exponential Membership Function
Theorem 2.44 Let ξ be a nonnegative fuzzy variable, and k a positive num-

ber. Then the k-th moment
Z +∞
k
E[ξ ] = k rk−1 Cr{ξ ≥ r}dr. (2.75)
0

Z ∞ Z ∞ Z ∞
E[ξ k ] = Cr{ξ k ≥ x}dx = Cr{ξ ≥ r}drk = k rk−1 Cr{ξ ≥ r}dr.
0 0 0

Theorem 2.45 (Li and Liu [111]) Let ξ be a fuzzy variable that takes val-
ues in [a, b] and has expected value e. Then for any positive integer k, the
kth absolute moment and kth absolute central moment satisfy the following
inequalities,
b−e k e−a k
E[|ξ|k ] ≤ |a| + |b| , (2.76)
b−a b−a
b−e e−a
E[|ξ − e|k ] ≤ (e − a)k + (b − e)k . (2.77)
b−a b−a

and f (x) = |x − e|k .
2.11 Critical Values

In order to rank fuzzy variables, we may use two critical values: optimistic
value and pessimistic value.
Definition 2.22 (Liu [130]) Let ξ be a fuzzy variable, and α ∈ (0, 1]. Then

ξsup (α) = sup r Cr {ξ ≥ r} ≥ α (2.78)
is called the α-optimistic value to ξ, and

ξinf (α) = inf r Cr {ξ ≤ r} ≥ α (2.79)
is called the α-pessimistic value to ξ.
This means that the fuzzy variable ξ will reach upwards of the α-optimistic
value ξsup (α) with credibility α, and will be below the α-pessimistic value
ξinf (α) with credibility α. In other words, the α-optimistic value ξsup (α) is
the supremum value that ξ achieves with credibility α, and the α-pessimistic
value ξinf (α) is the infimum value that ξ achieves with credibility α.
Example 2.44: Let ξ be an equipossible fuzzy variable on (a, b). Then its
α-optimistic and α-pessimistic values are
( (
b, if α ≤ 0.5 a, if α ≤ 0.5
ξsup (α) = ξinf (α) =
a, if α > 0.5, b, if α > 0.5.
Example 2.45: Let ξ = (a, b, c) be a triangular fuzzy variable. Then its

(
2αb + (1 − 2α)c, if α ≤ 0.5
ξsup (α) =
(2α − 1)a + (2 − 2α)b, if α > 0.5,
(
(1 − 2α)a + 2αb, if α ≤ 0.5
ξinf (α) =
(2 − 2α)b + (2α − 1)c, if α > 0.5.
Example 2.46: Let ξ = (a, b, c, d) be a trapezoidal fuzzy variable. Then its

(
2αc + (1 − 2α)d, if α ≤ 0.5
ξsup (α) =
(2α − 1)a + (2 − 2α)b, if α > 0.5,
(
(1 − 2α)a + 2αb, if α ≤ 0.5
ξinf (α) =
(2 − 2α)c + (2α − 1)d, if α > 0.5.
Theorem 2.46 Let ξ be a fuzzy variable. If α > 0.5, then we have
Cr{ξ ≤ ξinf (α)} ≥ α, Cr{ξ ≥ ξsup (α)} ≥ α. (2.80)
Proof: It follows from the definition of α-pessimistic value that there exists
a decreasing sequence {xi } such that Cr{ξ ≤ xi } ≥ α and xi ↓ ξinf (α) as
i → ∞. Since {ξ ≤ xi } ↓ {ξ ≤ ξinf (α)} and limi→∞ Cr{ξ ≤ xi } ≥ α > 0.5, it
follows from the credibility semicontinuity law that
Cr{ξ ≤ ξinf (α)} = lim Cr{ξ ≤ xi } ≥ α.

i→∞
Similarly, there exists an increasing sequence {xi } such that Cr{ξ ≥ xi } ≥

δ and xi ↑ ξsup (α) as i → ∞. Since {ξ ≥ xi } ↓ {ξ ≥ ξsup (α)} and
limi→∞ Cr{ξ ≥ xi } ≥ α > 0.5, it follows from the credibility semicontinuity
law that
Cr{ξ ≥ ξsup (α)} = lim Cr{ξ ≥ xi } ≥ α.
i→∞
Example 2.47: When α ≤ 0.5, it is possible that the inequalities
Cr{ξ ≤ ξinf (α)} < α, Cr{ξ ≥ ξsup (α)} < α
hold. Let ξ be an equipossible fuzzy variable on (−1, 1). It is clear that

ξinf (0.5) = −1. However, Cr{ξ ≤ ξinf (0.5)} = 0 < 0.5. In addition,
ξsup (0.5) = 1 and Cr{ξ ≥ ξsup (0.5)} = 0 < 0.5.

(a) ξinf (α) is an increasing and left-continuous function of α;
(b) ξsup (α) is a decreasing and left-continuous function of α.


i→∞
Thus Cr{ξ ≤ z ∗ } ≥ αi for each i. Letting i → ∞, we get Cr{ξ ≤ z ∗ } ≥ α.


then we have
1 ≥ Cr{ξ < ξ(α)} + Cr{ξ > ξ(α)} ≥ α + α > 1.
A contradiction proves ξinf (α) ≥ ξsup (α). Part (b): Assume that ξinf (α) >
ξsup (α). It follows from the definition of ξinf (α) that Cr{ξ ≤ ξ(α)} < α.
Similarly, it follows from the definition of ξsup (α) that Cr{ξ ≥ ξ(α)} < α.
Thus
1 ≤ Cr{ξ ≤ ξ(α)} + Cr{ξ ≥ ξ(α)} < α + α ≤ 1.
A contradiction proves ξinf (α) ≤ ξsup (α). The theorem is proved.

Proof: If c = 0, then the part (a) is obviously valid. When c > 0, we have
(cξ)sup (α) = sup {r | Cr{cξ ≥ r} ≥ α}

= c sup {r/c | Cr {ξ ≥ r/c} ≥ α}
= cξsup (α).
A similar way may prove that (cξ)inf (α) = cξinf (α).

In order to prove the part (b), it suffices to verify that (−ξ)sup (α) =
−ξinf (α) and (−ξ)inf (α) = −ξsup (α). In fact, for any α ∈ (0, 1], we have
(−ξ)sup (α) = sup{r | Cr{−ξ ≥ r} ≥ α}

= − inf{−r | Cr{ξ ≤ −r} ≥ α}
= −ξinf (α).
Theorem 2.50 Suppose that ξ and η are independent fuzzy variables. Then
for any α ∈ (0, 1], we have
(ξ + η)sup (α) = ξsup (α) + ηsup (α), (ξ + η)inf (α) = ξinf (α) + ηinf (α),
(ξη)sup (α) = ξsup (α)ηsup (α), (ξη)inf (α) = ξinf (α)ηinf (α), if ξ ≥ 0, η ≥ 0,
(ξ ∨ η)sup (α) = ξsup (α) ∨ ηsup (α), (ξ ∨ η)inf (α) = ξinf (α) ∨ ηinf (α),
(ξ ∧ η)sup (α) = ξsup (α) ∧ ηsup (α), (ξ ∧ η)inf (α) = ξinf (α) ∧ ηinf (α).
Proof: For any given number ε > 0, since ξ and η are independent fuzzy
variables, we have
Cr{ξ + η ≥ ξsup (α) + ηsup (α) − ε}

≥ Cr {{ξ ≥ ξsup (α) − ε/2} ∩ {η ≥ ηsup (α) − ε/2}}
= Cr{ξ ≥ ξsup (α) − ε/2} ∧ Cr{η ≥ ηsup (α) − ε/2} ≥ α
which implies
(ξ + η)sup (α) ≥ ξsup (α) + ηsup (α) − ε. (2.81)
On the other hand, by the independence, we have
Cr{ξ + η ≥ ξsup (α) + ηsup (α) + ε}

≤ Cr {{ξ ≥ ξsup (α) + ε/2} ∪ {η ≥ ηsup (α) + ε/2}}
= Cr{ξ ≥ ξsup (α) + ε/2} ∨ Cr{η ≥ ηsup (α) + ε/2} < α
which implies
(ξ + η)sup (α) ≤ ξsup (α) + ηsup (α) + ε. (2.82)
ξsup (α) + ηsup (α) + ε ≥ (ξ + η)sup (α) ≥ ξsup (α) + ηsup (α) − ε.
Letting ε → 0, we obtain (ξ + η)sup (α) = ξsup (α) + ηsup (α). The other
equalities may be proved similarly.
Example 2.48: The independence condition cannot be removed in Theo-

rem 2.50. For example, let Θ = {θ1 , θ2 }, Cr{θ1 } = Cr{θ2 } = 1/2, and let
fuzzy variables ξ and η be defined as
( (
0, if θ = θ1 1, if θ = θ1
ξ(θ) = η(θ) =
1, if θ = θ2 , 0, if θ = θ2 .
However, (ξ + η)sup (0.6) = 1 6= 0 = ξsup (0.6) + ηsup (0.6).

2.12 Entropy
Fuzzy entropy is a measure of uncertainty and has been studied by many
researchers such as De Luca and Termini [29], Kaufmann [76], Yager [239],
Kosko [85], Pal and Pal [188], Bhandari and Pal [7], and Pal and Bezdek
[192]. Those definitions of entropy characterize the uncertainty resulting
primarily from the linguistic vagueness rather than resulting from information
deficiency, and vanishes when the fuzzy variable is an equipossible one.
In order to measure the uncertainty of fuzzy variables, Liu [137] suggested
that an entropy of fuzzy variables should meet at least the following three
basic requirements:
(i) minimum: the entropy of a crisp number is minimum, i.e., 0;
(ii) maximum: the entropy of an equipossible fuzzy variable is maximum;
(iii) universality: the entropy is applicable not only to finite and infinite cases
but also to discrete and continuous cases.
In order to meet those requirements, Li and Liu [97] provided a new def-
inition of fuzzy entropy to characterize the uncertainty resulting from infor-
mation deficiency which is caused by the impossibility to predict the specified
value that a fuzzy variable takes.
Entropy of Discrete Fuzzy Variables

Definition 2.23 (Li and Liu [97]) Let ξ be a discrete fuzzy variable taking
values in {x1 , x2 , · · · }. Then its entropy is defined by
∞
X
H[ξ] = S(Cr{ξ = xi }) (2.83)
i=1
where S(t) = −t ln t − (1 − t) ln(1 − t).
Remark 2.6: It is easy to verify that S(t) is a symmetric function about

t = 0.5, strictly increases on the interval [0, 0.5], strictly decreases on the
interval [0.5, 1], and reaches its unique maximum ln 2 at t = 0.5.
Remark 2.7: It is clear that the entropy depends only on the number of
values and their credibilities and does not depend on the actual values that
the fuzzy variable takes.
Example 2.49: Suppose that ξ is a discrete fuzzy variable taking values in

{x1 , x2 , · · · }. If there exists some index k such that the membership function
µ(xk ) = 1, and 0 otherwise, then its entropy H[ξ] = 0.
Example 2.50: Suppose that ξ is a simple fuzzy variable taking values

in {x1 , x2 , · · · , xn }. If its membership function µ(x) ≡ 1, then its entropy
H[ξ] = n ln 2.
S(t)
...
..........
...
..
... . . . . . . . . . . . . . . . .......................
ln 2 ... .....
......
.
.
........
......
... ..... . .....
... ..... .....
... ..
..... .
. .....
... ..
.... . ....
....
... ....
.
.
. ...
...
... ... . ...
... .... .
. ...
... ... . ...
... ... . ...
...
... ..
. .
. ...
... .... . ...
... ... .
.
...
... ... . ...
... ... . ...
... ... . ...
...... . ...
. ...
...... . ...
..... .
....................................................................................................................................................................................
....
t
0 ..
. 0.5 1
Figure 2.5: Function S(t) = −t ln t − (1 − t) ln(1 − t)
Theorem 2.51 Suppose that ξ is a discrete fuzzy variable taking values in

{x1 , x2 , · · · }. Then
H[ξ] ≥ 0 (2.84)
and equality holds if and only if ξ is essentially a crisp number.

Cr{ξ = xi } = 0 or 1 for each i. That is, there exists one and only one index
k such that Cr{ξ = xk } = 1, i.e., ξ is essentially a crisp number.
This theorem states that the entropy of a fuzzy variable reaches its min-
imum 0 when the fuzzy variable degenerates to a crisp number. In this case,
there is no uncertainty.
Theorem 2.52 Suppose that ξ is a simple fuzzy variable taking values in

{x1 , x2 , · · · , xn }. Then
H[ξ] ≤ n ln 2 (2.85)
and equality holds if and only if ξ is an equipossible fuzzy variable.
Proof: Since the function S(t) reaches its maximum ln 2 at t = 0.5, we have
n
X
H[ξ] = S(Cr{ξ = xi }) ≤ n ln 2
i=1
and equality holds if and only if Cr{ξ = xi } = 0.5, i.e., µ(xi ) ≡ 1 for all
i = 1, 2, · · · , n.
This theorem states that the entropy of a fuzzy variable reaches its max-
imum when the fuzzy variable is an equipossible one. In this case, there is
no preference among all the values that the fuzzy variable will take.
Entropy of Continuous Fuzzy Variables

Definition 2.24 (Li and Liu [97]) Let ξ be a continuous fuzzy variable.
Then its entropy is defined by
Z +∞
H[ξ] = S(Cr{ξ = x})dx (2.86)
−∞
For any continuous fuzzy variable ξ with membership function µ, we have

Cr{ξ = x} = µ(x)/2 for each x ∈ <. Thus
Z +∞
µ(x) µ(x) µ(x) µ(x)
H[ξ] = − ln + 1− ln 1 − dx. (2.87)
−∞ 2 2 2 2
Example 2.51: Let ξ be an equipossible fuzzy variable (a, b). Then µ(x) = 1
if a ≤ x ≤ b, and 0 otherwise. Thus its entropy is
Z b
1 1 1 1
H[ξ] = − ln + 1 − ln 1 − dx = (b − a) ln 2.
a 2 2 2 2
Example 2.52: Let ξ be a triangular fuzzy variable (a, b, c). Then its
entropy is H[ξ] = (c − a)/2.
Example 2.53: Let ξ be a trapezoidal fuzzy variable (a, b, c, d). Then its
entropy is H[ξ] = (d − a)/2 + (ln 2 − 0.5)(c − b).
Example 2.54: Let ξ be an exponentially distributed

√ fuzzy variable with
second moment m2 . Then its entropy is H[ξ] = πm/ 6.
Example 2.55: Let ξ be a normally distributed fuzzy√variable with expected

value e and variance σ 2 . Then its entropy is H[ξ] = 6πσ/3.
Theorem 2.53 Let ξ be a continuous fuzzy variable. Then H[ξ] > 0.
Proof: The positivity is clear. In addition, when a continuous fuzzy variable

tends to a crisp number, its entropy tends to the minimum 0. However, a
crisp number is not a continuous fuzzy variable.
Theorem 2.54 Let ξ be a continuous fuzzy variable taking values on the

interval [a, b]. Then
H[ξ] ≤ (b − a) ln 2 (2.88)
and equality holds if and only if ξ is an equipossible fuzzy variable (a, b).
Proof: The theorem follows from the fact that the function S(t) reaches its
maximum ln 2 at t = 0.5.
Theorem 2.55 Let ξ and η be two continuous fuzzy variables with member-
ship functions µ(x) and ν(x), respectively. If µ(x) ≤ ν(x) for any x ∈ <,
then we have H[ξ] ≤ H[η].
Proof: Since µ(x) ≤ ν(x), we have S(µ(x)/2) ≤ S(ν(x)/2) for any x ∈ <.
It follows that H[ξ] ≤ H[η].
Theorem 2.56 Let ξ be a continuous fuzzy variable. Then for any real
numbers a and b, we have H[aξ + b] = |a|H[ξ].
Proof: It follows from the definition of entropy that

Z +∞ Z +∞
H[aξ+b] = S(Cr{aξ+b = x})dx = |a| S(Cr{ξ = y})dy = |a|H[ξ].
−∞ −∞
Maximum Entropy Principle

Given some constraints, for example, expected value and variance, there are
usually multiple compatible membership functions. Which membership func-
tion shall we take? The maximum entropy principle attempts to select the
membership function that maximizes the value of entropy and satisfies the
prescribed constraints.
Theorem 2.25 (Li and Liu [104]) Let ξ be a continuous nonnegative fuzzy
variable with finite second moment m2 . Then
πm
H[ξ] ≤ √ (2.89)
6
and the equality holds if ξ is an exponentially distributed fuzzy variable with
second moment m2 .
Proof: Let µ be the membership function of ξ. Note that µ is a continuous

function. The proof is based on the following two steps.
Step 1: Suppose that µ is a decreasing function on [0, +∞). For this
case, we have Cr{ξ ≥ x} = µ(x)/2 for any x > 0. Thus the second moment
Z +∞ Z +∞ Z +∞
E[ξ 2 ] = Cr{ξ 2 ≥ x}dx = 2xCr{ξ ≥ x}dx = xµ(x)dx.
0 0 0
The maximum entropy membership function µ should maximize the entropy

Z +∞
− ln + 1− ln 1 − dx
0 2 2 2 2
subject to the moment constraint
Z +∞
xµ(x)dx = m2 .
0
The Lagrangian is
Z +∞
L= − ln + 1− ln 1 − dx
0 2 2 2 2
Z +∞
2
−λ xµ(x)dx − m .
0
The maximum entropy membership function meets Euler-Lagrange equation

1 µ(x) 1 µ(x)
ln − ln 1 − + λx = 0
2 2 2 2
−1
and has the form µ(x) = 2 (1 + exp(2λx)) . Substituting it into the moment
constraint, we get
−1
πx
µ∗ (x) = 2 1 + exp √ , x≥0
6m
which is just the exponential membership function
√ with second moment m2 ,
∗
and the maximum entropy is H[ξ ] = πm/ 6.
Step 2: Let ξ be a general fuzzy variable with second moment m2 . Now
we define a fuzzy variable ξb via membership function
µ
b(x) = sup µ(y), x ≥ 0.
y≥x
Then µ
b is a decreasing function on [0, +∞), and
1 1 1
Cr{ξb2 ≥ x} = sup µ
b(y) = sup sup µ(z) = sup µ(z) ≤ Cr{ξ 2 ≥ x}
2 y≥√x 2 y≥√x z≥y 2 z≥√x
for any x > 0. Thus we have
Z +∞ Z +∞
E[ξb2 ] = Cr{ξb2 ≥ x}dx ≤ Cr{ξ 2 ≥ x}dx = E[ξ 2 ] = m2 .
0 0
It follows from µ(x) ≤ µ

b(x) and Step 1 that
q
π E[ξb2 ] πm
H[ξ] ≤ H[ξ]
b ≤ √ ≤ √ .
6 6
The theorem is thus proved.
Theorem 2.26 (Li and Liu [104]) Let ξ be a continuous fuzzy variable with
finite expected value e and variance σ 2 . Then
√
6πσ
H[ξ] ≤ (2.90)
3
and the equality holds if ξ is a normally distributed fuzzy variable with expected
value e and variance σ 2 .
Proof: Let µ be the continuous membership function of ξ. The proof is

based on the following two steps.
Step 1: Let µ(x) be a unimodal and symmetric function about x = e.
For this case, the variance is
Z +∞ Z +∞ √
V [ξ] = Cr{(ξ − e)2 ≥ x}dx = Cr{ξ − e ≥ x}dx
0 0
Z +∞ Z +∞
= 2(x − e)Cr{ξ ≥ x}dx = (x − e)µ(x)dx
e e
and the entropy is

Z +∞
H[ξ] = −2 ln + 1− ln 1 − dx.
e 2 2 2 2
The maximum entropy membership function µ should maximize the entropy

subject to the variance constraint. The Lagrangian is
Z +∞
L = −2 ln + 1− ln 1 − dx
e 2 2 2 2
Z +∞
2
−λ (x − e)µ(x)dx − σ .
e
The maximum entropy membership function meets Euler-Lagrange equation

µ(x) µ(x)
ln − ln 1 − + λ(x − e) = 0
2 2
−1
and has the form µ(x) = 2 (1 + exp (λ(x − e))) . Substituting it into the
variance constraint, we get
−1
∗ π|x − e|
µ (x) = 2 1 + exp √ , x∈<
6σ
which is just the normal membership function with √ expected value e and
variance σ 2 , and the maximum entropy is H[ξ ∗ ] = 6πσ/3.
Step 2: Let ξ be a general fuzzy variable with expected value e and
variance σ 2 . We define a fuzzy variable ξb by the membership function

 sup(µ(y) ∨ µ(2e − y)),
 y≤x if x ≤ e
µ
b(x) =
 sup (µ(y) ∨ µ(2e − y)) , if x > e.

y≥x
It is easy to verify that µ

b(x) is a unimodal and symmetric function about
x = e. Furthermore,
n o 1 1
Cr (ξb − e)2 ≥ r = sup µb(x) = sup sup(µ(y) ∨ µ(2e − y))
2 x≥e+√r 2 x≥e+√r y≥x
1 1
= sup√ (µ(y) ∨ µ(2e − y)) = sup µ(y)
2 y≥e+ r 2 (y−e)2 ≥r

≤ Cr (ξ − e)2 ≥ r
for any r > 0. Thus
Z +∞ Z +∞
2
V [ξ] =
b Cr{(ξ − e) ≥ r}dr ≤
b Cr{(ξ − e)2 ≥ r}dr = σ 2 .
0 0
It follows from µ(x) ≤ µ

b(x) and Step 1 that
√ q √
6π V [ξ]b 6πσ
H[ξ] ≤ H[ξ] ≤
b ≤ .
3 3
The proof is complete.
2.13 Distance
Distance between fuzzy variables has been defined in many ways, for exam-
ple, Hausdorff distance (Puri and Ralescu [204], Klement et. al. [81]), and
Hamming distance (Kacprzyk [73]). However, those definitions have no iden-
tification property. In order to overcome this shortage, Liu [135] proposed a
definition of distance as follows.
Definition 2.27 (Liu [135]) The distance between fuzzy variables ξ and η is
defined as
d(ξ, η) = E[|ξ − η|]. (2.91)
Example 2.56: Let ξ and η be equipossible fuzzy variables (a1 , b1 ) and

(a2 , b2 ), respectively, and (a1 , b1 )∩(a2 , b2 ) = ∅. Then |ξ −η| is an equipossible
fuzzy variable on the interval with endpoints |a1 − b2 | and |b1 − a2 |. Thus
the distance between ξ and η is the expected value of |ξ − η|, i.e.,
1
d(ξ, η) = (|a1 − b2 | + |b1 − a2 |) .
2
Example 2.57: Let ξ = (a1 , b1 , c1 ) and η = (a2 , b2 , c2 ) be triangular fuzzy

variables such that (a1 , c1 ) ∩ (a2 , c2 ) = ∅. Then
1
d(ξ, η) = (|a1 − c2 | + 2|b1 − b2 | + |c1 − a2 |) .
4
Example 2.58: Let ξ = (a1 , b1 , c1 , d1 ) and η = (a2 , b2 , c2 , d2 ) be trapezoidal

fuzzy variables such that (a1 , d1 ) ∩ (a2 , d2 ) = ∅. Then
1
d(ξ, η) = (|a1 − d2 | + |b1 − c2 | + |c1 − b2 | + |d1 − a2 |) .
4
Theorem 2.57 (Li and Liu [106]) Let ξ, η, τ be fuzzy variables, and let d(·, ·)
be the distance. Then we have
(d) (Triangle Inequality) d(ξ, η) ≤ 2d(ξ, τ ) + 2d(η, τ ).
Now we prove the part (d). It follows from the credibility subadditivity
theorem that
Z +∞
d(ξ, η) = Cr {|ξ − η| ≥ r} dr
0
Z +∞
≤ Cr {|ξ − τ | + |τ − η| ≥ r} dr
0
Z +∞
≤ Cr {{|ξ − τ | ≥ r/2} ∪ {|τ − η| ≥ r/2}} dr
0
Z +∞
≤ (Cr{|ξ − τ | ≥ r/2} + Cr{|τ − η| ≥ r/2}) dr
0
Z +∞ Z +∞
= Cr{|ξ − τ | ≥ r/2}dr + Cr{|τ − η| ≥ r/2}dr
0 0
= 2E[|ξ − τ |] + 2E[|τ − η|] = 2d(ξ, τ ) + 2d(τ, η).
Example 2.59: Let Θ = {θ1 , θ2 , θ3 } and Cr{θi } = 1/2 for i = 1, 2, 3. We

define fuzzy variables ξ, η and τ as follows,
( (
1, if θ 6= θ3 −1, if θ 6= θ1
ξ(θ) = η(θ) = τ (θ) ≡ 0.
0, otherwise, 0, otherwise,
It is easy to verify that d(ξ, τ ) = d(τ, η) = 1/2 and d(ξ, η) = 3/2. Thus
3
d(ξ, η) = (d(ξ, τ ) + d(τ, η)).
2
2.14 Inequalities
There are several useful inequalities for random variable, such as Markov
inequality, Chebyshev inequality, Hölder’s inequality, Minkowski inequality,
and Jensen’s inequality. This section introduces the analogous inequalities

for fuzzy variable.
Theorem 2.58 (Liu [134]) Let ξ be a fuzzy variable, and f a nonnegative

function. If f is even and increasing on [0, ∞), then for any given number
t > 0, we have
E[f (ξ)]
Cr{|ξ| ≥ t} ≤ . (2.92)
f (t)
Proof: It is clear that Cr{|ξ| ≥ f −1 (r)} is a monotone decreasing function

Z +∞
E[f (ξ)] = Cr{f (ξ) ≥ r}dr
0
Z +∞
= Cr{|ξ| ≥ f −1 (r)}dr
0
Z f (t)
≥ Cr{|ξ| ≥ f −1 (r)}dr
0
Z f (t)
≥ dr · Cr{|ξ| ≥ f −1 (f (t))}
0
= f (t) · Cr{|ξ| ≥ t}
Theorem 2.59 (Liu [134], Markov Inequality) Let ξ be a fuzzy variable.

Then for any given numbers t > 0 and p > 0, we have
E[|ξ|p ]
Cr{|ξ| ≥ t} ≤ . (2.93)
tp
Theorem 2.60 (Liu [134], Chebyshev Inequality) Let ξ be a fuzzy variable

whose variance V [ξ] exists. Then for any given number t > 0, we have
V [ξ]
Cr {|ξ − E[ξ]| ≥ t} ≤ . (2.94)
t2
Proof: It is a special case of Theorem 2.58 when the fuzzy variable ξ is
replaced with ξ − E[ξ], and f (x) = x2 .
Theorem 2.61 (Liu [134], Hölder’s Inequality) Let p and q be two positive
real numbers with 1/p+1/q = 1, and let ξ and η be independent fuzzy variables
with E[|ξ|p ] < ∞ and E[|η|q ] < ∞. Then we have
p p
E[|ξη|] ≤ p E[|ξ|p ] q E[|η|q ]. (2.95)
√
a and b such that
E[f (|ξ|p , |η|q )] ≤ f (E[|ξ|p ], E[|η|q ]).
Theorem 2.62 (Liu [134], Minkowski Inequality) Let p be a real number

with p ≥ 1, and let ξ and η be independent fuzzy variables with E[|ξ|p ] < ∞
and E[|η|p ] < ∞. Then we have
p
p
p p
E[|ξ + η|p ] ≤ p E[|ξ|p ] + p E[|η|p ]. (2.96)
√
E[f (|ξ|p , |η|p )] ≤ f (E[|ξ|p ], E[|η|p ]).
Theorem 2.63 (Liu [135], Jensen’s Inequality) Let ξ be a fuzzy variable,

and f : < → < a convex function. If E[ξ] and E[f (ξ)] are finite, then
f (E[ξ]) ≤ E[f (ξ)]. (2.97)

f (ξ) − f (E[ξ]) ≥ k · (ξ − E[ξ]).
E[f (ξ)] − f (E[ξ]) ≥ k · (E[ξ] − E[ξ]) = 0

This section discusses some convergence concepts of fuzzy sequence: conver-
gence almost surely (a.s.), convergence in credibility, convergence in mean,
and convergence in distribution.
Table 2.1: Relations among Convergence Concepts
Convergence Almost Surely

Convergence Convergence %
→
in Mean in Credibility &
Convergence in Distribution
Definition 2.28 (Liu [134]) Suppose that ξ, ξ1 , ξ2 , · · · are fuzzy variables de-
fined on the credibility space (Θ, P, Cr). The sequence {ξi } is said to be con-
vergent a.s. to ξ if and only if there exists an event A with Cr{A} = 1 such
that
lim |ξi (θ) − ξ(θ)| = 0 (2.98)
i→∞
for every θ ∈ A. In that case we write ξi → ξ, a.s.
Definition 2.29 (Liu [134]) Suppose that ξ, ξ1 , ξ2 , · · · are fuzzy variables de-
fined on the credibility space (Θ, P, Cr). We say that the sequence {ξi } con-
verges in credibility to ξ if
lim Cr {|ξi − ξ| ≥ ε} = 0 (2.99)
i→∞
for every ε > 0.

Definition 2.30 (Liu [134]) Suppose that ξ, ξ1 , ξ2 , · · · are fuzzy variables
with finite expected values defined on the credibility space (Θ, P, Cr). We
say that the sequence {ξi } converges in mean to ξ if
lim E[|ξi − ξ|] = 0. (2.100)
i→∞
lim E[|ξi − ξ|2 ] = 0. (2.101)

i→∞
Definition 2.31 (Liu [134]) Suppose that Φ, Φ1 , Φ2 , · · · are the credibility

distributions of fuzzy variables ξ, ξ1 , ξ2 , · · · , respectively. We say that {ξi }
converges in distribution to ξ if Φi → Φ at any continuity point of Φ.
Convergence in Mean vs. Convergence in Credibility

Theorem 2.64 (Liu [134]) Suppose that ξ, ξ1 , ξ2 , · · · are fuzzy variables de-
fined on the credibility space (Θ, P, Cr). If the sequence {ξi } converges in
mean to ξ, then {ξi } converges in credibility to ξ.
Proof: It follows from Theorem 2.59 that, for any given number ε > 0,
E[|ξi − ξ|]
Cr {|ξi − ξ| ≥ ε} ≤ →0
ε
as i → ∞. Thus {ξi } converges in credibility to ξ.
Example 2.60: Convergence in credibility does not imply convergence in

mean. For example, take (Θ, P, Cr) to be {θ1 , θ2 , · · · } with Cr{θ1 } = 1/2 and
Cr{θj } = 1/j for j = 2, 3, · · · The fuzzy variables are defined by
(
i, if j = i
ξi (θj ) =
0, otherwise

1
Cr {|ξi − ξ| ≥ ε} = → 0.
i
That is, the sequence {ξi } converges in credibility to ξ. However,
E [|ξi − ξ|] ≡ 1 6→ 0.
Convergence Almost Surely vs. Convergence in Credibility
Example 2.61: Convergence a.s. does not imply convergence in credibility.

For example, take (Θ, P, Cr) to be {θ1 , θ2 , · · · } with Cr{θj } = j/(2j + 1) for
j = 1, 2, · · · The fuzzy variables are defined by
(
i, if j = i
ξi (θj ) =
0, otherwise
for i = 1, 2, · · · and ξ = 0. Then the sequence {ξi } converges a.s. to ξ.

However, for any small number ε > 0, we have
i 1
Cr {|ξi − ξ| ≥ ε} = → .
2i + 1 2
That is, the sequence {ξi } does not converge in credibility to ξ.
Theorem 2.65 (Wang and Liu [233]) Suppose that ξ, ξ1 , ξ2 , · · · are fuzzy
variables defined on the credibility space (Θ, P, Cr). If the sequence {ξi } con-
verges in credibility to ξ, then {ξi } converges a.s. to ξ.
Proof: If {ξi } does not converge a.s. to ξ, then there exists an element
θ∗ ∈ Θ with Cr{θ∗ } > 0 such that ξi (θ∗ ) 6→ ξ(θ∗ ) as i → ∞. In other words,
there exists a small number ε > 0 and a subsequence {ξik (θ∗ )} such that
|ξik (θ∗ ) − ξ(θ∗ )| ≥ ε for any k. Since credibility measure is an increasing set
function, we have
Cr {|ξik − ξ| ≥ ε} ≥ Cr{θ∗ } > 0
for any k. It follows that {ξi } does not converge in credibility to ξ. A
contradiction proves the theorem.
Convergence in Credibility vs. Convergence in Distribution

Theorem 2.66 (Wang and Liu [233]) Suppose that ξ, ξ1 , ξ2 , · · · are fuzzy
variables. If the sequence {ξi } converges in credibility to ξ, then {ξi } con-
verges in distribution to ξ.
Proof: Let x be any given continuity point of the distribution Φ. On the

one hand, for any y > x, we have
{ξi ≤ x} = {ξi ≤ x, ξ ≤ y} ∪ {ξi ≤ x, ξ > y} ⊂ {ξ ≤ y} ∪ {|ξi − ξ| ≥ y − x}.
It follows from the credibility subadditivity theorem that
Φi (x) ≤ Φ(y) + Cr{|ξi − ξ| ≥ y − x}.
Since {ξi } converges in credibility to ξ, we have Cr{|ξi − ξ| ≥ y − x} → 0.

Thus we obtain lim supi→∞ Φi (x) ≤ Φ(y) for any y > x. Letting y → x, we
get
lim sup Φi (x) ≤ Φ(x). (2.102)
i→∞
{ξ ≤ z} = {ξ ≤ z, ξi ≤ x} ∪ {ξ ≤ z, ξi > x} ⊂ {ξi ≤ x} ∪ {|ξi − ξ| ≥ x − z}
which implies that
Φ(z) ≤ Φi (x) + Cr{|ξi − ξ| ≥ x − z}.

Since Cr{|ξi − ξ| ≥ x − z} → 0, we obtain Φ(z) ≤ lim inf i→∞ Φi (x) for any
Φ(x) ≤ lim inf Φi (x). (2.103)
i→∞
It follows from (2.102) and (2.103) that Φi (x) → Φ(x). The theorem is
proved.

in credibility. For example, take (Θ, P, Cr) to be {θ1 , θ2 } with Cr{θ1 } =
Cr{θ2 } = 1/2, and define

−1, if θ = θ1
ξ(θ) =
1, if θ = θ2 .
We also define ξi = −ξ for i = 1, 2, · · · Then ξi and ξ are identically dis-
tributed. Thus {ξi } converges in distribution to ξ. But, for any small number
ε > 0, we have Cr{|ξi − ξ| > ε} = Cr{Θ} = 1. That is, the sequence {ξi }
does not converge in credibility to ξ.
Convergence Almost Surely vs. Convergence in Distribution
Example 2.63: Convergence in distribution does not imply convergence a.s.

For example, take (Θ, P, Cr) to be {θ1 , θ2 } with Cr{θ1 } = Cr{θ2 } = 1/2, and
define
−1, if θ = θ1
ξ(θ) =
1, if θ = θ2 .
We also define ξi = −ξ for i = 1, 2, · · · Then {ξi } converges in distribution
to ξ. However, {ξi } does not converge a.s. to ξ.
Example 2.64: Convergence a.s. does not imply convergence in distribution.

For example, take (Θ, P, Cr) to be {θ1 , θ2 , · · · } with Cr{θj } = j/(2j + 1) for
j = 1, 2, · · · The fuzzy variables are defined by
(
i, if j = i
ξi (θj ) =
0, otherwise
for i = 1, 2, · · · and ξ = 0. Then the sequence {ξi } converges a.s. to ξ.
However, the credibility distributions of ξi are


 0, if x < 0
Φi (x) = (i + 1)/(2i + 1), if 0 ≤ x < i

1, if x ≥ i,

i = 1, 2, · · · , respectively. The credibility distribution of ξ is

(
0, if x < 0
Φ(x) =
1, if x ≥ 0.
It is clear that Φi (x) 6→ Φ(x) at x > 0. That is, the sequence {ξi } does not
converge in distribution to ξ.
2.16 Conditional Credibility

We now consider the credibility of an event A after it has been learned that
some other event B has occurred. This new credibility of A is called the
conditional credibility of A given B.
In order to define a conditional credibility measure Cr{A|B}, at first we
have to enlarge Cr{A ∩ B} because Cr{A ∩ B} < 1 for all events whenever
Cr{B} < 1. It seems that we have no alternative but to divide Cr{A ∩ B} by
Cr{B}. Unfortunately, Cr{A∩B}/Cr{B} is not always a credibility measure.
However, the value Cr{A|B} should not be greater than Cr{A ∩ B}/Cr{B}
(otherwise the normality will be lost), i.e.,
Cr{A ∩ B}
Cr{A|B} ≤ . (2.104)
Cr{B}
On the other hand, in order to preserve the self-duality, we should have
Cr{Ac ∩ B}
Cr{A|B} = 1 − Cr{Ac |B} ≥ 1 − . (2.105)
Cr{B}
Furthermore, since (A ∩ B) ∪ (Ac ∩ B) = B, we have Cr{B} ≤ Cr{A ∩ B} +

Cr{Ac ∩ B} by using the credibility subadditivity theorem. Thus
Cr{Ac ∩ B} Cr{A ∩ B}
0≤1− ≤ ≤ 1. (2.106)
Cr{B} Cr{B}
Hence any numbers between 1 − Cr{Ac ∩ B}/Cr{B} and Cr{A ∩ B}/Cr{B}

are reasonable values that the conditional credibility may take. Based on
the maximum uncertainty principle (see Appendix E), we have the following
conditional credibility measure.
Definition 2.32 (Liu [138]) Let (Θ, P, Cr) be a credibility space, and A, B ∈
P. Then the conditional credibility measure of A given B is defined by
Cr{A ∩ B} Cr{A ∩ B}


 , if < 0.5


 Cr{B} Cr{B}

Cr{A|B} = Cr{Ac ∩ B} Cr{Ac ∩ B} (2.107)
1− , if < 0.5
Cr{B} Cr{B}






0.5, otherwise
provided that Cr{B} > 0.

Section 2.16 - Conditional Credibility 109
It follows immediately from the definition of conditional credibility that

Cr{Ac ∩ B} Cr{A ∩ B}
1− ≤ Cr{A|B} ≤ . (2.108)
Cr{B} Cr{B}
Furthermore, the value of Cr{A|B} takes values as close to 0.5 as possible
in the interval. In other words, it accords with the maximum uncertainty
principle.
Theorem 2.67 Let (Θ, P, Cr) be a credibility space, and B an event with
Cr{B} > 0. Then Cr{·|B} defined by (2.107) is a credibility measure, and
(Θ, P, Cr{·|B}) is a credibility space.
Proof: It is sufficient to prove that Cr{·|B} satisfies the normality, mono-
tonicity, self-duality and maximality axioms. At first, it satisfies the normal-
ity axiom, i.e.,
Cr{Θc ∩ B} Cr{∅}
Cr{Θ|B} = 1 − =1− = 1.
Cr{B} Cr{B}
For any events A1 and A2 with A1 ⊂ A2 , if
Cr{A1 ∩ B} Cr{A2 ∩ B}
≤ < 0.5,
Cr{B} Cr{B}
then
Cr{A1 ∩ B} Cr{A2 ∩ B}
Cr{A1 |B} = ≤ = Cr{A2 |B}.
Cr{B} Cr{B}
If
Cr{A1 ∩ B} Cr{A2 ∩ B}
≤ 0.5 ≤ ,
Cr{B} Cr{B}
then Cr{A1 |B} ≤ 0.5 ≤ Cr{A2 |B}. If
Cr{A1 ∩ B} Cr{A2 ∩ B}
0.5 < ≤ ,
Cr{B} Cr{B}
then we have
Cr{Ac1 ∩ B} Cr{Ac2 ∩ B}

Cr{A1 |B} = 1 − ∨0.5 ≤ 1 − ∨0.5 = Cr{A2 |B}.
Cr{B} Cr{B}
This means that Cr{·|B} satisfies the monotonicity axiom. For any event A,
if
Cr{A ∩ B} Cr{Ac ∩ B}
≥ 0.5, ≥ 0.5,
Cr{B} Cr{B}
then we have Cr{A|B} + Cr{Ac |B} = 0.5 + 0.5 = 1 immediately. Otherwise,
without loss of generality, suppose
Cr{A ∩ B} Cr{Ac ∩ B}
< 0.5 < ,
Cr{B} Cr{B}
then we have

c Cr{A ∩ B} Cr{A ∩ B}
Cr{A|B} + Cr{A |B} = + 1− = 1.
Cr{B} Cr{B}
That is, Cr{·|B} satisfies the self-duality axiom. Finally, for any events {Ai }
with supi Cr{Ai |B} < 0.5, we have supi Cr{Ai ∩ B} < 0.5 and
supi Cr{Ai ∩ B} Cr{∪i Ai ∩ B}
sup Cr{Ai |B} = = = Cr{∪i Ai |B}.
i Cr{B} Cr{B}
Thus Cr{·|B} satisfies the maximality axiom. Hence Cr{·|B} is a credibility
measure. Furthermore, (Θ, P, Cr{·|B}) is a credibility space.
Example 2.65: Let ξ be a fuzzy variable, and X a set of real numbers such
that Cr{ξ ∈ X} > 0. Then for any x ∈ X, the conditional credibility of
ξ = x given ξ ∈ X is
Cr{ξ = x} Cr{ξ = x}

 , if < 0.5
∈ ∈ X}



 Cr{ξ X} Cr{ξ

Cr {ξ = x|ξ ∈ X} = Cr{ξ 6= x, ξ ∈ X} Cr{ξ 6= x, ξ ∈ X}
1− , if < 0.5
Cr{ξ ∈ X} Cr{ξ ∈ X}






0.5, otherwise.
Example 2.66: Let ξ and η be two fuzzy variables, and Y a set of real
numbers such that Cr{η ∈ Y } > 0. Then we have
Cr{ξ = x, η ∈ Y } Cr{ξ = x, η ∈ Y }

 , if < 0.5
∈ } Cr{η ∈ Y }



 Cr{η Y

Cr {ξ = x|η ∈ Y } = Cr{ξ 6= x, η ∈ Y } Cr{ξ 6= x, η ∈ Y }
1− , if < 0.5
Cr{η ∈ Y } Cr{η ∈ Y }






0.5, otherwise.
Definition 2.33 (Liu [138]) The conditional membership function of a fuzzy
variable ξ given B is defined by
µ(x|B) = (2Cr{ξ = x|B}) ∧ 1, x∈< (2.109)
Example 2.67: Let ξ be a fuzzy variable with membership function µ(x),

and X a set of real numbers such that µ(x) > 0 for some x ∈ X. Then the
conditional membership function of ξ given ξ ∈ X is

2µ(x)
∧ 1, if sup µ(x) < 1


sup µ(x)



 x∈X
x∈X
µ(x|X) = (2.110)
 2µ(x)

 ∧ 1, if sup µ(x) = 1
 2 − sup µ(x)

x∈X

x∈X c
for x ∈ X. Please mention that µ(x|X) ≡ 0 if x 6∈ X.
µ(x|X) µ(x|X)
. .
.... ....
....... .......
.. ..
...................................... .......................................
1 ...
... ...
................................................................
... ... ...
... .....
...
...
1 ...
...
.....
... ... ..
... ..... ..
..........................
...
...
... ... ... ... ... ... ... ..... ...
... ... .
.... ... ...
... ... ..... .... ...
.. .. ... .. ... ...
... . .... .... ... ... .
..
. .
. ... . ...
... ..
. .. .... ... ... .. .
. ... . ...
... ... .... .... ... ... .... . . ...
.. ..... .. .. . ...
... ...
... .... ... .
..
. .
. .... ...
... .. .... ... ... ... .. .
. .... ...
... .. .... ... ... . . .
. . ... ... .
.. . .... ....
... ..
..
. .... .... ... .
..
. .
. . ... ....
... ... .. ... .. ... .. . .... ...
... ..... .. ... ...
... .. ... .....
. ... ... ..
... .... .. ...... ... .... .. ... ...
.. . .....
...... ... .. .. ..
.................................................................................................................................................................. x ................................................................................................................................................................... x
. . .
.... ............................................... . .
............................................... .... . . .
.....................
0 ...
.
. X 0 ...
.
.
X
....................
Figure 2.6: Conditional Membership Function µ(x|X)
Example 2.68: Let ξ and η be two fuzzy variables with joint member-
ship function µ(x, y), and Y a set of real numbers. Then the conditional
membership function of ξ given η ∈ Y is
2 sup µ(x, y)


 y∈Y

 ∧ 1, if sup µ(x, y) < 1
sup µ(x, y)

x∈<,y∈Y



 x∈<,y∈Y
µ(x|Y ) = (2.111)


 2 sup µ(x, y)
 y∈Y
∧ 1, if sup µ(x, y) = 1


 2 − sup µ(x, y)

 x∈<,y∈Y
x∈<,y∈Y c
provided that µ(x, y) > 0 for some x ∈ < and y ∈ Y . Especially, the
conditional membership function of ξ given η = y is
2µ(x, y)


 ∧ 1, if sup µ(x, y) < 1


 sup µ(x, y) x∈<
x∈<

µ(x|y) =
 2µ(x, y)
∧ 1, if sup µ(x, y) = 1


 2 − sup µ(x, z)

 x∈<
x∈<,z6=y
provided that µ(x, y) > 0 for some x ∈ <.
Definition 2.34 Let ξ be a fuzzy variable on (Θ, P, Cr). A conditional fuzzy

variable of ξ given B is a function ξ|B from the conditional credibility space
(Θ, P, Cr{·|B}) to the set of real numbers such that
ξ|B (θ) ≡ ξ(θ), ∀θ ∈ Θ. (2.112)

Definition 2.35 (Liu [138]) The conditional credibility distribution Φ: < →

[0, 1] of a fuzzy variable ξ given B is defined by
Φ(x|B) = Cr {ξ ≤ x|B} (2.113)
Example 2.69: Let ξ and η be fuzzy variables. Then the conditional credi-
bility distribution of ξ given η = y is
Cr{ξ ≤ x, η = y} Cr{ξ ≤ x, η = y}


 , if < 0.5


 Cr{η = y} Cr{η = y}

Φ(x|η = y) = Cr{ξ > x, η = y} Cr{ξ > x, η = y}
1− , if < 0.5
Cr{η = y} Cr{η = y}






0.5, otherwise
provided that Cr{η = y} > 0.
Definition 2.36 (Liu [138]) The conditional credibility density function φ

of a fuzzy variable ξ given B is a nonnegative function such that
Z x
Φ(x|B) = φ(y|B)dy, ∀x ∈ <, (2.114)
−∞
Z +∞
φ(y|B)dy = 1 (2.115)
−∞
where Φ(x|B) is the conditional credibility distribution of ξ given B.
Definition 2.37 (Liu [138]) Let ξ be a fuzzy variable. Then the conditional
expected value of ξ given B is defined by
Z +∞ Z 0
E[ξ|B] = Cr{ξ ≥ r|B}dr − Cr{ξ ≤ r|B}dr (2.116)
0 −∞
Following conditional credibility and conditional expected value, we also

have conditional variance, conditional moments, conditional critical values,
conditional entropy as well as conditional convergence.
Hazard Rate
Definition 2.38 Let ξ be a nonnegative fuzzy variable representing lifetime.
Then the hazard rate (or failure rate) is

h(x) = lim Cr{ξ ≤ x + ∆ ξ > x}. (2.117)
∆↓0
The hazard rate tells us the credibility of a failure just after time x when it
is functioning at time x.
Example 2.70: Let ξ be an exponentially distributed fuzzy variable. Then

its hazard rate h(x) ≡ 0.5. In fact, the hazard rate is always 0.5 if the
membership function is positive and decreasing.
Example 2.71: Let ξ be a triangular fuzzy variable (a, b, c) with a ≥ 0.

Then its hazard rate


 0, if x ≤ a

x − a


, if a ≤ x ≤ (a + b)/2



h(x) = b−a
0.5, if (a + b)/2 ≤ x < c







0, if x ≥ c.

Chapter 3
Chance Theory
Fuzziness and randomness are two basic types of uncertainty. In many cases,
fuzziness and randomness simultaneously appear in a system. In order to
describe this phenomena, a fuzzy random variable was introduced by Kwak-
ernaak [87] as a random element taking “fuzzy variable” values. In addition,
a random fuzzy variable was proposed by Liu [130] as a fuzzy element taking
“random variable” values. For example, it might be known that the lifetime
of a modern engine is an exponentially distributed random variable with an
unknown parameter. If the parameter is provided as a fuzzy variable, then
the lifetime is a random fuzzy variable.
More generally, a hybrid variable was introduced by Liu [136] in 2006 as a
tool to describe the quantities with fuzziness and randomness. Fuzzy random
variable and random fuzzy variable are instances of hybrid variable. In order
to measure hybrid events, a concept of chance measure was introduced by Li
and Liu [107] in 2009. Chance theory is a hybrid of probability theory and
credibility theory. Perhaps the reader would like to know what axioms we
should assume for chance theory. In fact, chance theory will be based on the
three axioms of probability and five axioms of credibility.
The emphasis in this chapter is mainly on chance space, hybrid variable,
chance measure, chance distribution, expected value, variance, moments,
critical values, entropy, distance, convergence almost surely, convergence in
chance, convergence in mean, convergence in distribution, and conditional
chance.
3.1 Chance Space

Chance theory begins with the concept of chance space that inherits the
mathematical foundations of both probability theory and credibility theory.
Definition 3.1 (Liu [136]) Suppose that (Θ, P, Cr) is a credibility space and
(Ω, A, Pr) is a probability space. The product (Θ, P, Cr) × (Ω, A, Pr) is called
116 Chapter 3 - Chance Theory
a chance space.
The universal set Θ × Ω is clearly the set of all ordered pairs of the form
(θ, ω), where θ ∈ Θ and ω ∈ Ω. What is the product σ-algebra P × A? What
is the product measure Cr × Pr? Let us discuss these two basic problems.
What is the product σ-algebra P × A?

Generally speaking, it is not true that all subsets of Θ × Ω are measurable.
Let Λ be a subset of Θ × Ω. Write

Λ(θ) = ω ∈ Ω (θ, ω) ∈ Λ . (3.1)
It is clear that Λ(θ) is a subset of Ω. If Λ(θ) ∈ A holds for each θ ∈ Θ, then

Λ may be regarded as a measurable set.
Ω..
..
.........
....
..
...
... ...............................
... .......... ......
.............................................................. ...... .........
. ... ......
..... .
. ........ ... .......
..
.... .... .
... .. ...
... ... .... .. ...
...
... ... ... . ...
..
. ... .... ...
... .... ... ...
Λ(θ) ...
...
...
... Λ ..
..
..
...
... ... ... .. ...
... ... ... . ..
.
... ..
... ... ... .. ...
... ... ..
...
..... .. .....
........ ... ..... .. .......
.... ..
. .
..............................................................................
........... .
.... ........................... ..
... ..
... ..
... ..
... ..
.........................................................................................................................................................................
.
.
.... Θ
..
...
.
θ
Figure 3.1: Θ × Ω, Λ and Λ(θ)
Definition 3.2 (Liu [138]) Let (Θ, P, Cr) × (Ω, A, Pr) be a chance space. A
subset Λ ⊂ Θ × Ω is called an event if Λ(θ) ∈ A for each θ ∈ Θ.
Example 3.1: Empty set ∅ and universal set Θ × Ω are clearly events.
Example 3.2: Let X ∈ P and Y

∈ A. Then X × Y is a subset of Θ × Ω.
Since the set (
Y, if θ ∈ X
(X × Y )(θ) =
∅, if θ ∈ X c
is in the σ-algebra A for each θ ∈ Θ, the rectangle X × Y is an event.
Theorem 3.1 (Liu [138]) Let (Θ, P, Cr) × (Ω, A, Pr) be a chance space. The
class of all events is a σ-algebra over Θ × Ω, and denoted by P × A.
Section 3.1 - Chance Space 117
Proof: At first, it is obvious that Θ×Ω ∈ P × A. For any event Λ, we always

have
Λ(θ) ∈ A, ∀θ ∈ Θ.
Thus for each θ ∈ Θ, the set
Λc (θ) = ω ∈ Ω (θ, ω) ∈ Λc = (Λ(θ))c ∈ A

which implies that Λc ∈ P × A. Finally, let Λ1 , Λ2 , · · · be events. Then for

each θ ∈ Θ, we have
∞ ∞ ∞
! ( )
ω ∈ Ω (θ, ω) ∈ Λi ∈ A.
[ [ [
Λi (θ) = ω ∈ Ω (θ, ω) ∈
Λi =
i=1 i=1 i=1
That is, the countable union ∪i Λi ∈ P × A. HenceP × A is a σ-algebra.

Example 3.3: When Θ × Ω is countable, usually P × A is the power set.
Example 3.4: When Θ×Ω is uncountable, for example Θ×Ω = <2 , usually
P × A is a σ-algebra between the Borel algebra and power set of <2 . Let X
be a nonempty Borel set and let Y be a non-Borel set of real numbers. It
follows from X × Y 6∈ P × A that P × A is smaller than the power set. It is
also clear that Y × X ∈ P × A but Y × X is not a Borel set. Hence P × A is
larger than the Borel algebra.
What is the product measure Cr × Pr?

Product probability is a probability measure, and product credibility is a
credibility measure. What is the product measure Cr × Pr? We will call it
chance measure and define it as follows.
Definition 3.3 (Li and Liu [107]) Let (Θ, P, Cr) × (Ω, A, Pr) be a chance
space. Then a chance measure of an event Λ is defined as

 sup(Cr{θ} ∧ Pr{Λ(θ)}),
θ∈Θ




if sup(Cr{θ} ∧ Pr{Λ(θ)}) < 0.5




θ∈Θ

Ch{Λ} = (3.2)


 1 − sup(Cr{θ} ∧ Pr{Λc (θ)}),


 θ∈Θ

if sup(Cr{θ} ∧ Pr{Λ(θ)}) ≥ 0.5.



θ∈Θ
Example 3.5: Take a credibility space (Θ, P, Cr) to be {θ1 , θ2 } with Cr{θ1 } =
0.6 and Cr{θ2 } = 0.4, and take a probability space (Ω, A, Pr) to be {ω1 , ω2 }
with Pr{ω1 } = 0.7 and Pr{ω2 } = 0.3. Then
Ch{(θ1 , ω1 )} = 0.6, Ch{(θ2 , ω2 )} = 0.3.

Theorem 3.2 Let (Θ, P, Cr) × (Ω, A, Pr) be a chance space and Ch a chance
measure. Then we have
Ch{∅} = 0, (3.3)
Ch{Θ × Ω} = 1, (3.4)
0 ≤ Ch{Λ} ≤ 1 (3.5)
for any event Λ.
Proof: It follows from the definition immediately.
Theorem 3.3 Let (Θ, P, Cr) × (Ω, A, Pr) be a chance space and Ch a chance
measure. Then for any event Λ, we have
sup(Cr{θ} ∧ Pr{Λ(θ)}) ∨ sup(Cr{θ} ∧ Pr{Λc (θ)}) ≥ 0.5, (3.6)

θ∈Θ θ∈Θ
sup(Cr{θ} ∧ Pr{Λ(θ)}) + sup(Cr{θ} ∧ Pr{Λc (θ)}) ≤ 1, (3.7)

θ∈Θ θ∈Θ
sup(Cr{θ} ∧ Pr{Λ(θ)}) ≤ Ch{Λ} ≤ 1 − sup(Cr{θ} ∧ Pr{Λc (θ)}). (3.8)

θ∈Θ θ∈Θ
Proof: It follows from the basic properties of probability and credibility that
sup(Cr{θ} ∧ Pr{Λ(θ)}) ∨ sup(Cr{θ} ∧ Pr{Λc (θ)})

θ∈Θ θ∈Θ
≥ sup(Cr{θ} ∧ (Pr{Λ(θ)} ∨ Pr{Λc (θ)}))

θ∈Θ
≥ sup Cr{θ} ∧ 0.5 = 0.5

θ∈Θ
and
sup(Cr{θ} ∧ Pr{Λ(θ)}) + sup(Cr{θ} ∧ Pr{Λc (θ)})
θ∈Θ θ∈Θ
= sup (Cr{θ1 } ∧ Pr{Λ(θ1 )} + Cr{θ2 } ∧ Pr{Λc (θ2 )})

θ1 ,θ2 ∈Θ
≤ sup (Cr{θ1 } + Cr{θ2 }) ∨ sup(Pr{Λ(θ)} + Pr{Λc (θ)})

θ1 6=θ2 θ∈Θ
≤ 1 ∨ 1 = 1.
The inequalities (3.8) follows immediately from the above inequalities and
the definition of chance measure.
Theorem 3.4 (Li and Liu [107]) The chance measure is increasing. That
is,
Ch{Λ1 } ≤ Ch{Λ2 } (3.9)
for any events Λ1 and Λ2 with Λ1 ⊂ Λ2 .
Proof: Since Λ1 (θ) ⊂ Λ2 (θ) and Λc2 (θ) ⊂ Λc1 (θ) for each θ ∈ Θ, we have
sup(Cr{θ} ∧ Pr{Λ1 (θ)}) ≤ sup(Cr{θ} ∧ Pr{Λ2 (θ)}),

θ∈Θ θ∈Θ
sup(Cr{θ} ∧ Pr{Λc2 (θ)}) ≤ sup(Cr{θ} ∧ Pr{Λc1 (θ)}).

θ∈Θ θ∈Θ
The argument breaks down into three cases.

Case 1: sup(Cr{θ} ∧ Pr{Λ2 (θ)}) < 0.5. For this case, we have
θ∈Θ
sup(Cr{θ} ∧ Pr{Λ1 (θ)}) < 0.5,

θ∈Θ
Ch{Λ2 } = sup(Cr{θ} ∧ Pr{Λ2 (θ)}) ≥ sup(Cr{θ} ∧ Pr{Λ1 (θ)} = Ch{Λ1 }.

θ∈Θ θ∈Θ
Case 2: sup(Cr{θ} ∧ Pr{Λ2 (θ)}) ≥ 0.5 and sup(Cr{θ} ∧ Pr{Λ1 (θ)}) < 0.5.
θ∈Θ θ∈Θ
It follows from Theorem 3.3 that
Ch{Λ2 } ≥ sup(Cr{θ} ∧ Pr{Λ2 (θ)}) ≥ 0.5 > Ch{Λ1 }.

θ∈Θ
Case 3: sup(Cr{θ} ∧ Pr{Λ2 (θ)}) ≥ 0.5 and sup(Cr{θ} ∧ Pr{Λ1 (θ)}) ≥ 0.5.
θ∈Θ θ∈Θ
For this case, we have
Ch{Λ2 } = 1−sup(Cr{θ}∧Pr{Λc2 (θ)}) ≥ 1−sup(Cr{θ}∧Pr{Λc1 (θ)}) = Ch{Λ1 }.

θ∈Θ θ∈Θ
Thus Ch is an increasing measure.
Theorem 3.5 (Li and Liu [107]) The chance measure is self-dual. That is,
Ch{Λ} + Ch{Λc } = 1 (3.10)
for any event Λ.
Proof: For any event Λ, please note that

c
 sup(Cr{θ} ∧ Pr{Λ (θ)}), if sup(Cr{θ} ∧ Pr{Λc (θ)}) < 0.5

c θ∈Θ θ∈Θ
Ch{Λ } =
 1 − sup(Cr{θ} ∧ Pr{Λ(θ)}), if sup(Cr{θ} ∧ Pr{Λc (θ)}) ≥ 0.5.
θ∈Θ θ∈Θ
The argument breaks down into three cases.

Case 1: sup(Cr{θ} ∧ Pr{Λ(θ)}) < 0.5. For this case, we have
θ∈Θ
sup(Cr{θ} ∧ Pr{Λc (θ)}) ≥ 0.5,

θ∈Θ
Ch{Λ} + Ch{Λc } = sup(Cr{θ} ∧ Pr{Λ(θ)}) + 1 − sup(Cr{θ} ∧ Pr{Λ(θ)}) = 1.

θ∈Θ θ∈Θ
Case 2: sup(Cr{θ} ∧ Pr{Λ(θ)}) ≥ 0.5 and sup(Cr{θ} ∧ Pr{Λc (θ)}) < 0.5.
θ∈Θ θ∈Θ
For this case, we have
Ch{Λ}+Ch{Λc } = 1−sup(Cr{θ}∧Pr{Λc (θ)})+sup(Cr{θ}∧Pr{Λc (θ)}) = 1.

θ∈Θ θ∈Θ
Case 3: sup(Cr{θ} ∧ Pr{Λ(θ)}) ≥ 0.5 and sup(Cr{θ} ∧ Pr{Λc (θ)}) ≥ 0.5.

θ∈Θ θ∈Θ
For this case, it follows from Theorem 3.3 that
sup(Cr{θ} ∧ Pr{Λ(θ)}) = sup(Cr{θ} ∧ Pr{Λc (θ)}) = 0.5.

θ∈Θ θ∈Θ
Hence Ch{Λ} + Ch{Λc } = 0.5 + 0.5 = 1. The theorem is proved.
Theorem 3.6 (Li and Liu [107]) For any event X × Y , we have
Ch{X × Y } = Cr{X} ∧ Pr{Y }. (3.11)
Proof: The argument breaks down into three cases.

Case 1: Cr{X} < 0.5. For this case, we have
sup Cr{θ} ∧ Pr{Y } = Cr{X} ∧ Cr{Y } < 0.5,

θ∈X
Ch{X × Y } = sup Cr{θ} ∧ Pr{Y } = Cr{X} ∧ Pr{Y }.

θ∈X
Case 2: Cr{X} ≥ 0.5 and Pr{Y } < 0.5. Then we have
sup Cr{θ} ≥ 0.5,

θ∈X
sup Cr{θ} ∧ Pr{Y } = Pr{Y } < 0.5,

θ∈X
Ch{X × Y } = sup Cr{θ} ∧ Pr{Y } = Pr{Y } = Cr{X} ∧ Pr{Y }.

θ∈X
Case 3: Cr{X} ≥ 0.5 and Pr{Y } ≥ 0.5. Then we have
sup (Cr{θ} ∧ Pr{(X × Y )(θ)}) ≥ sup Cr{θ} ∧ Pr{Y } ≥ 0.5,

θ∈Θ θ∈X
Ch{X × Y } = 1 − sup (Cr{θ} ∧ Pr{(X × Y )c (θ)}) = Cr{X} ∧ Pr{Y }.

θ∈Θ
Example 3.6: It follows from Theorem 3.6 that for any events X × Ω and
Θ × Y , we have
Ch{X × Ω} = Cr{X}, Ch{Θ × Y } = Pr{Y }. (3.12)

Theorem 3.7 (Li and Liu [107], Chance Subadditivity Theorem) The chance
measure is subadditive. That is,
Ch{Λ1 ∪ Λ2 } ≤ Ch{Λ1 } + Ch{Λ2 } (3.13)
for any events Λ1 and Λ2 . In fact, chance measure is not only finitely sub-
additive but also countably subadditive.
Proof: The proof breaks down into three cases.
Case 1: Ch{Λ1 ∪ Λ2 } < 0.5. Then Ch{Λ1 } < 0.5, Ch{Λ2 } < 0.5 and
Ch{Λ1 ∪ Λ2 } = sup(Cr{θ} ∧ Pr{(Λ1 ∪ Λ2 )(θ)})
θ∈Θ
≤ sup(Cr{θ} ∧ (Pr{Λ1 (θ)} + Pr{Λ2 (θ)}))
θ∈Θ
≤ sup(Cr{θ} ∧ Pr{Λ1 (θ)} + Cr{θ} ∧ Pr{Λ2 (θ)})
θ∈Θ
≤ sup(Cr{θ} ∧ Pr{Λ1 (θ)}) + sup(Cr{θ} ∧ Pr{Λ2 (θ)})
θ∈Θ θ∈Θ
= Ch{Λ1 } + Ch{Λ2 }.
Case 2: Ch{Λ1 ∪ Λ2 } ≥ 0.5 and Ch{Λ1 } ∨ Ch{Λ2 } < 0.5. We first have
sup(Cr{θ} ∧ Pr{(Λ1 ∪ Λ2 )(θ)}) ≥ 0.5.
θ∈Θ
For any sufficiently small number ε > 0, there exists a point θ such that
Cr{θ} ∧ Pr{(Λ1 ∪ Λ2 )(θ)} > 0.5 − ε > Ch{Λ1 } ∨ Ch{Λ2 },
Cr{θ} > 0.5 − ε > Pr{Λ1 (θ)},
Cr{θ} > 0.5 − ε > Pr{Λ2 (θ)}.
Thus we have
Cr{θ} ∧ Pr{(Λ1 ∪ Λ2 )c (θ)} + Cr{θ} ∧ Pr{Λ1 (θ)} + Cr{θ} ∧ Pr{Λ2 (θ)}
= Cr{θ} ∧ Pr{(Λ1 ∪ Λ2 )c (θ)} + Pr{Λ1 (θ)} + Pr{Λ2 (θ)}
≥ Cr{θ} ∧ Pr{(Λ1 ∪ Λ2 )c (θ)} + Pr{(Λ1 ∪ Λ2 )(θ)} ≥ 1 − 2ε
because if Cr{θ} ≥ Pr{(Λ1 ∪ Λ2 )c (θ)}, then
Cr{θ} ∧ Pr{(Λ1 ∪ Λ2 )c (θ)} + Pr{(Λ1 ∪ Λ2 )(θ)}
= Pr{(Λ1 ∪ Λ2 )c (θ)} + Pr{(Λ1 ∪ Λ2 )(θ)}
= 1 ≥ 1 − 2ε
and if Cr{θ} < Pr{(Λ1 ∪ Λ2 )c (θ)}, then
Cr{θ} ∧ Pr{(Λ1 ∪ Λ2 )c (θ)} + Pr{(Λ1 ∪ Λ2 )(θ)}
= Cr{θ} + Pr{(Λ1 ∪ Λ2 )(θ)}
≥ (0.5 − ε) + (0.5 − ε) = 1 − 2ε.
Taking supremum on both sides and letting ε → 0, we obtain
Ch{Λ1 ∪ Λ2 } = 1 − sup(Cr{θ} ∧ Pr{(Λ1 ∪ Λ2 )c (θ)})

θ∈Θ
≤ sup(Cr{θ} ∧ Pr{Λ1 (θ)}) + sup(Cr{θ} ∧ Pr{Λ2 (θ)})
θ∈Θ θ∈Θ
= Ch{Λ1 } + Ch{Λ2 }.
Case 3: Ch{Λ1 ∪ Λ2 } ≥ 0.5 and Ch{Λ1 } ∨ Ch{Λ2 } ≥ 0.5. Without loss

of generality, suppose Ch{Λ1 } ≥ 0.5. For each θ, we first have
Cr{θ} ∧ Pr{Λc1 (θ)} = Cr{θ} ∧ Pr{(Λc1 (θ) ∩ Λc2 (θ)) ∪ (Λc1 (θ) ∩ Λ2 (θ))}
≤ Cr{θ} ∧ (Pr{(Λ1 ∪ Λ2 )c (θ)} + Pr{Λ2 (θ)})
≤ Cr{θ} ∧ Pr{(Λ1 ∪ Λ2 )c (θ)} + Cr{θ} ∧ Pr{Λ2 (θ)},
i.e., Cr{θ} ∧ Pr{(Λ1 ∪ Λ2 )c (θ)} ≥ Cr{θ} ∧ Pr{Λc1 (θ)} − Cr{θ} ∧ Pr{Λ2 (θ)}. It
follows from Theorem 3.3 that
Ch{Λ1 ∪ Λ2 } = 1 − sup(Cr{θ} ∧ Pr{(Λ1 ∪ Λ2 )c (θ)})

θ∈Θ
≤ 1 − sup(Cr{θ} ∧ Pr{Λc1 (θ)}) + sup(Cr{θ} ∧ Pr{Λ2 (θ)})
θ∈Θ θ∈Θ
≤ Ch{Λ1 } + Ch{Λ2 }.
Remark 3.1: For any events Λ1 and Λ2 , it follows from the chance subaddi-
tivity theorem that the chance measure is null-additive, i.e., Ch{Λ1 ∪ Λ2 } =
Ch{Λ1 } + Ch{Λ2 } if either Ch{Λ1 } = 0 or Ch{Λ2 } = 0.
Theorem 3.8 Let {Λi } be a decreasing sequence of events with Ch{Λi } → 0

as i → ∞. Then for any event Λ, we have
lim Ch{Λ ∪ Λi } = lim Ch{Λ\Λi } = Ch{Λ}. (3.14)

i→∞ i→∞
Proof: Since chance measure is increasing and subadditive, we immediately

have
Ch{Λ} ≤ Ch{Λ ∪ Λi } ≤ Ch{Λ} + Ch{Λi }
for each i. Thus we get Ch{Λ ∪ Λi } → Ch{Λ} by using Ch{Λi } → 0. Since

(Λ\Λi ) ⊂ Λ ⊂ ((Λ\Λi ) ∪ Λi ), we have
Ch{Λ\Λi } ≤ Ch{Λ} ≤ Ch{Λ\Λi } + Ch{Λi }.
Hence Ch{Λ\Λi } → Ch{Λ} by using Ch{Λi } → 0.

Theorem 3.9 (Li and Liu [107], Chance Semicontinuity Law) For events
Λ1 , Λ2 , · · · , we have
n o
lim Ch{Λi } = Ch lim Λi (3.15)
i→∞ i→∞
if one of the following conditions is satisfied:

(a) Ch{Λ} ≤ 0.5 and Λi ↑ Λ; (b) lim Ch{Λi } < 0.5 and Λi ↑ Λ;
i→∞
(c) Ch{Λ} ≥ 0.5 and Λi ↓ Λ; (d) lim Ch{Λi } > 0.5 and Λi ↓ Λ.
i→∞
Proof: (a) Assume Ch{Λ} ≤ 0.5 and Λi ↑ Λ. We first have
Ch{Λ} = sup(Cr{θ} ∧ Pr{Λ(θ)}), Ch{Λi } = sup(Cr{θ} ∧ Pr{Λi (θ)})

θ∈Θ θ∈Θ
for i = 1, 2, · · · For each θ ∈ Θ, since Λi (θ) ↑ Λ(θ), it follows from the

probability continuity theorem that
lim Cr{θ} ∧ Pr{Λi (θ)} = Cr{θ} ∧ Pr{Λ(θ)}.

i→∞
Taking supremum on both sides, we obtain
lim sup(Cr{θ} ∧ Pr{Λi (θ)}) = sup(Cr{θ} ∧ Pr{Λ(θ}).

i→∞ θ∈Θ θ∈Θ
The part (a) is verified.

(b) Assume limi→∞ Ch{Λi } < 0.5 and Λi ↑ Λ. For each θ ∈ Θ, since
Cr{θ} ∧ Pr{Λ(θ)} = lim Cr{θ} ∧ Pr{Λi (θ)},

i→∞
we have
sup(Cr{θ} ∧ Pr{Λ(θ)}) ≤ lim sup(Cr{θ} ∧ Pr{Λi (θ)}) < 0.5.

θ∈Θ i→∞ θ∈Θ
It follows that Ch{Λ} < 0.5 and the part (b) holds by using (a).
(c) Assume Ch{Λ} ≥ 0.5 and Λi ↓ Λ. We have Ch{Λc } ≤ 0.5 and Λci ↑ Λc .
It follows from (a) that
lim Ch{Λi } = 1 − lim Ch{Λci } = 1 − Ch{Λc } = Ch{Λ}.

i→∞ i→∞
(d) Assume limi→∞ Ch{Λi } > 0.5 and Λi ↓ Λ. We have lim Ch{Λci } <
i→∞
0.5 and Λci ↑ Λc . It follows from (b) that
lim Ch{Λi } = 1 − lim Ch{Λci } = 1 − Ch{Λc } = Ch{Λ}.

i→∞ i→∞

Theorem 3.10 (Chance Asymptotic Theorem) For any events Λ1 , Λ2 , · · · ,

we have
lim Ch{Λi } ≥ 0.5, if Λi ↑ Θ × Ω, (3.16)
i→∞
lim Ch{Λi } ≤ 0.5, if Λi ↓ ∅. (3.17)
i→∞
Proof: Assume Λi ↑ Θ × Ω. If limi→∞ Ch{Λi } < 0.5, it follows from the

chance semicontinuity law that
Ch{Θ × Ω} = lim Ch{Λi } < 0.5
i→∞
which is in contradiction with Cr{Θ × Ω} = 1. The first inequality is proved.

The second one may be verified similarly.
3.2 Hybrid Variables
Recall that a random variable is a measurable function from a probability

space to the set of real numbers, and a fuzzy variable is a function from a
credibility space to the set of real numbers. In order to describe a quan-
tity with both fuzziness and randomness, we introduce a concept of hybrid
variable as follows.
............................................................
.................... ............
............ .........
......... ........
....... ......
.
......... .....
.... ....
. ...
.... Set of Real Numbers ...
... .
... ..
...
.....
......
....... .........
•.
... • • ..
..........
. ...
......... ......
... ......
....
.......... ... . ...
..
... ............................ ..........
. ........... ....
... ........................................................................ ...
... ..
. ...
...
... .... ...
........................................................... ... .
. ... .........................................................
... ... .. ... ..... ... ... ....
... Fuzzy ...
... ..
..
.
.
...
.
.
... .
... ........
.....
...
...
...
.Random ...
... ... ... . ......
. ...... ... ..... ...
... . . ...
Variable .
......................................................................... .
...
.
........ .
......
.
......
.... ...... Variable
..................................................................
.
.
.
..
.
... .
..
.
.
.....
Hybrid .......
... .....
....
....
... ..... ..... ...
...
... .....
..... Variable
.....................................................................................................................
.....
.... ...
...
... ...
... ..
. ...
...
... ..
. ...
..
. ..
. ...
..
. ... . ...
... ............
. ...
... ....
.... ......
. ...
..... ...... ...
.. ......................
. ... .
..
....
............. ..............
. ..
...... . ......
. .. .
. . ...... ...............................................
........... ........ ....
. ...... . ...... .............
.... .......... ..... ..
.. ..
....... ...... .. . . ... .......
..
...... ... ... .. . ............ ... .....
....
.. .. ......... . .. ... .....
... ..
. .. ...... ..... ... ........... ... ...
• ....
.
• .. ..... ...
... . .
.
.
.
• • . ...
...
.... .. .... ..
... ... ...
Credibility Space
...
...
... . .
.
.. ...
... Probability Space ..
.
..
.
... ...
.... .... ..... ....
.....
...... ..... ......
......
....... ...... ....... ......
......... ....... ......... ........
..................................................... ...............
........................ .......
....
Figure 3.2: Graphical Representation of Hybrid Variable
Definition 3.4 (Liu [136]) A hybrid variable is a measurable function from

a chance space (Θ, P, Cr) × (Ω, A, Pr) to the set of real numbers, i.e., for any
Borel set B of real numbers, the set

{ξ ∈ B} = {(θ, ω) ∈ Θ × Ω ξ(θ, ω) ∈ B} (3.18)
Section 3.2 - Hybrid Variables 125
is an event.
Remark 3.2: A hybrid variable degenerates to a fuzzy variable if the value

of ξ(θ, ω) does not vary with ω. For example,
ξ(θ, ω) = θ, ξ(θ, ω) = θ2 + 1, ξ(θ, ω) = sin θ.
Remark 3.3: A hybrid variable degenerates to a random variable if the

value of ξ(θ, ω) does not vary with θ. For example,
ξ(θ, ω) = ω, ξ(θ, ω) = ω 2 + 1, ξ(θ, ω) = sin ω.
Remark 3.4: For each fixed θ ∈ Θ, it is clear that the hybrid variable ξ(θ, ω)
is a measurable function from the probability space (Ω, A, Pr) to the set of
real numbers. Thus it is a random variable and we will denote it by ξ(θ, ·).
Then a hybrid variable ξ(θ, ω) may also be regarded as a function from a
credibility space (Θ, P, Cr) to the set {ξ(θ, ·)|θ ∈ Θ} of random variables.
Thus ξ is a random fuzzy variable defined by Liu [130].
Remark 3.5: For each fixed ω ∈ Ω, it is clear that the hybrid variable
ξ(θ, ω) is a function from the credibility space (Θ, P, Cr) to the set of real
numbers. Thus it is a fuzzy variable and we will denote it by ξ(·, ω). Then a
hybrid variable ξ(θ, ω) may be regarded as a function from a probability space
(Ω, A, Pr) to the set {ξ(·, ω)|ω ∈ Ω} of fuzzy variables. If Cr{ξ(·, ω) ∈ B} is
a measurable function of ω for any Borel set B of real numbers, then ξ is a
fuzzy random variable in the sense of Liu and Liu [150].
Model I
If ã is a fuzzy variable and η is a random variable, then the sum ξ = ã + η is

a hybrid variable. The product ξ = ã · η is also a hybrid variable. Generally
speaking, if f : <2 → < is a measurable function, then
ξ = f (ã, η) (3.19)
is a hybrid variable. Suppose that ã has a membership function µ, and η has

a probability density function φ. Then for any Borel set B of real numbers,
we have
 Z !
µ(x)
sup ∧ φ(y)dy ,






 x 2 f (x,y)∈B

 !
 Z
µ(x)


if sup ∧ φ(y)dy < 0.5




 x 2 f (x,y)∈B
Ch{f (ã, η) ∈ B} = Z !

 µ(x)
1 − sup ∧ φ(y)dy ,


2

x


 f (x,y)∈B c



 Z !

 µ(x)

 if sup ∧ φ(y)dy ≥ 0.5.
2

x f (x,y)∈B
More generally, let ã1 , ã2 , · · · , ãm be fuzzy variables, and let η1 , η2 , · · · , ηn be
random variables. If f : <m+n → < is a measurable function, then
ξ = f (ã1 , ã2 , · · · , ãm ; η1 , η2 , · · · , ηn ) (3.20)
is a hybrid variable. The chance Ch{f (ã1 , ã2 , · · · , ãm ; η1 , η2 , · · · , ηn ) ∈ B}
may be calculated in a similar way provided that µ is the joint membership
function and φ is the joint probability density function.
Model II
Let ã1 , ã2 , · · · , ãm be fuzzy variables, and let p1 , p2 , · · · , pm be nonnegative
numbers with p1 + p2 + · · · + pm = 1. Then

 ã1 with probability p1


ξ= ã2 with probability p2 (3.21)

 ···

ãm with probability pm
is clearly a hybrid variable. If ã1 , ã2 , · · · , ãm have membership functions
µ1 , µ2 , · · · , µm , respectively, then for any set B of real numbers, we have
 X m
!
 µi (xi )
sup min ∧ {pi | xi ∈ B} ,


2


 x1 ,x2 ··· ,xm 1≤i≤m

 i=1


 X m
!
µ (x )


 i i

 if sup min ∧ {pi | xi ∈ B} < 0.5
2
 1≤i≤m
 x1 ,x2 ··· ,xm
i=1

Ch{ξ ∈ B} = !
X m
µi (xi )


c
1− sup min ∧ {pi | xi ∈ B } ,






 x1 ,x2 ··· ,xm 1≤i≤m 2 i=1


 !
 X m

 µ i (xi )
 if x ,xsup min ∧ {pi | xi ∈ B} ≥ 0.5.


2

1 2 ··· ,xm 1≤i≤m
i=1
Section 3.2 - Hybrid Variables 127
Model III
Let η1 , η2 , · · · , ηm be random variables, and let u1 , u2 , · · · , um be nonnegative

numbers with u1 ∨ u2 ∨ · · · ∨ um = 1. Then

 η1 with membership degree u1


ξ= η2 with membership degree u2 (3.22)

 ···

ηm with membership degree um
is clearly a hybrid variable. If η1 , η2 , · · · , ηm have probability density func-

tions φ1 , φ2 , · · · , φm , respectively, then for any Borel set B of real numbers,
we have
 Z
ui
max ∧ φi (x)dx ,


2


 1≤i≤m B



 Z

 ui

 if max ∧ φi (x)dx < 0.5
2
 1≤i≤m B

Ch{ξ ∈ B} = Z
 ui
 1 − max ∧

 φi (x)dx ,



 1≤i≤m 2 Bc

 Z
ui


if max ∧ φi (x)dx ≥ 0.5.



1≤i≤m 2 B
Model IV
In many statistics problems, the probability density function is completely

known except for the values of one or more parameters. For example, it
might be known that the lifetime ξ of a modern engine is an exponentially
distributed random variable with an unknown expected value β. Usually,
there is some relevant information in practice. It is thus possible to specify
an interval in which the value of β is likely to lie, or to give an approximate
estimate of the value of β. It is typically not possible to determine the value
of β exactly. If the value of β is provided as a fuzzy variable, then ξ is a
hybrid variable. More generally, suppose that ξ has a probability density
function
φ(x; ã1 , ã2 , · · · , ãm ), x∈< (3.23)
in which the parameters ã1 , ã2 , · · · , ãm are fuzzy variables rather than crisp
numbers. Then ξ is a hybrid variable provided that φ(x; y1 , y2 , · · · , ym ) is
a probability density function for any (y1 , y2 , · · · , ym ) that (ã1 , ã2 , · · · , ãm )
may take. If ã1 , ã2 , · · · , ãm have membership functions µ1 , µ2 , · · · , µm , re-
spectively, then for any Borel set B of real numbers, the chance Ch{ξ ∈ B}
is
Z
µi (yi )


 sup min ∧ φ(x; y1 , y2 , · · · , ym )dx ,



 y1 ,y2 ··· ,ym 1≤i≤m 2 B


 Z

 µi (yi )
if sup min ∧ φ(x; y1 , y2 , · · · , ym )dx < 0.5


2


 y1 ,y2 ,··· ,ym 1≤i≤m B
Z

 µi (yi )

 1 − sup min ∧ φ(x; y1 , y2 , · · · , ym )dx ,
2


 y1 ,y2 ··· ,ym 1≤i≤m Bc



 Z
µi (yi )


∧ φ(x; y1 , y2 , · · · , ym )dx ≥ 0.5.


 if sup min
y1 ,y2 ,··· ,ym 1≤i≤m 2 B
When are two hybrid variables equal to each other?

Definition 3.5 Let ξ1 and ξ2 be hybrid variables defined on the chance space
(Θ, P, Cr) × (Ω, A, Pr). We say ξ1 = ξ2 if ξ1 (θ, ω) = ξ2 (θ, ω) for almost all
(θ, ω) ∈ Θ × Ω.
Hybrid Vectors
Definition 3.6 An n-dimensional hybrid vector is a measurable function
from a chance space (Θ, P, Cr) × (Ω, A, Pr) to the set of n-dimensional real
vectors, i.e., for any Borel set B of <n , the set

{ξ ∈ B} = (θ, ω) ∈ Θ × Ω ξ(θ, ω) ∈ B (3.24)
is an event.
Theorem 3.11 The vector (ξ1 , ξ2 , · · · , ξn ) is a hybrid vector if and only if

ξ1 , ξ2 , · · · , ξn are hybrid variables.
Proof: Write ξ = (ξ1 , ξ2 , · · · , ξn ). Suppose that ξ is a hybrid vector on

the chance space (Θ, P, Cr) × (Ω, A, Pr). For any Borel set B of <, the set
B × <n−1 is a Borel set of <n . Thus the set
{ξ1 ∈ B} = {ξ1 ∈ B, ξ2 ∈ <, · · · , ξn ∈ <} = {ξ ∈ B × <n−1 }
is an event. Hence ξ1 is a hybrid variable. A similar process may prove that

ξ2 , ξ3 , · · · , ξn are hybrid variables.
Conversely, suppose that all ξ1 , ξ2 , · · · , ξn are hybrid variables on the
chance space (Θ, P, Cr) × (Ω, A, Pr). We define
B= B ⊂ <n {ξ ∈ B} is an event .

Section 3.3 - Chance Distribution 129
The vector ξ = (ξ1 , ξ2 , · · · , ξn ) is proved to be a hybrid vector if we can

( n
) n
Y \
ξ∈ (ai , bi ) = {ξi ∈ (ai , bi )}
i=1 i=1
is an event. Next, the class B is a σ-algebra of <n because (i) we have <n ∈ B
since {ξ ∈ <n } = Θ × Ω; (ii) if B ∈ B, then {ξ ∈ B} is an event, and
{ξ ∈ B c } = {ξ ∈ B}c
is an event. This means that B c ∈ B; (iii) if Bi ∈ B for i = 1, 2, · · · , then
{ξ ∈ Bi } are events and
∞ ∞
( )
[ [
ξ∈ Bi = {ξ ∈ Bi }
i=1 i=1
is an event. This means that ∪i Bi ∈ B. Since the smallest σ-algebra con-

taining all open intervals of <n is just the Borel algebra of <n , the class B
contains all Borel sets of <n . The theorem is proved.
Hybrid Arithmetic
Definition 3.7 Let f : <n → < be a measurable function, and ξ1 , ξ2 , · · · , ξn
hybrid variables on the chance space (Θ, P, Cr) × (Ω, A, Pr). Then ξ =
f (ξ1 , ξ2 , · · · , ξn ) is a hybrid variable defined as
ξ(θ, ω) = f (ξ1 (θ, ω), ξ2 (θ, ω), · · · , ξn (θ, ω)), ∀(θ, ω) ∈ Θ × Ω. (3.25)
Example 3.7: Let ξ1 and ξ2 be two hybrid variables. Then the sum ξ =
ξ1 + ξ2 is a hybrid variable defined by
ξ(θ, ω) = ξ1 (θ, ω) + ξ2 (θ, ω), ∀(θ, ω) ∈ Θ × Ω.
The product ξ = ξ1 ξ2 is also a hybrid variable defined by
ξ(θ, ω) = ξ1 (θ, ω) · ξ2 (θ, ω), ∀(θ, ω) ∈ Θ × Ω.
Theorem 3.12 Let ξ be an n-dimensional hybrid vector, and f : <n → < a
measurable function. Then f (ξ) is a hybrid variable.
Proof: Assume that ξ is a hybrid vector on the chance space (Θ, P, Cr) ×
(Ω, A, Pr). For any Borel set B of <, since f is a measurable function, the
f −1 (B) is a Borel set of <n . Thus the set
{f (ξ) ∈ B} = ξ ∈ f −1 (B)

is an event for any Borel set B. Hence f (ξ) is a hybrid variable.

3.3 Chance Distribution

Chance distribution has been defined in several ways. Here we accept the
following definition of chance distribution of hybrid variables.
Definition 3.8 (Li and Liu [107]) The chance distribution Φ: < → [0, 1] of
a hybrid variable ξ is defined by
Φ(x) = Ch {ξ ≤ x} . (3.26)
Example 3.8: Let η be a random variable on a probability space (Ω, A, Pr).

It is clear that η may be regarded as a hybrid variable on the chance space
(Θ, P, Cr) × (Ω, A, Pr) as follows,
ξ(θ, ω) = η(ω), ∀(θ, ω) ∈ Θ × Ω.
Thus its chance distribution is
Φ(x) = Ch{ξ ≤ x} = Ch{Θ × {η ≤ x}} = Cr{Θ} ∧ Pr{η ≤ x} = Pr{η ≤ x}
which is just the probability distribution of the random variable η.
Example 3.9: Let ã be a fuzzy variable on a credibility space (Θ, P, Cr).

It is clear that ã may be regarded as a hybrid variable on the chance space
(Θ, P, Cr) × (Ω, A, Pr) as follows,
ξ(θ, ω) = ã(θ), ∀(θ, ω) ∈ Θ × Ω.
Thus its chance distribution is
Φ(x) = Ch{ξ ≤ x} = Ch{{ã ≤ x} × Ω} = Cr{ã ≤ x} ∧ Pr{Ω} = Cr{ã ≤ x}
which is just the credibility distribution of the fuzzy variable ã.
Theorem 3.13 (Sufficient and Necessary Condition for Chance Distribu-

tion) A function Φ : < → [0, 1] is a chance distribution if and only if it is an
increasing function with
lim Φ(x) ≤ 0.5 ≤ lim Φ(x), (3.27)

x→−∞ x→+∞
lim Φ(y) = Φ(x) if lim Φ(y) > 0.5 or Φ(x) ≥ 0.5. (3.28)
y↓x y↓x
Proof: It is obvious that a chance distribution Φ is an increasing function.

The inequalities (3.27) follow from the chance asymptotic theorem immedi-
ately. Assume that x is a point at which limy↓x Φ(y) > 0.5. That is,
lim Ch{ξ ≤ y} > 0.5.

y↓x
Section 3.3 - Chance Distribution 131
Since {ξ ≤ y} ↓ {ξ ≤ x} as y ↓ x, it follows from the chance semicontinuity

law that
Φ(y) = Ch{ξ ≤ y} ↓ Ch{ξ ≤ x} = Φ(x)
as y ↓ x. When x is a point at which Φ(x) ≥ 0.5, if limy↓x Φ(y) 6= Φ(x), then
we have
lim Φ(y) > Φ(x) ≥ 0.5.
y↓x
For this case, we have proved that limy↓x Φ(y) = Φ(x). Thus (3.28) is proved.
Conversely, suppose Φ : < → [0, 1] is an increasing function satisfying
(3.27) and (3.28). Theorem 2.21 states that there is a fuzzy variable whose
credibility distribution is just Φ(x). Since a fuzzy variable is a special hybrid
variable, the theorem is proved.
Definition 3.9 The chance density function φ: < → [0, +∞) of a hybrid
variable ξ is a function such that
Z x
Φ(x) = φ(y)dy, ∀x ∈ <, (3.29)
−∞
Z +∞
φ(y)dy = 1 (3.30)
−∞
where Φ is the chance distribution of ξ.
Theorem 3.14 Let ξ be a hybrid variable whose chance density function φ

exists. Then we have
Z x Z +∞
Ch{ξ ≤ x} = φ(y)dy, Ch{ξ ≥ x} = φ(y)dy. (3.31)
−∞ x
by the self-duality of chance measure, we have
Z +∞ Z x Z +∞
Ch{ξ ≥ x} = 1 − Ch{ξ < x} = φ(y)dy − φ(y)dy = φ(y)dy.
−∞ −∞ x
Joint Chance Distribution

Definition 3.10 Let (ξ1 , ξ2 , · · · , ξn ) be a hybrid vector. Then the joint chance
distribution Φ : <n → [0, 1] is defined by
Φ(x1 , x2 , · · · , xn ) = Ch {ξ1 ≤ x1 , ξ2 ≤ x2 , · · · , ξn ≤ xn } .
Definition 3.11 The joint chance density function φ : <n → [0, +∞) of a
hybrid vector (ξ1 , ξ2 , · · · , ξn ) is a function such that
Z x1 Z x2 Z xn
−∞ −∞ −∞
n
holds for all (x1 , x2 , · · · , xn ) ∈ < , and
Z +∞ Z +∞ Z +∞
··· φ(y1 , y2 , · · · , yn )dy1 dy2 · · · dyn = 1
−∞ −∞ −∞
where Φ is the joint chance distribution of the hybrid vector (ξ1 , ξ2 , · · · , ξn ).
3.4 Expected Value

Expected value has been defined in several ways. This book uses the following
definition of expected value operator of hybrid variables.
Definition 3.12 (Li and Liu [107]) Let ξ be a hybrid variable. Then the
expected value of ξ is defined by
Z +∞ Z 0
E[ξ] = Ch{ξ ≥ r}dr − Ch{ξ ≤ r}dr (3.32)
0 −∞
Example 3.10: If a hybrid variable ξ degenerates to a random variable η,

then
Ch{ξ ≤ x} = Pr{η ≤ x}, Ch{ξ ≥ x} = Pr{η ≥ x}, ∀x ∈ <.
It follows from (3.32) that E[ξ] = E[η]. In other words, the expected value
operator of hybrid variable coincides with that of random variable.
Example 3.11: If a hybrid variable ξ degenerates to a fuzzy variable ã, then

Ch{ξ ≤ x} = Cr{ã ≤ x}, Ch{ξ ≥ x} = Cr{ã ≥ x}, ∀x ∈ <.
It follows from (3.32) that E[ξ] = E[ã]. In other words, the expected value
operator of hybrid variable coincides with that of fuzzy variable.
Example 3.12: Let ã be a fuzzy variable and η a random variable with

finite expected values. Then the hybrid variable ξ = ã + η has expected value
E[ξ] = E[ã] + E[η].
Theorem 3.15 Let ξ be a hybrid variable whose chance density function φ
exists. If the Lebesgue integral
Z +∞
xφ(x)dx
−∞

Z +∞
E[ξ] = xφ(x)dx. (3.33)
−∞
Theorem that
Z +∞ Z 0
E[ξ] = Ch{ξ ≥ r}dr − Ch{ξ ≤ r}dr
0 −∞
Z +∞ Z +∞ Z 0 Z r
0 r −∞ −∞
Z +∞ Z x Z 0 Z 0
0 0 −∞ x
Z +∞ Z 0
0 −∞
Z +∞
= xφ(x)dx.
−∞
Theorem 3.16 Let ξ be a hybrid variable with chance distribution Φ. If
lim Φ(x) = 0, lim Φ(x) = 1

x→−∞ x→+∞

Z +∞
xdΦ(x)
−∞

Z +∞
E[ξ] = xdΦ(x). (3.34)
−∞
R +∞
Proof: Since the Lebesgue-Stieltjes integral −∞
xdΦ(x) is finite, we imme-
diately have
Z y Z +∞ Z 0 Z 0
y→+∞ 0 0 y→−∞ y −∞
and Z +∞ Z y
y→+∞ y y→−∞ −∞
It follows from
Z +∞
y z→+∞
Z y
−∞ z→−∞
that
lim y (1 − Φ(y)) = 0, lim yΦ(y) = 0.
y→+∞ y→−∞
n−1
X Z y
xi (Φ(xi+1 ) − Φ(xi )) → xdΦ(x)
i=0 0
and
n−1
X Z y
(1 − Φ(xi+1 ))(xi+1 − xi ) → Ch{ξ ≥ r}dr
i=0 0
n−1
X n−1
X
i=0 i=0

Z +∞ Z +∞
Ch{ξ ≥ r}dr = xdΦ(x).
0 0
Z 0 Z 0
− Ch{ξ ≤ r}dr = xdΦ(x).
−∞ −∞

Theorem 3.17 Let ξ be a hybrid variable with finite expected values. Then
for any real numbers a and b, we have
E[aξ + b] = aE[ξ] + b. (3.35)
Z +∞ Z 0
E[ξ + b] = Ch{ξ + b ≥ r}dr − Ch{ξ + b ≤ r}dr
0 −∞
Z +∞ Z 0
= Ch{ξ ≥ r − b}dr − Ch{ξ ≤ r − b}dr
0 −∞
Z b
= E[ξ] + (Ch{ξ ≥ r − b} + Ch{ξ < r − b})dr
0
= E[ξ] + b.

Z 0
E[aξ + b] = E[ξ] − (Ch{ξ ≥ r − b} + Ch{ξ < r − b})dr = E[ξ] + b.
b
Step 2: We prove E[aξ] = aE[ξ]. If a = 0, then the equation E[aξ] =

aE[ξ] holds trivially. If a > 0, we have
Z +∞ Z 0
E[aξ] = Ch{aξ ≥ r}dr − Ch{aξ ≤ r}dr
0 −∞
Z +∞ Z 0
= Ch{ξ ≥ r/a}dr − Ch{ξ ≤ r/a}dr
0 −∞
Z +∞ Z 0
=a Ch{ξ ≥ t}dt − a Ch{ξ ≤ t}dt
0 −∞
= aE[ξ].
If a < 0, we have
Z +∞ Z 0
E[aξ] = Ch{aξ ≥ r}dr − Ch{aξ ≤ r}dr
0 −∞
Z +∞ Z 0
= Ch{ξ ≤ r/a}dr − Ch{ξ ≥ r/a}dr
0 −∞
Z +∞ Z 0
=a Ch{ξ ≥ t}dt − a Ch{ξ ≤ t}dt
0 −∞
= aE[ξ].
Step 3: For any real numbers a and b, it follows from Steps 1 and 2 that
E[aξ + b] = E[aξ] + b = aE[ξ] + b.
3.5 Variance
Definition 3.13 (Li and Liu [107]) Let ξ be a hybrid variable with finite
expected value e. Then the variance of ξ is defined by V [ξ] = E[(ξ − e)2 ].
Theorem 3.18 If ξ is a hybrid variable with finite expected value, a and b

are real numbers, then V [aξ + b] = a2 V [ξ].

Theorem 3.19 Let ξ be a hybrid variable with expected value e. Then V [ξ] =
0 if and only if Ch{ξ = e} = 1.

Z +∞
2
E[(ξ − e) ] = Ch{(ξ − e)2 ≥ r}dr
0
which implies Ch{(ξ − e)2 ≥ r} = 0 for any r > 0. Hence we have
Ch{(ξ − e)2 = 0} = 1.
That is, Ch{ξ = e} = 1.

Conversely, if Ch{ξ = e} = 1, then we have Ch{(ξ − e)2 = 0} = 1 and
Ch{(ξ − e)2 ≥ r} = 0 for any r > 0. Thus
Z +∞
V [ξ] = Ch{(ξ − e)2 ≥ r}dr = 0.
0

Theorem 3.20 Let f be a convex function on [a, b], and ξ a hybrid variable
b−e e−a
E[f (ξ)] ≤ f (a) + f (b). (3.36)
b−a b−a
Proof: For each (θ, ω) ∈ Θ × Ω, we have a ≤ ξ(θ, ω) ≤ b and
b − ξ(θ, ω) ξ(θ, ω) − a
ξ(θ, ω) = a+ b.
b−a b−a
b − ξ(θ, ω) ξ(θ, ω) − a
f (ξ(θ, ω)) ≤ f (a) + f (b).
b−a b−a
Theorem 3.21 (Maximum Variance Theorem) Let ξ be a hybrid variable

V [ξ] ≤ (e − a)(b − e). (3.37)

3.6 Moments
Definition 3.14 (Li and Liu [107]) Let ξ be a hybrid variable. Then for any
positive integer k,
Theorem 3.22 Let ξ be a nonnegative hybrid variable, and k a positive num-
ber. Then the k-th moment
Z +∞
k
E[ξ ] = k rk−1 Ch{ξ ≥ r}dr. (3.38)
0

Z ∞ Z ∞ Z ∞
k k k
E[ξ ] = Ch{ξ ≥ x}dx = Ch{ξ ≥ r}dr = k rk−1 Ch{ξ ≥ r}dr.
0 0 0

Theorem 3.23 (Li and Liu [111]) Let ξ be a hybrid variable that takes val-
ues in [a, b] and has expected value e. Then for any positive integer k, the
kth absolute moment and kth absolute central moment satisfy the following
inequalities,
b−e k e−a k
E[|ξ|k ] ≤ |a| + |b| , (3.39)
b−a b−a
b−e e−a
E[|ξ − e|k ] ≤ (e − a)k + (b − e)k . (3.40)
b−a b−a
and f (x) = |x − e|k .
3.7 Critical Values

In order to rank hybrid variables, we introduce the following definition of
critical values of hybrid variables.
Definition 3.15 (Li and Liu [107]) Let ξ be a hybrid variable, and α ∈ (0, 1].
Then
ξsup (α) = sup r Ch {ξ ≥ r} ≥ α (3.41)

ξinf (α) = inf r Ch {ξ ≤ r} ≥ α (3.42)
The hybrid variable ξ reaches upwards of the α-optimistic value ξsup (α),
and is below the α-pessimistic value ξinf (α) with chance α.
Example 3.13: If a hybrid variable ξ degenerates to a random variable η,

then
Ch{ξ ≤ x} = Pr{η ≤ x}, Ch{ξ ≥ x} = Pr{η ≥ x}, ∀x ∈ <.
It follows from the definition of critical values that
ξsup (α) = ηsup (α), ξinf (α) = ηinf (α), ∀α ∈ (0, 1].
In other words, the critical values of hybrid variable coincide with that of
random variable.
Example 3.14: If a hybrid variable ξ degenerates to a fuzzy variable ã, then
Ch{ξ ≤ x} = Cr{ã ≤ x}, Ch{ξ ≥ x} = Cr{ã ≥ x}, ∀x ∈ <.
It follows from the definition of critical values that
ξsup (α) = ãsup (α), ξinf (α) = ãinf (α), ∀α ∈ (0, 1].
In other words, the critical values of hybrid variable coincide with that of
fuzzy variable.
Theorem 3.24 Let ξ be a hybrid variable. If α > 0.5, then we have
Ch{ξ ≤ ξinf (α)} ≥ α, Ch{ξ ≥ ξsup (α)} ≥ α. (3.43)
Proof: It follows from the definition of α-pessimistic value that there exists
a decreasing sequence {xi } such that Ch{ξ ≤ xi } ≥ α and xi ↓ ξinf (α) as
i → ∞. Since {ξ ≤ xi } ↓ {ξ ≤ ξinf (α)} and limi→∞ Ch{ξ ≤ xi } ≥ α > 0.5, it
follows from the chance semicontinuity theorem that
Ch{ξ ≤ ξinf (α)} = lim Ch{ξ ≤ xi } ≥ α.

i→∞
Similarly, there exists an increasing sequence {xi } such that Ch{ξ ≥

xi } ≥ α and xi ↑ ξsup (α) as i → ∞. Since {ξ ≥ xi } ↓ {ξ ≥ ξsup (α)}
and limi→∞ Ch{ξ ≥ xi } ≥ α > 0.5, it follows from the chance semicontinuity
theorem that
Ch{ξ ≥ ξsup (α)} = lim Ch{ξ ≥ xi } ≥ α.
i→∞
Theorem 3.25 Let ξ be a hybrid variable and α a number between 0 and 1.

We have
Section 3.8 - Distance 139
Proof: (a) If c = 0, then the part (a) is obvious. In the case of c > 0, we
have
(cξ)sup (α) = sup{r Ch{cξ ≥ r} ≥ α}
= c sup{r/c | Ch{ξ ≥ r/c} ≥ α}
= cξsup (α).
A similar way may prove (cξ)inf (α) = cξinf (α). In order to prove the part (b),
it suffices to prove that (−ξ)sup (α) = −ξinf (α) and (−ξ)inf (α) = −ξsup (α).
In fact, we have

(−ξ)sup (α) = sup{r Ch{−ξ ≥ r} ≥ α}
= − inf{−r | Ch{ξ ≤ −r} ≥ α}
= −ξinf (α).
Theorem 3.26 Let ξ be a hybrid variable. Then we have
¯
then we have
¯
1 ≥ Ch{ξ < ξ(α)} ¯
+ Ch{ξ > ξ(α)} ≥ α + α > 1.
A contradiction proves ξinf (α) ≥ ξsup (α).
Part (b): Assume that ξinf (α) > ξsup (α). It follows from the definition of
¯
ξinf (α) that Ch{ξ ≤ ξ(α)} < α. Similarly, it follows from the definition of
¯
ξsup (α) that Ch{ξ ≥ ξ(α)} < α. Thus
¯
1 ≤ Ch{ξ ≤ ξ(α)} ¯
+ Ch{ξ ≥ ξ(α)} < α + α ≤ 1.
A contradiction proves ξinf (α) ≤ ξsup (α). The theorem is verified.
Theorem 3.27 Let ξ be a hybrid variable. Then we have
(a) ξsup (α) is a decreasing and left-continuous function of α;
(b) ξinf (α) is an increasing and left-continuous function of α.
i→∞
Thus Ch{ξ ≤ z ∗ } ≥ αi for each i. Letting i → ∞, we get Ch{ξ ≤ z ∗ } ≥ α.

3.8 Distance
Definition 3.16 (Li and Liu [107]) The distance between hybrid variables ξ
and η is defined as
d(ξ, η) = E[|ξ − η|]. (3.44)
Theorem 3.28 (Li and Liu [107]) Let ξ, η, τ be hybrid variables, and let
d(·, ·) be the distance. Then we have
Now we prove the part (d). It follows from the chance subadditivity theorem
that
Z +∞
d(ξ, η) = Ch {|ξ − η| ≥ r} dr
0
Z +∞
≤ Ch {|ξ − τ | + |τ − η| ≥ r} dr
0
Z +∞
≤ Ch {{|ξ − τ | ≥ r/2} ∪ {|τ − η| ≥ r/2}} dr
0
Z +∞
≤ (Ch{|ξ − τ | ≥ r/2} + Ch{|τ − η| ≥ r/2}) dr
0
Z +∞ Z +∞
= Ch{|ξ − τ | ≥ r/2}dr + Ch{|τ − η| ≥ r/2}dr
0 0
= 2E[|ξ − τ |] + 2E[|τ − η|] = 2d(ξ, τ ) + 2d(τ, η).
3.9 Inequalities
Theorem 3.29 (Li and Liu [107]) Let ξ be a hybrid variable, and f a non-
negative function. If f is even and increasing on [0, ∞), then for any given
number t > 0, we have
E[f (ξ)]
Ch{|ξ| ≥ t} ≤ . (3.45)
f (t)
Proof: It is clear that Ch{|ξ| ≥ f −1 (r)} is a monotone decreasing function

Z +∞
E[f (ξ)] = Ch{f (ξ) ≥ r}dr
0
Z +∞
= Ch{|ξ| ≥ f −1 (r)}dr
0
Z f (t)
≥ Ch{|ξ| ≥ f −1 (r)}dr
0
Z f (t)
≥ dr · Ch{|ξ| ≥ f −1 (f (t))}
0
= f (t) · Ch{|ξ| ≥ t}
Theorem 3.30 (Li and Liu [107], Markov Inequality) Let ξ be a hybrid vari-
able. Then for any given numbers t > 0 and p > 0, we have
E[|ξ|p ]
Ch{|ξ| ≥ t} ≤ . (3.46)
tp
Theorem 3.31 (Li and Liu [107], Chebyshev Inequality) Let ξ be a hybrid
variable whose variance V [ξ] exists. Then for any given number t > 0, we
have
V [ξ]
Ch {|ξ − E[ξ]| ≥ t} ≤ 2 . (3.47)
t
Proof: It is a special case of Theorem 3.29 when the hybrid variable ξ is
Theorem 3.32 (Hölder’s Inequality) Let p and q be positive real numbers

with 1/p + 1/q = 1. If ξ is a fuzzy variable with E[|ξ|p ] < ∞ and η is a
random variable with E[|η|q ] < ∞, then we have
p p
E[|ξη|] ≤ p E[|ξ|p ] q E[|η|q ]. (3.48)
√
a and b such that

E[f (|ξ|p , |η|q )] ≤ f (E[|ξ|p ], E[|η|q ]).
Theorem 3.33 (Minkowski Inequality) Let p be a real number with p ≥ 1.

If ξ is a fuzzy variable with E[|ξ|p ] < ∞ and η is a random variable with
E[|η|p ] < ∞, then we have
p
p
p p
E[|ξ + η|p ] ≤ p E[|ξ|p ] + p E[|η|p ]. (3.49)
√
E[f (|ξ|p , |η|p )] ≤ f (E[|ξ|p ], E[|η|p ]).
Theorem 3.34 (Jensen’s Inequality) Let ξ be a hybrid variable, and f : < →

< a convex function. If E[ξ] and E[f (ξ)] are finite, then
f (E[ξ]) ≤ E[f (ξ)]. (3.50)
f (ξ) − f (E[ξ]) ≥ k · (ξ − E[ξ]).
E[f (ξ)] − f (E[ξ]) ≥ k · (E[ξ] − E[ξ]) = 0


Li and Liu [107] discussed the convergence concepts of hybrid sequence: con-
vergence almost surely (a.s.), convergence in chance, convergence in mean,
Table 3.1: Relationship among Convergence Concepts
Convergence Convergence Convergence

⇒ ⇒
in Mean in Chance in Distribution
Definition 3.17 Suppose that ξ, ξ1 , ξ2 , · · · are hybrid variables defined on

the chance space (Θ, P, Cr) × (Ω, A, Pr). The sequence {ξi } is said to be
convergent a.s. to ξ if there exists an event Λ with Ch{Λ} = 1 such that
lim |ξi (θ, ω) − ξ(θ, ω)| = 0 (3.51)

i→∞
for every (θ, ω) ∈ Λ. In that case we write ξi → ξ, a.s.
Definition 3.18 Suppose that ξ, ξ1 , ξ2 , · · · are hybrid variables. We say that

the sequence {ξi } converges in chance to ξ if
lim Ch {|ξi − ξ| ≥ ε} = 0 (3.52)

i→∞
for every ε > 0.
Definition 3.19 Suppose that ξ, ξ1 , ξ2 , · · · are hybrid variables with finite

expected values. We say that the sequence {ξi } converges in mean to ξ if
lim E[|ξi − ξ|] = 0. (3.53)

i→∞
lim E[|ξi − ξ|2 ] = 0. (3.54)

i→∞
Definition 3.20 Suppose that Φ, Φ1 , Φ2 , · · · are the chance distributions of

hybrid variables ξ, ξ1 , ξ2 , · · · , respectively. We say that {ξi } converges in dis-
tribution to ξ if Φi → Φ at any continuity point of Φ.
Convergence Almost Surely vs. Convergence in Chance
Example 3.15: Convergence a.s. does not imply convergence in chance.

Take a credibility space (Θ, P, Cr) to be {θ1 , θ2 , · · · } with Cr{θj } = j/(2j +1)
for j = 1, 2, · · · and take an arbitrary probability space (Ω, A, Pr). Then we

define hybrid variables as
(
i, if j = i
ξi (θj , ω) =
0, otherwise
for i = 1, 2, · · · and ξ ≡ 0. The sequence {ξi } convergence a.s. to ξ. However,

for some small number ε > 0, we have
i 1
Ch{|ξi − ξ| ≥ ε} = Cr{|ξi − ξ| ≥ ε} = → .
2i + 1 2
That is, the sequence {ξi } does not converge in chance to ξ.
Example 3.16: Convergence in chance does not imply convergence a.s.

Take an arbitrary credibility space (Θ, P, Cr) and take a probability space
(Ω, A, Pr) to be [0, 1] with Borel algebra and Lebesgue measure. For any
positive integer i, there is an integer j such that i = 2j + k, where k is an
integer between 0 and 2j − 1. Then we define hybrid variables as
(
i, if k/2j ≤ ω ≤ (k + 1)/2j
ξi (θ, ω) =
0, otherwise
for i = 1, 2, · · · and ξ ≡ 0. For some small number ε > 0, we have
1
Ch{|ξi − ξ| ≥ ε} = Pr{|ξi − ξ| ≥ ε} = →0
2i
as i → ∞. That is, the sequence {ξi } converges in chance to ξ. However, for
any ω ∈ [0, 1], there is an infinite number of intervals of the form [k/2j , (k +
1)/2j ] containing ω. Thus ξi (θ, ω) does converges to 0. In other words, the
sequence {ξi } does not converge a.s. to ξ.
Convergence in Mean vs. Convergence in Chance

Theorem 3.35 Suppose that ξ, ξ1 , ξ2 , · · · are hybrid variables. If {ξi } con-
verges in mean to ξ, then {ξi } converges in chance to ξ.
Proof: It follows from the Markov inequality that for any given number
ε > 0, we have
E[|ξi − ξ|]
Ch{|ξi − ξ| ≥ ε} ≤ →0
ε
as i → ∞. Thus {ξi } converges in chance to ξ. The theorem is proved.
Example 3.17: Convergence in chance does not imply convergence in mean.

Take a credibility space (Θ, P, Cr) to be {θ1 , θ2 , · · · } with Cr{θ1 } = 1/2
and Cr{θj } = 1/j for j = 2, 3, · · · and take an arbitrary probability space

(Ω, A, Pr). The hybrid variables are defined by
(
i, if j = i
ξi (θj , ω) =
0, otherwise
for i = 1, 2, · · · and ξ ≡ 0. For some small number ε > 0, we have

1
Ch{|ξi − ξ| ≥ ε} = Cr{|ξi − ξ| ≥ ε} = → 0.
i
That is, the sequence {ξi } converges in chance to ξ. However,
E[|ξi − ξ|] = 1, ∀i.
Convergence Almost Surely vs. Convergence in Mean
Example 3.18: Convergence a.s. does not imply convergence in mean.

(Ω, A, Pr) to be {ω1 , ω2 , · · · } with Pr{ωj } = 1/2j for j = 1, 2, · · · The hybrid
variables are defined by
(
2i , if j = i
ξi (θ, ωj ) =
0, otherwise
for i = 1, 2, · · · and ξ ≡ 0. Then ξi converges a.s. to ξ. However, the sequence

{ξi } does not converges in mean to ξ because E[|ξi − ξ|] ≡ 1.
Example 3.19: Convergence in chance does not imply convergence a.s.

(Ω, A, Pr) to be [0, 1] with Borel algebra and Lebesgue measure. For any
positive integer i, there is an integer j such that i = 2j + k, where k is an
integer between 0 and 2j − 1. The hybrid variables are defined by
(
i, if k/2j ≤ ω ≤ (k + 1)/2j
ξi (θ, ω) =
0, otherwise
for i = 1, 2, · · · and ξ ≡ 0. Then

1
E[|ξi − ξ|] = → 0.
2j
That is, the sequence {ξi } converges in mean to ξ. However, for any ω ∈ [0, 1],
there is an infinite number of intervals of the form [k/2j , (k+1)/2j ] containing
ω. Thus ξi (θ, ω) does converges to 0. In other words, the sequence {ξi } does
not converge a.s. to ξ.
Convergence in Chance vs. Convergence in Distribution

Theorem 3.36 Suppose ξ, ξ1 , ξ2 , · · · are hybrid variables. If {ξi } converges
in chance to ξ, then {ξi } converges in distribution to ξ.
Proof: Let x be a given continuity point of the distribution Φ. On the one

hand, for any y > x, we have
It follows from the chance subadditivity theorem that
Φi (x) ≤ Φ(y) + Ch{|ξi − ξ| ≥ y − x}.
Since {ξi } converges in chance to ξ, we have Ch{|ξi − ξ| ≥ y − x} → 0 as

i → ∞. Thus we obtain lim supi→∞ Φi (x) ≤ Φ(y) for any y > x. Letting
y → x, we get
lim sup Φi (x) ≤ Φ(x). (3.55)
i→∞
{ξ ≤ z} = {ξi ≤ x, ξ ≤ z} ∪ {ξi > x, ξ ≤ z} ⊂ {ξi ≤ x} ∪ {|ξi − ξ| ≥ x − z}
which implies that
Φ(z) ≤ Φi (x) + Ch{|ξi − ξ| ≥ x − z}.
Since Ch{|ξi − ξ| ≥ x − z} → 0, we obtain Φ(z) ≤ lim inf i→∞ Φi (x) for any
Φ(x) ≤ lim inf Φi (x). (3.56)

i→∞

in chance. Take a credibility space (Θ, P, Cr) to be {θ1 , θ2 } with Cr{θ1 } =
Cr{θ2 } = 1/2 and take an arbitrary probability space (Ω, A, Pr). We define
a hybrid variables as
(
−1, if θ = θ1
ξ(θ, ω) =
1, if θ = θ2 .
We also define ξi = −ξ for i = 1, 2, · · · . Then ξi and ξ have the same chance

distribution. Thus {ξi } converges in distribution to ξ. However, for some
small number ε > 0, we have
Ch{|ξi − ξ| ≥ ε} = Cr{|ξi − ξ| ≥ ε} = 1.
That is, the sequence {ξi } does not converge in chance to ξ.

Section 3.11 - Conditional Chance 147
Convergence Almost Surely vs. Convergence in Distribution

a.s. Take a credibility space to be (Θ, P, Cr) to be {θ1 , θ2 } with Cr{θ1 } =
Cr{θ2 } = 1/2 and take an arbitrary probability space (Ω, A, Pr). We define
a hybrid variable ξ as
(
−1, if θ = θ1
ξ(θ, ω) =
1, if θ = θ2 .
We also define ξi = −ξ for i = 1, 2, · · · . Then ξi and ξ have the same chance

distribution. Thus {ξi } converges in distribution to ξ. However, the sequence
{ξi } does not converge a.s. to ξ.
Example 3.22: Convergence a.s. does not imply convergence in chance.

Take a credibility space (Θ, P, Cr) to be {θ1 , θ2 , · · · } with Cr{θj } = j/(2j +1)
for j = 1, 2, · · · and take an arbitrary probability space (Ω, A, Pr). The hybrid
variables are defined by
(
i, if j = i
ξi (θj , ω) =
0, otherwise
for i = 1, 2, · · · and ξ ≡ 0. Then the sequence {ξi } converges a.s. to ξ.

However, the chance distributions of ξi are


 0, if x < 0
Φi (x) = (i + 1)/(2i + 1), if 0 ≤ x < i

1, if x ≥ i

for i = 1, 2, · · · , respectively. The chance distribution of ξ is

0, if x < 0
Φ(x) =
1, if x ≥ 0.
It is clear that Φi (x) does not converge to Φ(x) at x > 0. That is, the
sequence {ξi } does not converge in distribution to ξ.
3.11 Conditional Chance

We consider the chance measure of an event A after it has been learned that
some other event B has occurred. This new chance measure of A is called
the conditional chance measure of A given B.
In order to define a conditional chance measure Ch{A|B}, at first we
have to enlarge Ch{A ∩ B} because Ch{A ∩ B} < 1 for all events whenever
Ch{B} < 1. It seems that we have no alternative but to divide Ch{A∩B} by
Ch{B}. Unfortunately, Ch{A ∩ B}/Ch{B} is not always a chance measure.

However, the value Ch{A|B} should not be greater than Ch{A ∩ B}/Ch{B}
Ch{A ∩ B}
Ch{A|B} ≤ . (3.57)
Ch{B}
Ch{Ac ∩ B}
Ch{A|B} = 1 − Ch{Ac |B} ≥ 1 − . (3.58)
Ch{B}
Furthermore, since (A ∩ B) ∪ (Ac ∩ B) = B, we have Ch{B} ≤ Ch{A ∩ B} +

Ch{Ac ∩ B} by using the chance subadditivity theorem. Thus
Ch{Ac ∩ B} Ch{A ∩ B}
0≤1− ≤ ≤ 1. (3.59)
Ch{B} Ch{B}
Hence any numbers between 1 − Ch{Ac ∩ B}/Ch{B} and Ch{A ∩ B}/Ch{B}

are reasonable values that the conditional chance may take. Based on the
maximum uncertainty principle (see Appendix E), we have the following con-
ditional chance measure.
Definition 3.21 (Li and Liu [110]) Let (Θ, P, Cr) × (Ω, A, Pr) be a chance
space and A, B two events. Then the conditional chance measure of A given
B is defined by
Ch{A ∩ B} Ch{A ∩ B}


 , if < 0.5


 Ch{B} Ch{B}

Ch{A|B} = Ch{Ac ∩ B} Ch{Ac ∩ B} (3.60)
1− , if < 0.5
Ch{B} Ch{B}






0.5, otherwise
provided that Ch{B} > 0.
Remark 3.6: It follows immediately from the definition of conditional

chance that
Ch{Ac ∩ B} Ch{A ∩ B}
1− ≤ Ch{A|B} ≤ . (3.61)
Ch{B} Ch{B}
Furthermore, it is clear that the conditional chance measure obeys the max-
imum uncertainty principle.
Remark 3.7: Let X and Y be events in the credibility space. Then the
conditional chance measure of X × Ω given Y × Ω is
Cr{X ∩ Y } Cr{X ∩ Y }

 , if < 0.5
Cr{Y } Cr{Y }





Ch{X × Ω|Y × Ω} = Cr{X c ∩ Y } Cr{X c ∩ Y }
1− , if < 0.5
Cr{Y } Cr{Y }






0.5, otherwise
which is just the conditional credibility of X given Y .
Remark 3.8: Let X and Y be events in the probability space. Then the
conditional chance measure of Θ × X given Θ × Y is
Pr{X ∩ Y }
Ch{Θ × X|Θ × Y } =
Pr{Y }
which is just the conditional probability of X given Y .
Example 3.23: Let ξ and η be two hybrid variables. Then we have

Ch{ξ = x, η = y} Ch{ξ = x, η = y}


 , if < 0.5


 Ch{η = y} Ch{η = y}

Ch {ξ = x|η = y} = Ch{ξ 6= x, η = y} Ch{ξ 6= x, η = y}
1− , if < 0.5
Ch{η = y} Ch{η = y}






0.5, otherwise
provided that Ch{η = y} > 0.
Theorem 3.37 (Li and Liu [110]) Conditional chance measure is normal,
increasing, self-dual and countably subadditive.
Proof: At first, the conditional chance measure Ch{·|B} is normal, i.e.,

Ch{∅}
Ch{Θ × Ω|B} = 1 − = 1.
Ch{B}
Ch{A1 ∩ B} Ch{A2 ∩ B}
≤ < 0.5,
Ch{B} Ch{B}
then
Ch{A1 ∩ B} Ch{A2 ∩ B}
Ch{A1 |B} = ≤ = Ch{A2 |B}.
Ch{B} Ch{B}
If
Ch{A1 ∩ B} Ch{A2 ∩ B}
≤ 0.5 ≤ ,
Ch{B} Ch{B}
then Ch{A1 |B} ≤ 0.5 ≤ Ch{A2 |B}. If
Ch{A1 ∩ B} Ch{A2 ∩ B}
0.5 < ≤ ,
Ch{B} Ch{B}
then we have
Ch{Ac1 ∩ B} Ch{Ac2 ∩ B}

Ch{A1 |B} = 1 − ∨0.5 ≤ 1 − ∨0.5 = Ch{A2 |B}.
Ch{B} Ch{B}
This means that Ch{·|B} is increasing. For any event A, if
Ch{A ∩ B} Ch{Ac ∩ B}
≥ 0.5, ≥ 0.5,
Ch{B} Ch{B}
then we have Ch{A|B} + Ch{Ac |B} = 0.5 + 0.5 = 1 immediately. Otherwise,

Ch{A ∩ B} Ch{Ac ∩ B}
< 0.5 < ,
Ch{B} Ch{B}
then we have

Ch{A ∩ B} Ch{A ∩ B}
Ch{A|B} + Ch{Ac |B} = + 1− = 1.
Ch{B} Ch{B}
That is, Ch{·|B} is self-dual. Finally, for any countable sequence {Ai } of
events, if Ch{Ai |B} < 0.5 for all i, it follows from the countable subadditivity
of chance measure that
(∞ ) ∞
[ X
(∞ ) Ch Ai ∩ B Ch{Ai ∩ B} ∞
[ i=1 i=1
X
Ch Ai ∩ B ≤ ≤ = Ch{Ai |B}.
i=1
Ch{B} Ch{B} i=1
Suppose there is one term greater than 0.5, say
Ch{A1 |B} ≥ 0.5, Ch{Ai |B} < 0.5, i = 2, 3, · · ·
If Ch{∪i Ai |B} = 0.5, then we immediately have

(∞ ) ∞
[ X
Ch Ai ∩ B ≤ Ch{Ai |B}.
i=1 i=1
If Ch{∪i Ai |B} > 0.5, we may prove the above inequality by the following
facts:
∞ ∞
!
[ \
c c
A1 ∩ B ⊂ (Ai ∩ B) ∪ Ai ∩ B ,
i=2 i=1
∞
(∞ )
X \
Ch{Ac1 ∩ B} ≤ Ch{Ai ∩ B} + Ch Aci ∩B ,
i=2 i=1
(∞ )
\
(∞ ) Ch Aci ∩B
[ i=1
Ch Ai |B =1− ,
i=1
Ch{B}
∞
X
∞
Ch{Ai ∩ B}
X Ch{Ac1 ∩ B} i=2
Ch{Ai |B} ≥ 1 − + .
i=1
Ch{B} Ch{B}
If there are at least two terms greater than 0.5, then the countable subad-
ditivity is clearly true. Thus Ch{·|B} is countably subadditive. Hence the
theorem is verified.
Definition 3.22 Let ξ be a hybrid variable on (Θ, P, Cr) × (Ω, A, Pr). A
conditional hybrid variable of ξ given B is a measurable function ξ|B from
the conditional chance space to the set of real numbers such that
ξ|B (θ, ω) ≡ ξ(θ, ω), ∀(θ, ω) ∈ Θ × Ω. (3.62)
Definition 3.23 (Li and Liu [110]) The conditional chance distribution Φ:
< → [0, 1] of a hybrid variable ξ given B is defined by
Φ(x|B) = Ch {ξ ≤ x|B} (3.63)
provided that Ch{B} > 0.
Example 3.24: Let ξ and η be hybrid variables. Then the conditional

chance distribution of ξ given η = y is
Ch{ξ ≤ x, η = y} Ch{ξ ≤ x, η = y}


 , if < 0.5


 Ch{η = y} Ch{η = y}

Φ(x|η = y) = Ch{ξ > x, η = y} Ch{ξ > x, η = y}
1− , if < 0.5
Ch{η = y} Ch{η = y}






0.5, otherwise
provided that Ch{η = y} > 0.
Definition 3.24 (Li and Liu [110]) The conditional chance density function
φ of a hybrid variable ξ given B is a nonnegative function such that
Z x
Φ(x|B) = φ(y|B)dy, ∀x ∈ <, (3.64)
−∞
Z +∞
φ(y|B)dy = 1 (3.65)
−∞
where Φ(x|B) is the conditional chance distribution of ξ given B.
Definition 3.25 (Li and Liu [110]) Let ξ be a hybrid variable. Then the
conditional expected value of ξ given B is defined by
Z +∞ Z 0
E[ξ|B] = Ch{ξ ≥ r|B}dr − Ch{ξ ≤ r|B}dr (3.66)
0 −∞
Following conditional chance and conditional expected value, we also have

conditional variance, conditional moments, conditional critical values, condi-
tional entropy as well as conditional convergence.
Chapter 4
Uncertainty Theory
A lot of surveys showed that the measure of union of events is not necessarily
the maximum measure except those events are independent. This fact im-
plies that many subjectively uncertain systems do not behave fuzziness and
membership function is no longer helpful. In order to deal with this type of
subjective uncertainty, Liu [138] founded an uncertainty theory in 2007 that
is a branch of mathematics based on normality, monotonicity, self-duality,
countable subadditivity, and product measure axioms.
The emphasis in this chapter is mainly on uncertain measure, uncer-
tainty space, uncertain variable, identification function, uncertainty distri-
bution, expected value, variance, moments, independence, identical distribu-
tion, critical values, entropy, distance, convergence almost surely, convergence
in measure, convergence in mean, convergence in distribution, and conditional
uncertainty.
4.1 Uncertainty Space
Let Γ be a nonempty set, and let L be a σ-algebra over Γ. Each element Λ ∈ L

is called an event. In order to present an axiomatic definition of uncertain
measure, it is necessary to assign to each event Λ a number M{Λ} which
indicates the level that Λ will occur. In order to ensure that the number
M{Λ} has certain mathematical properties, Liu [138] proposed the following
four axioms:
Axiom 1. (Normality) M{Γ} = 1.
Axiom 2. (Monotonicity) M{Λ1 } ≤ M{Λ2 } whenever Λ1 ⊂ Λ2 .
Axiom 3. (Self-Duality) M{Λ} + M{Λc } = 1 for any event Λ.
Axiom 4. (Countable Subadditivity) For every countable sequence of events
154 Chapter 4 - Uncertainty Theory
{Λi }, we have
(∞ ) ∞
M M{Λi }.
[ X
Λi ≤ (4.1)
i=1 i=1
Remark 4.1: The law of contradiction tells us that a proposition cannot

be both true and false at the same time, and the law of excluded middle
tells us that a proposition is either true or false. Self-duality is in fact a
generalization of the law of contradiction and law of excluded middle. In
other words, a mathematical system without self-duality assumption will be
inconsistent with the laws. This is the main reason why self-duality axiom is
assumed.
Remark 4.2: Pathology occurs if subadditivity is not assumed. For ex-

ample, suppose that a universal set contains 3 elements. We define a set
function that takes value 0 for each singleton, and 1 for each set with at least
2 elements. Then such a set function satisfies all axioms but subadditivity.
Is it not strange if such a set function serves as a measure?
Remark 4.3: Pathology occurs if countable subadditivity axiom is replaced

with finite subadditivity axiom. For example, assume the universal set con-
sists of all real numbers. We define a set function that takes value 0 if the
set is bounded, 0.5 if both the set and complement are unbounded, and 1 if
the complement of the set is bounded. Then such a set function is finitely
subadditive but not countably subadditive. Is it not strange if such a set
function serves as a measure? This is the main reason why we accept the
countable subadditivity axiom.
Definition 4.1 (Liu [138]) The set function M is called an uncertain mea-
sure if it satisfies the normality, monotonicity, self-duality, and countable
subadditivity axioms.
Example 4.1: Let Γ = {γ1 , γ2 , γ3 }. For this case, there are only 8 events.
Define
M{γ1 } = 0.6, M{γ2 } = 0.3, M{γ3 } = 0.2,
M{γ1 , γ2 } = 0.8, M{γ1 , γ3 } = 0.7, M{γ2 , γ3 } = 0.4,
M{∅} = 0, M{Γ} = 1.
Then M is an uncertain measure because it satisfies the four axioms.
Example 4.2: Suppose that λ(x) is a nonnegative function on < satisfying
sup (λ(x) + λ(y)) = 1. (4.2)

x6=y
Section 4.1 - Uncertainty Space 155
Then for any set Λ of real numbers, the set function


 sup λ(x), if sup λ(x) < 0.5
M{Λ} = 
x∈Λ x∈Λ

(4.3)
 1 − sup λ(x), if sup λ(x) ≥ 0.5
x∈Λc x∈Λ
is an uncertain measure on <.
Example 4.3: Suppose ρ(x) is a nonnegative and integrable function on <

such that Z
ρ(x)dx ≥ 1. (4.4)
<
Then for any Borel set Λ of real numbers, the set function
 Z Z


 ρ(x)dx, if ρ(x)dx < 0.5


 Λ Λ
M{Λ} =  1 − ρ(x)dx,
 Z Z
(4.5)
if ρ(x)dx < 0.5
Λc Λc






0.5, otherwise
Example 4.4: Suppose λ(x) is a nonnegative function and ρ(x) is a non-

negative and integrable function on < such that
Z Z
sup λ(x) + ρ(x)dx ≥ 0.5 and/or sup λ(x) + ρ(x)dx ≥ 0.5 (4.6)
x∈Λ Λ x∈Λc Λc
for any Borel set Λ of real numbers. Then the set function
 Z Z


 sup λ(x) + ρ(x)dx, if sup λ(x) + ρ(x)dx < 0.5


 x∈Λ Λ x∈Λ Λ
M{Λ} =  1 − sup λ(x) − ρ(x)dx, if sup λ(x) + ρ(x)dx < 0.5

 Z Z
x∈Λc Λc x∈Λc Λc






0.5, otherwise
Theorem 4.1 Suppose that M is an uncertain measure. Then we have

M{∅} = 0, (4.7)
0 ≤ M{Λ} ≤ 1 (4.8)
for any event Λ.
Proof: It follows from the normality and self-duality axioms that M{∅} =
1 − M{Γ} = 1 − 1 = 0. It follows from the monotonicity axiom that 0 ≤
M{Λ} ≤ 1 because ∅ ⊂ Λ ⊂ Γ.
Theorem 4.2 Suppose that M is an uncertain measure. Then for any events
Λ1 and Λ2 , we have
M{Λ1 } ∨ M{Λ2 } ≤ M{Λ1 ∪ Λ2 } ≤ M{Λ1 } + M{Λ2 }. (4.9)
Proof: The left-hand inequality follows from the monotonicity axiom and
the right-hand inequality follows from the countable subadditivity axiom im-
mediately.
Theorem 4.3 Suppose that M is an uncertain measure. Then for any events
Λ1 and Λ2 , we have
M{Λ1 } + M{Λ2 } − 1 ≤ M{Λ1 ∩ Λ2 } ≤ M{Λ1 } ∧ M{Λ2 }. (4.10)
Proof: The right-hand inequality follows from the monotonicity axiom and
the left-hand inequality follows from the self-duality and countable subaddi-
tivity axioms, i.e.,
M{λ1 ∩ Λ2 } = 1 − M{(Λ1 ∩ Λ2 )c } = 1 − M{Λc1 ∪ Λc2 }

≥ 1 − (M{Λc1 } + M{Λc2 })
= 1 − (1 − M{Λ1 }) − (1 − M{Λ2 })
= M{Λ1 } + M{Λ2 } − 1.
The inequalities are verified.
Theorem 4.4 Let Γ = {γ1 , γ2 , · · · }. If M is an uncertain measure, then

∞
M{γi } + M{γj } ≤ 1 ≤ M{γk }
X
(4.11)
k=1
for any i and j.
Proof: Since M is increasing and self-dual, we have, for any i and j,

M{γi } + M{γj } ≤ M{Γ\{γj }} + M{γj } = 1.
Since Γ = ∪k {γk } and M is countably subadditive, we have
∞ ∞
( )
1 = M{Γ} = M M{γk }.
[ X
{γk } ≤
k=1 k=1

Uncertainty Null-Additivity Theorem

Null-additivity is a direct deduction from subadditivity. We first prove a
more general theorem.
Theorem 4.5 Let {Λi } be a decreasing sequence of events with M{Λi } → 0

as i → ∞. Then for any event Λ, we have
lim
i→∞
M{Λ ∪ Λi } = i→∞
lim M{Λ\Λi } = M{Λ}. (4.12)
Proof: It follows from the monotonicity and countable subadditivity axioms

that
M{Λ} ≤ M{Λ ∪ Λi } ≤ M{Λ} + M{Λi }
for each i. Thus we get M{Λ ∪ Λi } → M{Λ} by using M{Λi } → 0. Since
(Λ\Λi ) ⊂ Λ ⊂ ((Λ\Λi ) ∪ Λi ), we have
M{Λ\Λi } ≤ M{Λ} ≤ M{Λ\Λi } + M{Λi }.

Hence M{Λ\Λi } → M{Λ} by using M{Λi } → 0.
Remark 4.4: It follows from the above theorem that the uncertain measure
is null-additive, i.e., M{Λ1 ∪ Λ2 } = M{Λ1 } + M{Λ2 } if either M{Λ1 } = 0
or M{Λ2 } = 0. In other words, the uncertain measure remains unchanged if
the event is enlarged or reduced by an event with measure zero.
Theorem 4.6 An uncertain measure is identical with not only probability

measure but also credibility measure if there are effectively two elements in
the universal set.
Proof: Suppose there are at most two elements, say γ1 and γ2 , taking
nonzero uncertain measure values. Let Λ1 and Λ2 be two disjoint events.
The argument breaks down into two cases.
Case 1: M{Λ1 } = 0 or M{Λ2 } = 0. For this case, we may verify that
M{Λ1 ∪ Λ2 } = M{Λ1 } + M{Λ2 } by using the countable subadditivity axiom.
Case 2: M{Λ1 } > 0 or M{Λ2 } > 0. For this case, without loss of general-
ity, we suppose that γ1 ∈ Λ1 and γ2 ∈ Λ2 . Note that M{(Λ1 ∪ Λ2 )c } = 0. It
follows from the self-duality axiom and uncertainty null-additivity theorem
that
M{Λ1 ∪ Λ2 } = M{Λ1 ∪ Λ2 ∪ (Λ1 ∪ Λ2 )c } = M{Γ} = 1,
M{Λ1 } + M{Λ2 } = M{Λ1 ∪ (Λ1 ∪ Λ2 )c } + M{Λ2 } = 1.
Hence M{Λ1 ∪ Λ2 } = M{Λ1 } + M{Λ2 }. The additivity is proved. Hence such
an uncertain measure is identical with probability measure. Furthermore, it
follows from Theorem 2.6 that it is also identical with credibility measure.
Uncertainty Asymptotic Theorem

Theorem 4.7 (Uncertainty Asymptotic Theorem) For any events Λ1 , Λ2 , · · · ,
we have
lim M{Λi } > 0, if Λi ↑ Γ, (4.13)
i→∞
lim
i→∞
M{Λi } < 1, if Λi ↓ ∅. (4.14)
Proof: Assume Λi ↑ Γ. Since Γ = ∪i Λi , it follows from the countable

subadditivity axioms that
∞
1 = M{Γ} ≤ M{Λi }.
X
i=1
Since M{Λi } is increasing with respect to i, we have limi→∞ M{Λi } > 0. If

Λi ↓ ∅, then Λci ↑ Γ. It follows from the first inequality and self-duality axiom
that
lim M{Λi } = 1 − lim M{Λci } < 1.
i→∞ i→∞
Example 4.5: Assume Γ is the set of real numbers. Let α be a number with
0 < α ≤ 0.5. Define a set function as follows,


 0, if Λ = ∅

 α, if Λ is upper bounded



M{Λ} =  0.5, if both Λ and Λc are upper unbounded (4.15)
c
1 − α, if Λ is upper bounded





1, if Λ = Γ.

It is easy to verify that M is an uncertain measure. Write Λi = (−∞, i] for

i = 1, 2, · · · Then Λi ↑ Γ and limi→∞ M{Λi } = α. Furthermore, we have
Λci ↓ ∅ and limi→∞ M{Λci } = 1 − α.
Uncertainty Space
Definition 4.2 (Liu [138]) Let Γ be a nonempty set, L a σ-algebra over
Γ, and M an uncertain measure. Then the triplet (Γ, L, M) is called an
uncertainty space.
Product Measure Axiom and Product Uncertain Measure

Product uncertain measure was defined by Liu [141] in 2009, thus producing
the fifth axiom of uncertainty theory called product measure axiom. Let
(Γk , Lk , Mk ) be uncertainty spaces for k = 1, 2, · · · , n. Write
Γ = Γ 1 × Γ2 × · · · × Γn , L = L1 × L2 × · · · × Ln .
The product uncertain measure is defined as follows.

Axiom 5. (Liu [141], Product Measure Axiom) Let Γk be nonempty sets
on which Mk are uncertain measures, k = 1, 2, · · · , n, respectively. Then the
product uncertain measure on Γ is
min Mk {Λk },

 sup


 Λ1 ×Λ2 ×···×Λn ⊂Λ 1≤k≤n
min Mk {Λk } > 0.5



if sup



Λ1 ×Λ2 ×···×Λn ⊂Λ 1≤k≤n




M{Λ} =  1 − min Mk {Λk },

sup (4.16)
 Λ1 ×Λ2 ×···×Λn ⊂Λc 1≤k≤n
min Mk {Λk } > 0.5






 if sup
Λ1 ×Λ2 ×···×Λn ⊂Λc 1≤k≤n






0.5, otherwise
for each event Λ ∈ L, denoted byM = M1 ∧ M2 ∧ · · · ∧ Mn .
Example 4.6: Let Λi be events in (Γk , Lk , Mk ), k = 1, 2, · · · , n, respectively.
Then the rectangle Λ1 × Λ2 × · · · × Λn is an event and the product uncertain
measure is
M{Λ1 × Λ2 × · · · × Λn } = min Mk {Λk }.
1≤k≤n
Especially, M{Γ} = 1 and M{∅} = 0.

Theorem 4.8 (Peng [203]) The product uncertain measure (4.16) is an un-
certain measure.
Proof: In order to prove that the product uncertain measure (4.16) is indeed
an uncertain measure, we should verify that the product uncertain measure
satisfies the normality, monotonicity, self-duality and countable subadditivity
axioms.
Step 1: At first, for any event Λ ∈ L, it is easy to verify that
sup min Mk {Λk } + sup min Mk {Λk } ≤ 1.
Λ1 ×Λ2 ×···×Λn ⊂Λ 1≤k≤n Λ1 ×Λ2 ×···×Λn ⊂Λc 1≤k≤n
This means that at most one of

sup min Mk {Λk } and sup min Mk {Λk }
Λ1 ×Λ2 ×···×Λn ⊂Λ 1≤k≤n Λ1 ×Λ2 ×···×Λn ⊂Λc 1≤k≤n
is greater than 0.5. Thus the expression (4.16) is reasonable.

Step 2: The product uncertain measure is clearly normal, i.e., M{Γ} = 1.
Step 3: We proves the self-duality, i.e., M{Λ} + M{Λc } = 1. The argu-
ment breaks down into three cases. Case 1: Assume
sup min Mk {Λk } > 0.5.
Λ1 ×Λ2 ×···×Λn ⊂Λ 1≤k≤n
Then we immediately have
sup min Mk {Λk } < 0.5.

Λ1 ×Λ2 ×···×Λn ⊂Λc 1≤k≤n
It follows from (4.16) that
M{Λ} = sup min Mk {Λk },

Λ1 ×Λ2 ×···×Λn ⊂Λ 1≤k≤n
M{Λc } = 1 − sup min Mk {Λk } = 1 − M{Λ}.

Λ1 ×Λ2 ×···×Λn ⊂(Λc )c 1≤k≤n
The self-duality is proved. Case 2: Assume

Λ1 ×Λ2 ×···×Λn ⊂Λc 1≤k≤n
This case may be proved by a similar process. Case 3: Assume
sup min Mk {Λk } ≤ 0.5

Λ1 ×Λ2 ×···×Λn ⊂Λ 1≤k≤n
and
sup min Mk {Λk } ≤ 0.5.
Λ1 ×Λ2 ×···×Λn ⊂Λc 1≤k≤n
It follows from (4.16) that M{Λ} = M{Λc } = 0.5 which proves the self-
duality.
Step 4: Let us prove that M is increasing. Suppose Λ and ∆ are two
events in L with Λ ⊂ ∆. The argument breaks down into three cases. Case 1:
Assume
Λ1 ×Λ2 ×···×Λn ⊂Λ 1≤k≤n
Then
sup min Mk {∆k } ≥ sup min Mk {Λk } > 0.5.

∆1 ×∆2 ×···×∆n ⊂∆ 1≤k≤n Λ1 ×Λ2 ×···×Λn ⊂Λ 1≤k≤n
It follows from (4.16) that M{Λ} ≤ M{∆}. Case 2: Assume

sup min Mk {∆k } > 0.5.
∆1 ×∆2 ×···×∆n ⊂∆c 1≤k≤n
Then
sup min Mk {Λk } ≥ sup min Mk {∆k } > 0.5.

Λ1 ×Λ2 ×···×Λn ⊂Λc 1≤k≤n ∆1 ×∆2 ×···×∆n ⊂∆c 1≤k≤n
Thus
M{Λ} = 1 − sup min Mk {Λk }
Λ1 ×Λ2 ×···×Λn ⊂Λc 1≤k≤n
≤1− sup min Mk {∆k } = M{∆}.

∆1 ×∆2 ×···×∆n ⊂∆c 1≤k≤n
Section 4.2 - Uncertain Variables 161
Case 3: Assume
sup min Mk {Λk } ≤ 0.5
Λ1 ×Λ2 ×···×Λn ⊂Λ 1≤k≤n
and
sup min Mk {∆k } ≤ 0.5.
∆1 ×∆2 ×···×∆n ⊂∆c 1≤k≤n
Then
M{Λ} ≤ 0.5 ≤ 1 − M{∆c } = M{∆}.
Step 5: Finally, we prove the countable subadditivity of M. For sim-
plicity, we only prove the case of two events Λ and ∆. The argument breaks
down into three cases. Case 1: Assume M{Λ} < 0.5 and M{∆} < 0.5. For
any given ε > 0, there are two rectangles Λ1 × Λ2 × · · · × Λn ⊂ Λc and
∆1 × ∆2 × · · · × ∆n ⊂ ∆c such that
1 − min Mk {Λk } ≤ M{Λ} + ε/2,
1≤k≤n
1 − min Mk {∆k } ≤ M{∆} + ε/2.

1≤k≤n
Note that
(Λ1 ∩ ∆1 ) × (Λ2 ∩ ∆2 ) × · · · × (Λn ∩ ∆n ) ⊂ (Λ ∪ ∆)c .
It follows from Theorem 4.3 that Mk {Λk ∩ ∆k } ≥ Mk {Λk } + Mk {∆k } − 1
for any k. Thus
M{Λ ∪ ∆} ≤ 1 − 1≤k≤n
min Mk {Λk ∩ ∆k }
≤ 1 − min Mk {Λk } + 1 − min Mk {∆k }

1≤k≤n 1≤k≤n
≤ M{Λ} + M{∆} + ε.
Letting ε → 0, we obtain M{Λ ∪ ∆} ≤ M{Λ} + M{∆}. Case 2: Assume
M{Λ} ≥ 0.5 and M{∆} < 0.5. When M{Λ ∪ ∆} = 0.5, the subadditivity is
obvious. Now we consider the case M{Λ ∪ ∆} > 0.5, i.e., M{Λc ∩ ∆c } < 0.5.
By using Λc ∪ ∆ = (Λc ∩ ∆c ) ∪ ∆ and Case 1, we get
M{Λc ∪ ∆} ≤ M{Λc ∩ ∆c } + M{∆}.
Thus
M{Λ ∪ ∆} = 1 − M{Λc ∩ ∆c } ≤ 1 − M{Λc ∪ ∆} + M{∆}
≤ 1 − M{Λc } + M{∆} = M{Λ} + M{∆}.
Case 3: If both M{Λ} ≥ 0.5 and M{∆} ≥ 0.5, then the subadditivity is
obvious because M{Λ} + M{∆} ≥ 1. The theorem is proved.
Definition 4.3 Let (Γk , Lk , Mk ), k = 1, 2, · · · , n be uncertainty spaces, Γ =
Γ1 × Γ2 × · · · × Γn , L = L1 × L2 × · · · × Ln , and M = M1 ∧ M2 ∧ · · · ∧ Mn .
Then (Γ, L, M) is called the product uncertainty space of (Γk , Lk , Mk ), k =
1, 2, · · · , n.
4.2 Uncertain Variables

Definition 4.4 (Liu [138]) An uncertain variable is a measurable function
ξ from an uncertainty space (Γ, L, M) to the set of real numbers, i.e., for any
Borel set B of real numbers, the set

{ξ ∈ B} = {γ ∈ Γ ξ(γ) ∈ B} (4.17)
is an event.
Definition 4.5 An uncertain variable ξ on the uncertainty space (Γ, L, M)

is said to be
(a) nonnegative if M{ξ < 0} = 0;
(b) positive if M{ξ ≤ 0} = 0;
(c) continuous if M{ξ = x} is a continuous function of x;
M {ξ 6= x1 , ξ 6= x2 , · · · , ξ 6= xm } = 0; (4.18)
M {ξ 6= x1 , ξ 6= x2 , · · · } = 0. (4.19)
Definition 4.6 Let ξ1 and ξ2 be uncertain variables defined on the uncer-

tainty space (Γ, L, M). We say ξ1 = ξ2 if ξ1 (γ) = ξ2 (γ) for almost all γ ∈ Γ.
Uncertain Vector
Definition 4.7 An n-dimensional uncertain vector is a measurable function
from an uncertainty space (Γ, L, M) to the set of n-dimensional real vectors,
i.e., for any Borel set B of <n , the set

{ξ ∈ B} = γ ∈ Γ ξ(γ) ∈ B (4.20)
is an event.
Theorem 4.9 The vector (ξ1 , ξ2 , · · · , ξn ) is an uncertain vector if and only

if ξ1 , ξ2 , · · · , ξn are uncertain variables.
Proof: Write ξ = (ξ1 , ξ2 , · · · , ξn ). Suppose that ξ is an uncertain vector on

the uncertainty space (Γ, L, M). For any Borel set B of <, the set B × <n−1
is a Borel set of <n . Thus the set
{ξ1 ∈ B} = {ξ1 ∈ B, ξ2 ∈ <, · · · , ξn ∈ <} = {ξ ∈ B × <n−1 }
is an event. Hence ξ1 is an uncertain variable. A similar process may

prove that ξ2 , ξ3 , · · · , ξn are uncertain variables. Conversely, suppose that
Section 4.2 - Uncertain Variables 163
all ξ1 , ξ2 , · · · , ξn are uncertain variables on the uncertainty space (Γ, L, M).

We define
B = B ⊂ <n {ξ ∈ B} is an event .

The vector ξ = (ξ1 , ξ2 , · · · , ξn ) is proved to be an uncertain vector if we can

( n
) n
Y \
ξ∈ (ai , bi ) = {ξi ∈ (ai , bi )}
i=1 i=1
is an event. Next, the class B is a σ-algebra of <n because (i) we have <n ∈ B
since {ξ ∈ <n } = Γ; (ii) if B ∈ B, then {ξ ∈ B} is an event, and
{ξ ∈ B c } = {ξ ∈ B}c
is an event. This means that B c ∈ B; (iii) if Bi ∈ B for i = 1, 2, · · · , then

{ξ ∈ Bi } are events and
∞ ∞
( )
[ [
ξ∈ Bi = {ξ ∈ Bi }
i=1 i=1
is an event. This means that ∪i Bi ∈ B. Since the smallest σ-algebra con-

taining all open intervals of <n is just the Borel algebra of <n , the class B
contains all Borel sets of <n . The theorem is proved.
Uncertain Arithmetic
Definition 4.8 Suppose that f : <n → < is a measurable function, and
ξ1 , ξ2 , · · · , ξn uncertain variables on the uncertainty space (Γ, L, M). Then
ξ = f (ξ1 , ξ2 , · · · , ξn ) is an uncertain variable defined as
ξ(γ) = f (ξ1 (γ), ξ2 (γ), · · · , ξn (γ)), ∀γ ∈ Γ. (4.21)
Example 4.7: Let ξ1 and ξ2 be two uncertain variables. Then the sum
ξ = ξ1 + ξ2 is an uncertain variable defined by
ξ(γ) = ξ1 (γ) + ξ2 (γ), ∀γ ∈ Γ.
The product ξ = ξ1 ξ2 is also an uncertain variable defined by
ξ(γ) = ξ1 (γ) · ξ2 (γ), ∀γ ∈ Γ.
The reader may wonder whether ξ(γ1 , γ2 , · · · , γn ) defined by (4.21) is an

uncertain variable. The following theorem answers this question.
Theorem 4.10 Let ξ be an n-dimensional uncertain vector, and f : <n → <

a measurable function. Then f (ξ) is an uncertain variable such that
M{f (ξ) ∈ B} = M{ξ ∈ f −1 (B)} (4.22)
for any Borel set B of real numbers.
Proof: Assume that ξ is an uncertain vector on the uncertainty space
(Γ, L, M). For any Borel set B of <, since f is a measurable function, the
f −1 (B) is a Borel set of <n . Thus the set {f (ξ) ∈ B} = {ξ ∈ f −1 (B)} is an
event for any Borel set B. Hence f (ξ) is an uncertain variable.
Example 4.8: Let ξ be a simple uncertain variable defined by



 x1 with uncertain measure µ1
x2 with uncertain measure µ2

ξ=

 ···
xm with uncertain measure µm

where x1 , x2 , · · · , xm are distinct numbers. If f is a one-to-one function, then

f (ξ) is also a simple uncertain variable


 f (x1 ) with uncertain measure µ1
f (x2 ) with uncertain measure µ2

f (ξ) =

 ···
f (xm ) with uncertain measure µm .

4.3 Identification Function

A random variable may be characterized by a probability density function,
and a fuzzy variable may be described by a membership function. This
section will introduce an identification function to characterize an uncertain
variable.
First Identification Function

Definition 4.9 An uncertain variable ξ is said to have a first identification
function λ if
(i) λ(x) is a nonnegative function on < such that
sup (λ(x) + λ(y)) = 1; (4.23)
x6=y
(ii) for any set B of real numbers, we have


M{ξ ∈ B} = 
x∈B x∈B

(4.24)
 1 − sup λ(x), if sup λ(x) ≥ 0.5.
x∈B c x∈B
Section 4.3 - Identification Function 165
Example 4.9: By a rectangular uncertain variable we mean the uncertain

variable fully determined by the pair (a, b) of crisp numbers with a < b, whose
first identification function is
(
0.5, if a ≤ x ≤ b
λ1 (x) =
0, otherwise.
Example 4.10: By a triangular uncertain variable we mean the uncertain

variable fully determined by the triplet (a, b, c) of crisp numbers with a <
b < c, whose first identification function is
 x−a
 , if a ≤ x ≤ b
 2(b − a)



λ2 (x) = x−c
, if b ≤ x ≤ c



 2(b − c)

0, otherwise.
Example 4.11: By a trapezoidal uncertain variable we mean the uncertain

variable fully determined by the quadruplet (a, b, c, d) of crisp numbers with
a < b < c < d, whose first identification function is
x−a


 , if a ≤ x ≤ b



 2(b − a)


 0.5, if b ≤ x ≤ c
λ3 (x) =
x−d
, if c ≤ x ≤ d






 2(c − d)

 0, otherwise.
λ1 (x) λ2 (x) λ3 (x)

.... .... ....
......... ......... .........
.... . . . . ......................................................... . . . . . . . . . . . ...... . . . . . . . . . . ...... . . . . . . . . . . . . . . . . . . .... . . . . . . . . ........................................
0.5 ... . . .
. .....
. ... .. ...
... .
.
.
. .... ... .. ... ... ..... ....
..
... . . ... ... . ... ... ... . . ....
... . . .
. ... . .... .
. .. .
. . ..
... . . .
.
. .
.. . ... .
.
. .. .
. . ...
. . .. .. .... .. .. . .
... . . .... . ..
. . . ....
... . . .
. ..
. . ... .
.
. .. .
. . ...
. . . . . ... . . . . ..
... . . .
. .
. . ... .
. .
.
. .. . .
... . . .... ... . .
...
.
.
.
. ..
. . . ....
... . . .
. ..
. . ... .
. .
. . ...
... . . .
.
. ..
. . . .
.
. .
. . . ...
. . . ... . . . ...
... . . .... ... . ...
.
.
.
. .
.
.
. . . ...
... . . .
. ..
. . .
... . .
. .. . . ..
.................................................................................................... . . . . . . .
x .
. .
.
x ... .............................................................................................................................
.
. .
.
x
0 a ...
a c
.
......................................................................................................
.. ..
a b c d
...
..
b ...
..
b ...
..
Figure 4.1: First Identification Functions λ1 , λ2 and λ3
Theorem 4.11 Suppose λ is a nonnegative function satisfying (4.23). Then

there is an uncertain variable ξ such that (4.24) holds.
Proof: Let < be the universal set. For each set B of real numbers, we define
a set function

M{B} = 
x∈B x∈B

 1 − sup λ(x), if sup λ(x) ≥ 0.5.

x∈B c x∈B
It is clear that M is normal, increasing, self-dual, and countably subadditive.

That is, the set function M is indeed an uncertain measure. Now we define
an uncertain variable ξ as an identity function from the uncertainty space
(<, A, M) to <. We may verify that ξ meets (4.24). The theorem is proved.
Theorem 4.12 (First Measure Inversion Theorem) Let ξ be an uncertain

variable with first identification function λ. Then for any set B of real num-
bers, we have

M{ξ ∈ B} = 
x∈B x∈B

(4.25)
 1 − sup λ(x), if sup λ(x) ≥ 0.5.
x∈B c x∈B
Proof: It follows from the definition of first identification function immedi-

ately.
Example 4.12: Suppose ξ is an uncertain variable with first identification

function λ. Then
M{ξ = x} = λ(x), ∀x ∈ <.
Example 4.13: Suppose ξ is a discrete uncertain variable taking values in
{x1 , x2 , · · · } with first identification function λ. Note that λ must meet
sup (λ(xi ) + λ(xj )) = 1. (4.26)

i6=j
For any set B of real numbers, we have


 sup λ(xi ), if sup λ(xi ) < 0.5
M{ξ ∈ B} = 
 xi ∈B xi ∈B
(4.27)
 1 − sup λ(xi ), if sup λ(xi ) ≥ 0.5.
xi ∈B c xi ∈B
Second Identification Function

Definition 4.10 (Liu [147]) An uncertain variable ξ is said to have a second
identification function ρ if
(i) ρ(x) is a nonnegative and integrable function on < such that
Z
ρ(x)dx ≥ 1; (4.28)
<
(ii) for any Borel set B of real numbers, we have

 Z Z




 B B
M{ξ ∈ B} =  1 − ρ(x)dx,
 Z Z
(4.29)
if ρ(x)dx < 0.5
Bc Bc






0.5, otherwise.
Example 4.14: By an exponential uncertain variable we mean the uncertain

variable whose second identification function is

 1 exp − x , if x ≥ 0


ρ1 (x) = β α (4.30)

 0, if x < 0
where α and β are real numbers with α ≥ β > 0. Note that

Z +∞
α
ρ1 (x)dx = ≥ 1.
−∞ β
Example 4.15: By a bell-like uncertain variable we mean the uncertain

variable whose second identification function is
(x − µ)2

1
ρ2 (x) = √ exp − , x∈< (4.31)
β π α2
where µ, α and β are real numbers with α ≥ β > 0. Note that

Z +∞
α
ρ2 (x)dx = ≥ 1.
−∞ β
ρ1 (x) ρ2 (x)
.... ....
......... .........
........ .... ...........
.... .. ....
... .... .. .... .. .....
... .... ... ... .. ...
... .... ... .... .. ...
...
... ... ... ... .. ...
... ... ... ... .. ...
... ... ... ... . ...
....
... .... ... ... ... ...
... ..... ... ... .. ...
..... .. ...
... ..... ... .. .. ...
... ..... ... .. . ...
...
......
....... ... .
... .
. ...
... ........
.......... ... ....... ... .....
.....
... ............. ... ....... .. ......
... ................... .. . . ...
. . ......
........................................................................................................................................................... x ............................................................................................................................................................... x
.... .... µ
. .
Figure 4.2: Second Identification Functions ρ1 and ρ2

Theorem 4.13 Suppose ρ is a nonnegative and integrable function satisfying

(4.28). Then there is an uncertain variable ξ such that (4.29) holds.
Proof: Let < be the universal set. For each Borel set B of real numbers, we
define a set function
 Z Z




 B B
M{B} =  1 − ρ(x)dx, if ρ(x)dx < 0.5

 Z Z
Bc Bc






0.5, otherwise.

Theorem 4.14 (Second Measure Inversion Theorem) Let ξ be an uncertain

variable with second identification function ρ. Then for any Borel set B of
real numbers, we have
 Z Z




 B B
M{ξ ∈ B} =  1 − ρ(x)dx, if ρ(x)dx < 0.5

 Z Z
(4.32)
c c

 B B




0.5, otherwise.
Proof: It follows from the definition of second identification function imme-

diately.

{x1 , x2 , · · · } with second identification function ρ. Note that ρ must meet
∞
X
ρ(xi ) ≥ 1. (4.33)
i=1
For any set B of real numbers, we have

 X X

 ρ(xi ), if ρ(xi ) < 0.5


 xi ∈B xi ∈B
M{ξ ∈ B} =  1 − X ρ(xi ), if X ρ(xi ) < 0.5


(4.34)
xi ∈B c xi ∈B c





0.5, otherwise.
Third Identification Function
Definition 4.11 (Liu [141]) An uncertain variable ξ is said to have a third

identification function (λ, ρ) if (i) λ(x) is a nonnegative function and ρ(x) is
a nonnegative and integrable function such that
Z Z
sup λ(x) + ρ(x)dx ≥ 0.5 and/or sup λ(x) + ρ(x)dx ≥ 0.5 (4.35)
x∈B B x∈B c Bc
for any Borel set B of real numbers; (ii) we have

 Z


 sup λ(x) + ρ(x)dx,
x∈B

 B



 Z

if sup λ(x) + ρ(x)dx < 0.5






 x∈B B
M{ξ ∈ B} = 
 Z
(4.36)
1 − sup λ(x) − ρ(x)dx,
x∈B c Bc





 Z






 x∈B c Bc



0.5, otherwise.

Theorem 4.15 Suppose λ is a nonnegative function and ρ is a nonnegative

and integrable function satisfying the condition (4.35). Then there is an
uncertain variable ξ such that (4.36) holds.
Proof: Let < be the universal set. For each Borel set B of real numbers, we
define a set function
 Z Z


 sup λ(x) + ρ(x)dx, if sup λ(x) + ρ(x)dx < 0.5


 x∈B B x∈B B
M{B} =  1 − sup λ(x) − ρ(x)dx, if sup λ(x) + ρ(x)dx < 0.5

 Z Z
x∈B c Bc x∈B c Bc






0.5, otherwise.

Theorem 4.16 (Third Measure Inversion Theorem) Let ξ be an uncertain

variable with third identification function (λ, ρ). Then for any Borel set B of
real numbers, we have

 Z


 sup λ(x) + ρ(x)dx,
x∈B

 B



 Z







 x∈B B
M{ξ ∈ B} =  1 − sup λ(x) − ρ(x)dx,

 Z
x∈B c Bc





 Z






 x∈B c Bc



0.5, otherwise.

Proof: It follows from the definition of third identification function imme-

diately.

{x1 , x2 , · · · } with third identification function (λ, ρ). Assume that
X X
sup λ(xi ) + ρ(xi ) ≥ 0.5 and/or sup λ(xi ) + ρ(xi ) ≥ 0.5
xi ∈B xi ∈B c
xi ∈B xi ∈B c
for any set B of real numbers. Then we have

 X

 sup λ(xi ) + ρ(xi ),
 xi ∈B

 xi ∈B


 X
if sup λ(xi ) + ρ(xi ) < 0.5




 xi ∈B

 xi ∈B
M{ξ ∈ B} =  1 − sup λ(xi ) − X ρ(xi ),


xi ∈B c
xi ∈B c




 X
if sup λ(xi ) + ρ(xi ) < 0.5




 xi ∈B c c


 x i ∈B

0.5, otherwise.

4.4 Uncertainty Distribution

Following the idea of probability distribution, credibility distribution and
chance distribution, this section introduces uncertainty distribution in order
to describe uncertain variables.
Definition 4.12 (Liu [138]) The uncertainty distribution Φ: < → [0, 1] of

an uncertain variable ξ is defined by
Φ(x) = M {ξ ≤ x} . (4.37)
Section 4.4 - Uncertainty Distribution 171
Example 4.18: Let ξ be an uncertain variable with first identification func-

tion λ. Then its uncertainty distribution is

 sup λ(y), if sup λ(y) < 0.5
 y≤x y≤x
Φ(x) =
 1 − sup λ(y), if sup λ(y) ≥ 0.5.

y>x y≤x
Example 4.19: Let ξ be an uncertain variable with second identification

function ρ. Then its uncertainty distribution is
 Z x Z x
ρ(y)dy, if ρ(y)dy < 0.5






 −∞ −∞
 Z +∞ Z +∞
Φ(x) =

 1 − ρ(y)dy, if ρ(y)dy < 0.5


 x x


0.5, otherwise.

Example 4.20: Let ξ be an uncertain variable with third identification

function (λ, ρ). Then its uncertainty distribution is
 Z x


 sup λ(y) + ρ(y)dy,

 y≤x −∞



 Z x



 if sup λ(y) + ρ(y)dy < 0.5
y≤x

 −∞


 Z +∞
Φ(x) =
 1 − sup λ(y) − ρ(y)dy,
y>x


 x



 Z +∞

if sup λ(y) + ρ(y)dy < 0.5



y>x



 x


0.5, otherwise.
Theorem 4.17 An uncertainty distribution is an increasing function such

that
0 ≤ lim Φ(x) < 1, 0 < lim Φ(x) ≤ 1. (4.38)
x→−∞ x→+∞
Proof: It is obvious that an uncertainty distribution Φ is an increasing

function, and the inequalities (4.38) follow from the uncertainty asymptotic
theorem immediately.
Example 4.21: Let c be a number with 0 < c < 1. Then Φ(x) ≡ c is an

uncertainty distribution. When c ≤ 0.5, we define a set function over < as
follows,


 0, if Λ=∅

c, if Λ is upper bounded



M{Λ} = 

0.5, if both Λ and Λc are upper unbounded
1 − c, if Λc is upper bounded





1, if Λ = Γ.

Then (<, L, M) is an uncertainty space. It is easy to verify that the identity

function ξ(γ) = γ is an uncertain variable whose uncertainty distribution is
just Φ(x) ≡ c. When c > 0.5, we define


 0, if Λ = ∅

 1 − c, if Λ is upper bounded



M{Λ} =  0.5, if both Λ and Λc are upper unbounded
c, if Λc is upper bounded





1, if Λ = Γ.

Then the function ξ(γ) = −γ is an uncertain variable whose uncertainty

distribution is just Φ(x) ≡ c.
Definition 4.13 A continuous uncertain variable is said to be (a) singular

if its uncertainty distribution is a singular function; (b) absolutely continuous
if its uncertainty distribution is absolutely continuous.
Definition 4.14 (Liu [138]) The uncertainty density function φ: < → [0, +∞)
of an uncertain variable ξ is a function such that
Z x
Φ(x) = φ(y)dy, ∀x ∈ <, (4.39)
−∞
Z +∞
φ(y)dy = 1 (4.40)
−∞
where Φ is the uncertainty distribution of ξ.
Example 4.22: The uncertainty density function may not exist even if the
uncertainty distribution is continuous and differentiable a.e. Suppose f is the
Cantor function, and set

 0,
 if x < 0
Φ(x) = f (x), if 0 ≤ x ≤ 1 (4.41)

1, if x > 1.

Then Φ is an increasing and continuous function, and is an uncertainty dis-

tribution. Note that Φ0 (x) = 0 almost everywhere, and
Z +∞
Φ0 (x)dx = 0 6= 1.
−∞
Thus the uncertainty density function does not exist.

Theorem 4.18 Let ξ be an uncertain variable whose uncertainty density
function φ exists. Then we have
Z x Z +∞
M{ξ ≤ x} = φ(y)dy, M{ξ ≥ x} = φ(y)dy. (4.42)
−∞ x
by the self-duality of uncertain measure, we have
Z +∞ Z x Z +∞
M{ξ ≥ x} = 1 − M{ξ < x} = φ(y)dy − φ(y)dy = φ(y)dy.
−∞ −∞ x
Joint Uncertainty Distribution

Definition 4.15 Let (ξ1 , ξ2 , · · · , ξn ) be an uncertain vector. Then the joint
uncertainty distribution Φ : <n → [0, 1] is defined by
Φ(x1 , x2 , · · · , xn ) = M {ξ1 ≤ x1 , ξ2 ≤ x2 , · · · , ξn ≤ xn } .
Definition 4.16 The joint uncertainty density function φ : <n → [0, +∞)
of an uncertain vector (ξ1 , ξ2 , · · · , ξn ) is a function such that
Z x1 Z x2 Z xn
−∞ −∞ −∞
n
holds for all (x1 , x2 , · · · , xn ) ∈ < , and
Z +∞ Z +∞ Z +∞
··· φ(y1 , y2 , · · · , yn )dy1 dy2 · · · dyn = 1
−∞ −∞ −∞
where Φ is the joint uncertainty distribution of (ξ1 , ξ2 , · · · , ξn ).
4.5 Independence
Definition 4.17 (Liu [141]) The uncertain variables ξ1 , ξ2 , · · · , ξm are said
to be independent if
(m )
M {ξi ∈ Bi } = min M {ξi ∈ Bi }
\
(4.43)
1≤i≤m
i=1
for any Borel sets B1 , B2 , · · · , Bm of real numbers.

Theorem 4.18 The uncertain variables ξ1 , ξ2 , · · · , ξm are independent if and

only if (m )
M {ξi ∈ Bi } = max M {ξi ∈ Bi }
[
(4.44)
1≤i≤m
i=1
for any Borel sets B1 , B2 , · · · , Bm of real numbers.
Proof: It follows from the self-duality of uncertain measure that ξ1 , ξ2 , · · · , ξm
are independent if and only if
(m ) (m )
M {ξi ∈ Bi } = 1 − M
[ \
{ξi ∈ Bic }
i=1 i=1
= 1 − min
1≤i≤m
M{ξi ∈ Bic } = max
1≤i≤m
M {ξi ∈ Bi } .
Thus the proof is complete.

This section introduces the concept of identical distribution of uncertain vari-
ables.
Definition 4.19 The uncertain variables ξ and η are identically distributed
if
M{ξ ∈ B} = M{η ∈ B} (4.45)
for any Borel set B of real numbers.
Theorem 4.19 Let ξ and η be identically distributed uncertain variables,
and f : < → < a measurable function. Then f (ξ) and f (η) are identically
distributed uncertain variables.
Proof: For any Borel set B of real numbers, we have
M{f (ξ) ∈ B} = M{ξ ∈ f −1 (B)} = M{η ∈ f −1 (B)} = M{f (η) ∈ B}.
Hence f (ξ) and f (η) are identically distributed uncertain variables.
Theorem 4.20 If ξ and η are identically distributed uncertain variables,
then they have the same uncertainty distribution.
Proof: Since ξ and η are identically distributed uncertain variables, we have
M{ξ ∈ (−∞, x]} = M{η ∈ (−∞, x]} for any x. Thus ξ and η have the same
uncertainty distribution.
Theorem 4.21 If ξ and η are identically distributed uncertain variables
whose uncertainty density functions exist, then they have the same uncer-
tainty density function.
Proof: It follows from Theorem 4.20 immediately.
Section 4.7 - Operational Law 175
4.7 Operational Law

Theorem 4.22 (Liu [141], Operational Law) Let ξ1 , ξ2 , · · · , ξn be indepen-
dent uncertain variables, and f : <n → < a measurable function. Then
ξ = f (ξ1 , ξ2 , · · · , ξn ) is an uncertain variable such that
min Mk {ξk ∈ Bk },

 sup
f (B1 ,B2 ,··· ,Bn )⊂B 1≤k≤n



min Mk {ξk ∈ Bk } > 0.5


if sup



f (B1 ,B2 ,··· ,Bn )⊂B 1≤k≤n




M{ξ ∈ B} =  1 − min Mk {ξk ∈ Bk },

sup
 f (B1 ,B2 ,··· ,Bn )⊂B c 1≤k≤n
min Mk {ξk ∈ Bk } > 0.5






 if sup
f (B1 ,B2 ,··· ,Bn )⊂B c 1≤k≤n






0.5, otherwise
where B, B1 , B2 , · · · , Bn are Borel sets of real numbers.
Proof: It follows from the product measure axiom immediately.
Example 4.23: The sum of independent rectangular uncertain variables

ξ = (a1 , a2 ) and η = (b1 , b2 ) is a rectangular uncertain variable, and
ξ + η = (a1 + b1 , a2 + b2 ).
The product of a rectangular uncertain variable ξ = (a1 , a2 ) and a scalar

number h is also a rectangular uncertain variable
(
(ha1 , ha2 ), if h ≥ 0
h·ξ =
(ha2 , ha1 ), if h < 0.
Example 4.24: The sum of independent triangular uncertain variables ξ =

(a1 , a2 , a3 ) and η = (b1 , b2 , b3 ) is a triangular uncertain variable, and
ξ + η = (a1 + b1 , a2 + b2 , a3 + b3 ).
The product of a triangular uncertain variable ξ = (a1 , a2 , a3 ) and a scalar

number h is also a triangular uncertain variable
(
(ha1 , ha2 , ha3 ), if h ≥ 0
h·ξ =
(ha3 , ha2 , ha1 ), if h < 0.
Example 4.25: The sum of independent trapezoidal uncertain variables

ξ = (a1 , a2 , a3 , a4 ) and η = (b1 , b2 , b3 , b4 ) is a trapezoidal uncertain variable,
and
ξ + η = (a1 + b1 , a2 + b2 , a3 + b3 , a4 + b4 ).
The product of a trapezoidal uncertain variable ξ = (a1 , a2 , a3 , a4 ) and a

scalar number h is also a trapezoidal uncertain variable
(
(ha1 , ha2 , ha3 , ha4 ), if h ≥ 0
h·ξ =
(ha4 , ha3 , ha2 , ha1 ), if h < 0.
Example 4.26: Let ξ1 , ξ2 , · · · , ξn be independent uncertain variables with

first identification functions λ1 , λ2 , · · · , λn , respectively, and f : <n → < a
function. It follows from the operational law that ξ = f (ξ1 , ξ2 , · · · , ξn ) is an
uncertain variable with first identification function
sup min λ (x ), if sup min λi (xi ) < 0.5

 f (x1 ,··· ,xn )=x 1≤i≤n i i

f (x1 ,··· ,xn )=x 1≤i≤n
λ(x) =
 1−
 sup min λi (xi ), if sup min λi (xi ) ≥ 0.5.
f (x1 ,··· ,xn )6=x 1≤i≤n f (x1 ,··· ,xn )=x 1≤i≤n
Example 4.27: Let ξ and η be two independent exponential uncertain

variables whose second identification function is
 
 1 exp − x , if x ≥ 0  1 exp − x , if x ≥ 0

 
ρξ (x) = a a ρη (x) = b b
 
 0, if x < 0,  0, if x < 0.
Then their sum is also an exponential uncertain variable with second identi-
fication function

 1 exp − x

, if x ≥ 0
ρ(x) = a+b a+b

0, if x < 0.

4.8 Expected Value

Expected value is the average value of uncertain variable in the sense of
uncertain measure, and represents the size of uncertain variable.
Definition 4.20 (Liu [138]) Let ξ be an uncertain variable. Then the ex-
pected value of ξ is defined by
Z +∞ Z 0
E[ξ] = M{ξ ≥ r}dr − M{ξ ≤ r}dr (4.46)
0 −∞
Example 4.28: The rectangular uncertain variable ξ = (a, b) has an ex-

pected value (a + b)/2.
Example 4.29: The triangular uncertain variable ξ = (a, b, c) has an ex-

pected value E[ξ] = (a + 2b + c)/4.
Example 4.30: The trapezoidal uncertain variable ξ = (a, b, c, d) has an

expected value E[ξ] = (a + b + c + d)/4.
Example 4.31: Let ξ be a continuous uncertain variable with first iden-

tification function λ. If its expected value exists, and there is a point x0
such that λ(x) is increasing on (−∞, x0 ) and decreasing on (x0 , +∞), then
M{ξ ≥ x} = λ(x) for any x > x0 and M{ξ ≤ x} = λ(x) for any x < x0 .
Thus
Z +∞ Z x0
E[ξ] = x0 + λ(x)dx − λ(x)dx.
x0 −∞
Example 4.32: The definition of expected value operator is also applicable

to discrete case. Assume that ξ is a simple uncertain variable whose first
identification function is


 λ1 , if x = x1
λ2 , if x = x2

λ(x) = (4.47)

 ···
λm , if x = xm

where x1 , x2 , · · · , xm are distinct numbers. The expected value of ξ is
m
X
E[ξ] = wi xi (4.48)
i=1
max {λj |xj ≤ xi } − max {λj |xj < xi }


1≤j≤m 1≤j≤m





wi = + max {λj |xj ≥ xi } − max {λj |xj > xi }, if λi < 0.5
 1≤j≤m 1≤j≤m


 1 − max {λj |xj < xi } − max {λj |xj > xi },

if λi ≥ 0.5
1≤j≤m 1≤j≤m
for i = 1, 2, · · · , m. It is easy to verify that all wi ≥ 0 and the sum of all

weights is just 1.
Example 4.33: Consider the uncertain variable ξ defined by (4.47). Suppose

x1 < x2 < · · · < xm and there exists an index k with 1 < k < m such that
λ 1 ≤ λ2 ≤ · · · ≤ λ k and λk ≥ λk+1 ≥ · · · ≥ λm .
Then the expected value is determined by (4.48) and the weights are given
by 

 λ1 , if i = 1

λ − λ , if i = 2, 3, · · · , k − 1

i i−1



wi = 1 − λk−1 − λk+1 , if i = k

λi − λi+1 , if i = k + 1, k + 2, · · · , m − 1





λm , if i = m.

Example 4.34: The exponential uncertain variable ξ with second identifi-

cation function 
 1 exp − x , if x ≥ 0


ρ(x) = β α

 0, if x < 0
has an expected value
α(β − α) 2α α 2α − β
E[ξ] = α + ln + ln .
β 2α − β 2 β
Example 4.35: The bell-like uncertain variable ξ with second identification

function
(x − µ)2

1
ρ(x) = √ exp −
β π α2
has an expected value E[ξ] = µ.
Example 4.36: Let ξ be a continuous uncertain variable with second iden-

tification function ρ. It is clear that there are two points x1 and x2 such
that Z x1 Z +∞
ρ(x)dx = ρ(x)dx = 0.5.
−∞ x2
If the expected value exists, then we have

Z x1 Z +∞
E[ξ] = xρ(x)dx + xρ(x)dx.
−∞ x2
Example 4.37: Assume that ξ is a simple uncertain variable whose second



 ρ1 , if x = x1
ρ2 , if x = x2

ρ(x) = (4.49)

 ···
ρm , if x = xm


m
X
E[ξ] = wi xi (4.50)
i=1

 X X

 ρi , if ρj < 0.5 or ρj < 0.5

 xj ≤xi xj ≥xi
wi = X + X +



 0.5 − ρ i + 0.5 − ρi , otherwise
xj <xi xj >xi
for i = 1, 2, · · · , m.
Example 4.38: Assume that ξ is a simple uncertain variable whose third



 (λ1 , ρ1 ), if x = x1
(λ1 , ρ2 ), if x = x2

(λ, ρ)(x) = (4.51)

 ···
(λm , ρm ), if x = xm


m
X
E[ξ] = wi xi (4.52)
i=1

 X

 sup λj − sup λj + ρi , if sup λj + ρi < 0.5
xj ≤xi xj <xi xj ≤xi
x ≤x


 j i

 X
sup λj − sup λj + ρi , if sup λj + ρi < 0.5

wi = xj ≥xi xj >xi xj ≥xi

 xj ≥x i

 X + X +
− − − −



 0.5 sup λ j ρ i + 0.5 sup λ j ρi , otherwise
 xj <xi xj >xi
xj <xi xj >xi
for i = 1, 2, · · · , m.
Theorem 4.23 Let ξ be an uncertain variable whose uncertainty density

function φ exists. If the Lebesgue integral
Z +∞
xφ(x)dx
−∞

E[ξ] = xφ(x)dx. (4.53)
−∞
Theorem that
Z +∞ Z 0
E[ξ] = M{ξ ≥ r}dr − M{ξ ≤ r}dr
0 −∞
Z +∞ Z +∞ Z 0 Z r
0 r −∞ −∞
Z +∞ Z x Z 0 Z 0
0 0 −∞ x
Z +∞ Z 0
0 −∞
Z +∞
= xφ(x)dx.
−∞
Theorem 4.24 Let ξ be an uncertain variable with uncertainty distribution

Φ. If
lim Φ(x) = 0, lim Φ(x) = 1
x→−∞ x→∞

Z +∞
xdΦ(x)
−∞

Z +∞
E[ξ] = xdΦ(x). (4.54)
−∞
R +∞
Proof: Since the Lebesgue-Stieltjes integral −∞
xdΦ(x) is finite, we imme-
diately have
Z y Z +∞ Z 0 Z 0
y→+∞ 0 0 y→−∞ y −∞
and Z +∞ Z y
y→+∞ y y→−∞ −∞
It follows from
Z +∞
y z→+∞
Z y
−∞ z→−∞
that
lim y (1 − Φ(y)) = 0, lim yΦ(y) = 0.
y→+∞ y→−∞
n−1
X Z y
xi (Φ(xi+1 ) − Φ(xi )) → xdΦ(x)
i=0 0
and
n−1 y
M{ξ ≥ r}dr
X Z
(1 − Φ(xi+1 ))(xi+1 − xi ) →
i=0 0
n−1
X n−1
X
i=0 i=0

Z +∞ +∞
M{ξ ≥ r}dr =
Z
xdΦ(x).
0 0

Z 0 0
M{ξ ≤ r}dr =
Z
− xdΦ(x).
−∞ −∞
Theorem 4.25 Let ξ be an uncertain variable with finite expected value.

Then for any real numbers a and b, we have
E[aξ + b] = aE[ξ] + b. (4.55)
Z +∞ Z 0
E[ξ + b] = M{ξ + b ≥ r}dr − M{ξ + b ≤ r}dr
0 −∞
+∞ 0
M{ξ ≥ r − b}dr − M{ξ ≤ r − b}dr
Z Z
=
0 −∞
b
(M{ξ ≥ r − b} + M{ξ < r − b})dr
Z
= E[ξ] +
0
= E[ξ] + b.

Z 0
E[aξ + b] = E[ξ] − (M{ξ ≥ r − b} + M{ξ < r − b})dr = E[ξ] + b.
b
Step 2: We prove E[aξ] = aE[ξ]. If a = 0, then the equation E[aξ] =

aE[ξ] holds trivially. If a > 0, we have
Z +∞ Z 0
E[aξ] = M{aξ ≥ r}dr − M{aξ ≤ r}dr
0 −∞
+∞ 0
M{ξ ≥ r/a}dr − M{ξ ≤ r/a}dr
Z Z
=
0 −∞
+∞ Z 0
M{ξ ≥ t}dt − a M{ξ ≤ t}dt
Z
=a
0 −∞
= aE[ξ].
If a < 0, we have
+∞ 0
M{aξ ≥ r}dr − M{aξ ≤ r}dr
Z Z
E[aξ] =
0 −∞
+∞ Z 0
M{ξ ≤ r/a}dr − M{ξ ≥ r/a}dr
Z
=
0 −∞
+∞ Z 0
M{ξ ≥ t}dt − a M{ξ ≤ t}dt
Z
=a
0 −∞
= aE[ξ].
Step 3: Finally, for any real numbers a and b, it follows from Steps 1
and 2 that
E[aξ + b] = E[aξ] + b = aE[ξ] + b.
Theorem 4.26 Let f be a convex function on [a, b], and ξ an uncertain
variable that takes values in [a, b] and has expected value e. Then
b−e e−a
E[f (ξ)] ≤ f (a) + f (b). (4.56)
b−a b−a
Proof: For each γ ∈ Γ, we have a ≤ ξ(γ) ≤ b and
b − ξ(γ) ξ(γ) − a
ξ(γ) = a+ b.
b−a b−a
b − ξ(γ) ξ(γ) − a
f (ξ(γ)) ≤ f (a) + f (b).
b−a b−a
Taking expected values on both sides, we obtain the inequality.
4.9 Variance
The variance of an uncertain variable provides a measure of the spread of the
distribution around its expected value. A small value of variance indicates
that the uncertain variable is tightly concentrated around its expected value;
and a large value of variance indicates that the uncertain variable has a wide
spread around its expected value.
Definition 4.21 (Liu [138]) Let ξ be an uncertain variable with finite ex-
pected value e. Then the variance of ξ is defined by V [ξ] = E[(ξ − e)2 ].
Example 4.39: Let ξ be a rectangular uncertain variable (a, b). Then its
expected value is e = (a + b)/2, and for any positive number r, we have
(
1/2, if r ≤ (b − a)2 /4
M{(ξ − e) ≥ r} =
2
0, if r > (b − a)2 /4.
Thus the variance is
Z +∞ (b−a)2 /4
(b − a)2
M{(ξ − e) ≥ r}dr =
Z
2 1
V [ξ] = dr = .
0 0 2 8
Example 4.40: Let ξ = (a, b, c) be a triangular uncertain variable. Then

its variance is
33α3 + 21α2 β + 11αβ 2 − β 3
V [ξ] =
384α
where α = (b − a) ∨ (c − b) and β = (b − a) ∧ (c − b). Especially, if ξ is
symmetric, i.e., b − a = c − b, then its variance is V [ξ] = (c − a)2 /24.
Example 4.41: Let ξ = (a, b, c, d) be a symmetric trapezoidal uncertain

variable, i.e., b − a = d − c. Then its variance is
(d − a)2 + (d − a)(c − b) + (c − b)2
V [ξ] = .
24
Example 4.42: An uncertain variable ξ is called normal if it has a normal

uncertainty distribution
−1
π(e − x)
Φ(x) = 1 + exp √ , x ∈ <. (4.57)
3σ
Note that the uncertain variable with first identification function
 −1
π(e − x)
√


 1 + exp , if x ≤ e
3σ


λ(x) = −1

 π(x − e)
 1 + exp √ , if x > e


3σ
Φ(x)
....
.........
....
1 .........................................................................
..
.... .....................................
..............
... ..........
... ..
...
.........
.
... ......
.....
... ......
... .....
... ..
......
.
... .....
...........
0.5 .........................................................................
...
...
.
..... ...
... ..
...... .. ....
.
... ..... ... ...
..... ...
... ...... .. ...
... ....... .. ...
... ...
...
........ . ....
.......................
.
.
. .....
.........
................. .
...........................................................................................................................................................................................................
........................ . ...............
......................................... x
..
0 ..
..... e
.
Figure 4.3: Normal Uncertainty Distribution
has the normal uncertainty distribution Φ; and the uncertain variable with
second identification function
−1
πx π(e − x) π(x − e)
ρ(x) = √ 2 + exp √ + exp √
3σ 3σ 3σ
has also the normal uncertainty distribution Φ. It is easy to verify that
the normal uncertain variable ξ has expected value e. It follows from the
definition of variance that
Z +∞ Z +∞
V [ξ] = M{(ξ − e)2 ≥ r}dr = M{(ξ ≥ e + √r) ∪ (ξ ≤ e − √r)}dr.
0 0
On the one hand, we have

Z +∞ +∞ √ −1
σ2
M{ξ ≥ e + √r}dr =
Z
π r
V [ξ] ≥ 1 + exp √ dr = .
0 0 3σ 2
On the other hand, we have
Z +∞
√ √
V [ξ] ≤ (M{ξ ≥ e + r} + M{ξ ≤ e − r})dr
0
Z +∞ √ −1
π r
=2 1 + exp √ dr = σ 2 .
0 3σ
This means that the variance of a normal uncertain variable is an interval
[σ 2 /2, σ 2 ]. If a scalar variance is needed, then we take the maximum value,
i.e.,
V [ξ] = σ 2 . (4.58)
Let ξ1 and ξ2 be independent normal uncertain variables with expected values
e1 and e2 , variances σ12 and σ22 , respectively. Then for any real numbers a1
and a2 , the uncertain variable a1 ξ1 + a2 ξ2 is also normal with expected value
a1 e1 + a2 e2 and variance (|a1 |σ1 + |a2 |σ2 )2 .
Theorem 4.27 If ξ is an uncertain variable with finite expected value, a and

b are real numbers, then V [aξ + b] = a2 V [ξ].

Theorem 4.28 Let ξ be an uncertain variable with expected value e. Then

V [ξ] = 0 if and only if M{ξ = e} = 1.

+∞
M{(ξ − e)2 ≥ r}dr
Z
2
E[(ξ − e) ] =
0
which implies M{(ξ − e)2 ≥ r} = 0 for any r > 0. Hence we have

M{(ξ − e)2 = 0} = 1.
That is, M{ξ = e} = 1.
Conversely, if M{ξ = e} = 1, then we have M{(ξ − e)2 = 0} = 1 and
M{(ξ − e)2 ≥ r} = 0 for any r > 0. Thus
+∞
M{(ξ − e)2 ≥ r}dr = 0.
Z
V [ξ] =
0
Theorem 4.29 Let ξ be an uncertain variable that takes values in [a, b] and
has expected value e. Then
V [ξ] ≤ (e − a)(b − e). (4.59)
4.10 Moments
Definition 4.22 (Liu [138]) Let ξ be an uncertain variable. Then for any
positive integer k,
Theorem 4.30 Let ξ be a nonnegative uncertain variable, and k a positive

number. Then the k-th moment
Z +∞
E[ξ k ] = k rk−1 M{ξ ≥ r}dr. (4.60)
0

Z ∞ Z ∞ ∞
M{ξ k ≥ x}dx = M{ξ ≥ r}drk = k rk−1 M{ξ ≥ r}dr.
Z
E[ξ k ] =
0 0 0

Theorem 4.31 Let ξ be an uncertain variable, and t a positive number. If
E[|ξ|t ] < ∞, then
lim xt M{|ξ| ≥ x} = 0. (4.61)
x→∞
Conversely, if (4.61) holds for some positive number t, then E[|ξ|s ] < ∞ for
any 0 ≤ s < t.
Proof: It follows from the definition of expected value operator that
Z +∞
t
E[|ξ| ] = M{|ξ|t ≥ r}dr < ∞.
0
Thus we have
+∞
M{|ξ|t ≥ r}dr = 0.
Z
lim
x→∞ xt /2
The equation (4.61) is proved by the following relation,

Z +∞ Z xt
t
M {|ξ|t ≥ r}dr ≥
t
M{|ξ|t ≥ r}dr ≥ 12 xt M{|ξ| ≥ x}.
x /2 x /2
Conversely, if (4.61) holds, then there exists a number a such that

xt M{|ξ| ≥ x} ≤ 1, ∀x ≥ a.
Thus we have
a +∞
M{|ξ|s ≥ r}dr + M{|ξ|s ≥ r}dr
Z Z
E[|ξ|s ] =
0 a
a +∞
M{|ξ|s ≥ r}dr + srs−1 M{|ξ| ≥ r}dr
Z Z
=
0 a
a +∞
M{|ξ|
Z Z
s
≤ ≥ r}dr + s rs−t−1 dr
0 0
Z +∞
p
< +∞. by r dr < ∞ for any p < −1
0

Theorem 4.32 Let ξ be an uncertain variable that takes values in [a, b] and
has expected value e. Then for any positive integer k, the kth absolute moment
and kth absolute central moment satisfy the following inequalities,
b−e k e−a k
E[|ξ|k ] ≤ |a| + |b| , (4.62)
b−a b−a
b−e e−a
E[|ξ − e|k ] ≤ (e − a)k + (b − e)k . (4.63)
b−a b−a

and f (x) = |x − e|k .
4.11 Critical Values

Definition 4.23 (Liu [138]) Let ξ be an uncertain variable, and α ∈ (0, 1].
Then
ξsup (α) = sup r M {ξ ≥ r} ≥ α

(4.64)
ξinf (α) = inf r M {ξ ≤ r} ≥ α

(4.65)
Theorem 4.33 Let ξ be an uncertain variable and α a number between 0

and 1. We have
Proof: (a) If c = 0, then the part (a) is obvious. In the case of c > 0, we
have
(cξ)sup (α) = sup{r M{cξ ≥ r} ≥ α}

= c sup{r/c | M{ξ ≥ r/c} ≥ α}

= cξsup (α).
A similar way may prove (cξ)inf (α) = cξinf (α). In order to prove the part (b),
it suffices to prove that (−ξ)sup (α) = −ξinf (α) and (−ξ)inf (α) = −ξsup (α).
In fact, we have
(−ξ)sup (α) = sup{r M{−ξ ≥ r} ≥ α}

= − inf{−r | M{ξ ≤ −r} ≥ α}

= −ξinf (α).
Theorem 4.34 Let ξ be an uncertain variable. Then we have

¯
then we have
1 ≥ M{ξ < ξ(α)}

¯ + M{ξ > ξ(α)}
¯ ≥ α + α > 1.
A contradiction proves ξinf (α) ≥ ξsup (α).

Part (b): Assume that ξinf (α) > ξsup (α). It follows from the definition
of ξinf (α) that M{ξ ≤ ξ(α)}
¯ < α. Similarly, it follows from the definition of
ξsup (α) that M{ξ ≥ ξ(α)}
¯ < α. Thus
1 ≤ M{ξ ≤ ξ(α)}
¯ + M{ξ ≥ ξ(α)}
¯ < α + α ≤ 1.
A contradiction proves ξinf (α) ≤ ξsup (α). The theorem is verified.
Theorem 4.35 Let ξ be an uncertain variable. Then ξsup (α) is a decreasing

function of α, and ξinf (α) is an increasing function of α.
Proof: It follows from the definition immediately.
4.12 Entropy
This section provides a definition of entropy to characterize the uncertainty
of uncertain variables resulting from information deficiency.
Entropy of Discrete Uncertain Variables

Definition 4.24 (Liu [138]) Suppose that ξ is a discrete uncertain variable
taking values in {x1 , x2 , · · · }. Then its entropy is defined by
∞
S(M{ξ = xi })
X
H[ξ] = (4.66)
i=1
Example 4.43: Suppose that ξ is a discrete uncertain variable taking values

in {x1 , x2 , · · · }. If there exists some index k such that M{ξ = xk } = 1, and
0 otherwise, then its entropy H[ξ] = 0.
Example 4.44: Suppose that ξ is a simple uncertain variable taking values

in {x1 , x2 , · · · , xn }. If M{ξ = xi } = 0.5 for all i = 1, 2, · · · , n, then its
entropy H[ξ] = n ln 2.
Theorem 4.36 Suppose that ξ is a discrete uncertain variable taking values

in {x1 , x2 , · · · }. Then
H[ξ] ≥ 0 (4.67)
and equality holds if and only if ξ is essentially a deterministic/crisp number.

M{ξ = xi } = 0 or 1 for each i. That is, there exists one and only one index k
such that M{ξ = xk } = 1, i.e., ξ is essentially a deterministic/crisp number.
This theorem states that the entropy of an uncertain variable reaches its
minimum 0 when the uncertain variable degenerates to a deterministic/crisp
number. In this case, there is no uncertainty.
Theorem 4.37 Suppose that ξ is a simple uncertain variable taking values

in {x1 , x2 , · · · , xn }. Then
H[ξ] ≤ n ln 2 (4.68)
and equality holds if and only if M{ξ = xi } = 0.5 for all i = 1, 2, · · · , n.
Proof: Since the function S(t) reaches its maximum ln 2 at t = 0.5, we have
n
S(M{ξ = xi }) ≤ n ln 2
X
H[ξ] =
i=1
and equality holds if and only if M{ξ = xi } = 0.5 for all i = 1, 2, · · · , n.

This theorem states that the entropy of an uncertain variable reaches its
maximum when the uncertain variable is an equipossible one. In this case,
there is no preference among all the values that the uncertain variable will
take.
Entropy of Continuous Uncertain Variables

Definition 4.25 (Liu [141]) Suppose that ξ is a continuous uncertain vari-
able. Then its entropy is defined by
Z +∞
H[ξ] = S(M{ξ ≤ x})dx (4.69)
−∞
Example 4.45: Let ξ be a continuous uncertain variable with uncertainty

distribution 
 0, if x < a

Φ(x) = 0.5, if a ≤ x < b (4.70)

1, if x ≥ b.

Its entropy is
Z b
H[ξ] = − (0.5 ln 0.5 + (1 − 0.5) ln (1 − 0.5)) dx = (b − a) ln 2.
a
Theorem 4.38 Let ξ be a continuous uncertain variable. Then H[ξ] > 0.
Proof: The positivity is clear. In addition, when a continuous uncertain

variable tends to a crisp number, its entropy tends to the minimum 0. How-
ever, a crisp number is not a continuous uncertain variable.
Theorem 4.39 Let ξ be a continuous uncertain variable taking values on

the interval [a, b]. Then
H[ξ] ≤ (b − a) ln 2 (4.71)
and equality holds if and only if ξ has an uncertainty distribution (4.70).
Proof: The theorem follows from the fact that the function S(t) reaches its
maximum ln 2 at t = 0.5.
4.13 Distance
Definition 4.26 (Liu [138]) The distance between uncertain variables ξ and
η is defined as
d(ξ, η) = E[|ξ − η|]. (4.72)
Example 4.46: Let ξ and η be rectangular uncertain variables (a1 , b1 ) and

(a2 , b2 ), respectively, and (a1 , b1 ) ∩ (a2 , b2 ) = ∅. Then |ξ − η| is a rectangular
uncertain variable on the interval with endpoints |a1 − b2 | and |b1 − a2 |. Thus
the distance between ξ and η is the expected value of |ξ − η|, i.e.,
1
d(ξ, η) = (|a1 − b2 | + |b1 − a2 |) .
2
Example 4.47: Let ξ = (a1 , b1 , c1 ) and η = (a2 , b2 , c2 ) be triangular uncer-

tain variables such that (a1 , c1 ) ∩ (a2 , c2 ) = ∅. Then
1
d(ξ, η) = (|a1 − c2 | + 2|b1 − b2 | + |c1 − a2 |) .
4
Example 4.48: Let ξ = (a1 , b1 , c1 , d1 ) and η = (a2 , b2 , c2 , d2 ) be trapezoidal

uncertain variables such that (a1 , d1 ) ∩ (a2 , d2 ) = ∅. Then
1
d(ξ, η) = (|a1 − d2 | + |b1 − c2 | + |c1 − b2 | + |d1 − a2 |) .
4
Theorem 4.40 Let ξ, η, τ be uncertain variables, and let d(·, ·) be the dis-
tance. Then we have
Now we prove the part (d). It follows from the countable subadditivity axiom
that
Z +∞
d(ξ, η) = M {|ξ − η| ≥ r} dr
0
+∞
M {|ξ − τ | + |τ − η| ≥ r} dr
Z
≤
0
+∞
M {{|ξ − τ | ≥ r/2} ∪ {|τ − η| ≥ r/2}} dr
Z
≤
0
+∞
(M{|ξ − τ | ≥ r/2} + M{|τ − η| ≥ r/2}) dr
Z
≤
0
+∞ +∞
M{|ξ − τ | ≥ r/2}dr + M{|τ − η| ≥ r/2}dr
Z Z
=
0 0
= 2E[|ξ − τ |] + 2E[|τ − η|] = 2d(ξ, τ ) + 2d(τ, η).
Example 4.49: Let Γ = {γ1 , γ2 , γ3 }. Define M{∅} = 0, M{Γ} = 1 and

M{Λ} = 1/2 for any subset Λ (excluding ∅ and Γ). We set uncertain variables
ξ, η and τ as follows,
( (
1, if γ 6= γ3 −1, if γ 6= γ1
ξ(γ) = η(γ) = τ (γ) ≡ 0.
0, otherwise, 0, otherwise,
It is easy to verify that d(ξ, τ ) = d(τ, η) = 1/2 and d(ξ, η) = 3/2. Thus
3
d(ξ, η) = (d(ξ, τ ) + d(τ, η)).
2
4.14 Inequalities
Theorem 4.41 (Liu [138]) Let ξ be an uncertain variable, and f a non-
negative function. If f is even and increasing on [0, ∞), then for any given
number t > 0, we have
M{|ξ| ≥ t} ≤ E[f (ξ)]

f (t)
. (4.73)
Proof: It is clear that M{|ξ| ≥ f −1 (r)} is a monotone decreasing function

+∞
M{f (ξ) ≥ r}dr
Z
E[f (ξ)] =
0
+∞
M{|ξ| ≥ f −1 (r)}dr
Z
=
0
f (t)
M{|ξ| ≥ f −1 (r)}dr
Z
≥
0
f (t)
dr · M{|ξ| ≥ f −1 (f (t))}
Z
≥
0
= f (t) · M{|ξ| ≥ t}
Theorem 4.42 (Liu [138], Markov Inequality) Let ξ be an uncertain vari-

able. Then for any given numbers t > 0 and p > 0, we have
p
M{|ξ| ≥ t} ≤ E[|ξ|
tp
]
. (4.74)
Theorem 4.43 (Liu [138], Chebyshev Inequality) Let ξ be an uncertain vari-

able whose variance V [ξ] exists. Then for any given number t > 0, we have
M {|ξ − E[ξ]| ≥ t} ≤ Vt[ξ]

2
. (4.75)
Proof: It is a special case of Theorem 4.41 when the uncertain variable ξ is

Theorem 4.44 (Liu [138], Hölder’s Inequality) Let p and q be positive real
numbers with 1/p+1/q = 1, and let ξ and η be strongly independent uncertain
variables with E[|ξ|p ] < ∞ and E[|η|q ] < ∞. Then we have
p p
E[|ξη|] ≤ p E[|ξ|p ] q E[|η|q ]. (4.76)
√
a and b such that

E[f (|ξ|p , |η|q )] ≤ f (E[|ξ|p ], E[|η|q ]).
Theorem 4.45 (Liu [138], Minkowski Inequality) Let p be a real number

with p ≥ 1, and let ξ and η be strongly independent uncertain variables with
E[|ξ|p ] < ∞ and E[|η|p ] < ∞. Then we have
p
p
p p
E[|ξ + η|p ] ≤ p E[|ξ|p ] + p E[|η|p ]. (4.77)
√
E[f (|ξ|p , |η|p )] ≤ f (E[|ξ|p ], E[|η|p ]).
Theorem 4.46 (Liu [138], Jensen’s Inequality) Let ξ be an uncertain vari-

able, and f : < → < a convex function. If E[ξ] and E[f (ξ)] are finite, then
f (E[ξ]) ≤ E[f (ξ)]. (4.78)
f (ξ) − f (E[ξ]) ≥ k · (ξ − E[ξ]).
E[f (ξ)] − f (E[ξ]) ≥ k · (E[ξ] − E[ξ]) = 0


We have the following four convergence concepts of uncertain sequence: con-
vergence almost surely (a.s.), convergence in measure, convergence in mean,
Table 4.1: Relationship among Convergence Concepts
Convergence Convergence Convergence

⇒ ⇒
in Mean in Measure in Distribution
Definition 4.27 (Liu [138]) Suppose that ξ, ξ1 , ξ2 , · · · are uncertain vari-

ables defined on the uncertainty space (Γ, L, M). The sequence {ξi } is said
to be convergent a.s. to ξ if there exists an event Λ with M{Λ} = 1 such that
lim |ξi (γ) − ξ(γ)| = 0 (4.79)
i→∞
for every γ ∈ Λ. In that case we write ξi → ξ, a.s.

ables. We say that the sequence {ξi } converges in measure to ξ if
lim
i→∞
M {|ξi − ξ| ≥ ε} = 0 (4.80)
for every ε > 0.

ables with finite expected values. We say that the sequence {ξi } converges in
mean to ξ if
lim E[|ξi − ξ|] = 0. (4.81)
i→∞
lim E[|ξi − ξ|2 ] = 0. (4.82)
i→∞
Definition 4.30 (Liu [138]) Suppose that Φ, Φ1 , Φ2 , · · · are the uncertainty

distributions of uncertain variables ξ, ξ1 , ξ2 , · · · , respectively. We say that
{ξi } converges in distribution to ξ if Φi → Φ at any continuity point of Φ.
Theorem 4.31 (Liu [138]) Suppose that ξ, ξ1 , ξ2 , · · · are uncertain variables.
If {ξi } converges in mean to ξ, then {ξi } converges in measure to ξ.
Proof: It follows from the Markov inequality that for any given number
ε > 0, we have
M{|ξi − ξ| ≥ ε} ≤ E[|ξiε− ξ|] → 0
as i → ∞. Thus {ξi } converges in measure to ξ. The theorem is proved.
Section 4.16 - Conditional Uncertainty 195
Theorem 4.32 (Liu [138]) Suppose ξ, ξ1 , ξ2 , · · · are uncertain variables. If

{ξi } converges in measure to ξ, then {ξi } converges in distribution to ξ.
Proof: Let x be a given continuity point of the uncertainty distribution Φ.

On the one hand, for any y > x, we have
It follows from the countable subadditivity axiom that
Φi (x) ≤ Φ(y) + M{|ξi − ξ| ≥ y − x}.
Since {ξi } converges in measure to ξ, we have M{|ξi − ξ| ≥ y − x} → 0 as

i → ∞. Thus we obtain lim supi→∞ Φi (x) ≤ Φ(y) for any y > x. Letting
y → x, we get
lim sup Φi (x) ≤ Φ(x). (4.83)
i→∞
{ξ ≤ z} = {ξi ≤ x, ξ ≤ z} ∪ {ξi > x, ξ ≤ z} ⊂ {ξi ≤ x} ∪ {|ξi − ξ| ≥ x − z}
which implies that
Φ(z) ≤ Φi (x) + M{|ξi − ξ| ≥ x − z}.
Since M{|ξi − ξ| ≥ x − z} → 0, we obtain Φ(z) ≤ lim inf i→∞ Φi (x) for any
Φ(x) ≤ lim inf Φi (x). (4.84)

i→∞
4.16 Conditional Uncertainty

We consider the uncertain measure of an event A after it has been learned
that some other event B has occurred. This new uncertain measure of A is
called the conditional uncertain measure of A given B.
In order to define a conditional uncertain measure M{A|B}, at first we
have to enlarge M{A ∩ B} because M{A ∩ B} < 1 for all events whenever
M{B} < 1. It seems that we have no alternative but to divide M{A ∩ B} by
M{B}. Unfortunately, M{A∩B}/M{B} is not always an uncertain measure.
However, the value M{A|B} should not be greater than M{A ∩ B}/M{B}
M{A|B} ≤ MM
{A ∩ B}
{B}
. (4.85)
M{A|B} = 1 − M{Ac |B} ≥ 1 − M{A

c
∩ B}
M{B} . (4.86)
Furthermore, since (A ∩ B) ∪ (Ac ∩ B) = B, we have M{B} ≤ M{A ∩ B} +

M{Ac ∩ B} by using the countable subadditivity axiom. Thus
M{Ac ∩ B} ≤ M{A ∩ B} ≤ 1.
0≤1−
M{B} M{B} (4.87)
Hence any numbers between 1− M{Ac ∩B}/M{B} and M{A∩B}/M{B} are

reasonable values that the conditional uncertain measure may take. Based on
the maximum uncertainty principle (see Appendix E), we have the following
conditional uncertain measure.
Definition 4.33 (Liu [138]) Let (Γ, L, M) be an uncertainty space, and A, B ∈

L. Then the conditional uncertain measure of A given B is defined by

M{A ∩ B} , if M{A ∩ B} < 0.5
M{B} M{B}




M{A|B} =  1 − M{A ∩ B} , if M{Ac ∩ B} < 0.5


c
(4.88)




M {B} M {B}

0.5, otherwise
provided that M{B} > 0.

It follows immediately from the definition of conditional uncertain mea-
sure that
M{Ac ∩ B} ≤ M{A|B} ≤ M{A ∩ B} .
1−
M{B} M{B} (4.89)
Furthermore, the conditional uncertain measure obeys the maximum uncer-

tainty principle, and takes values as close to 0.5 as possible.
Remark 4.5: Conditional uncertain measure coincides with conditional

probability, conditional credibility, and conditional chance.
Theorem 4.47 Let (Γ, L, M) be an uncertainty space, and B an event with

M{B} > 0. Then M{·|B} defined by (4.88) is an uncertain measure, and
(Γ, L, M{·|B}) is an uncertainty space.
Proof: It is sufficient to prove that M{·|B} satisfies the normality, mono-

tonicity, self-duality and countable subadditivity axioms. At first, it satisfies
the normality axiom, i.e.,
M{Γ|B} = 1 − M{Γ M{∅} = 1.

c
∩ B}
M{B} =1−
M{B}

M{A1 ∩ B} ≤ M{A2 ∩ B} < 0.5,
M{B} M{B}
then
M{A1 |B} = M{A M{A2 ∩ B}
M{B} ≤ M{B} = M{A2 |B}.
1 ∩ B}
If
M{A1 ∩ B} ≤ 0.5 ≤ M{A2 ∩ B} ,
M{B} M{B}
then M{A1 |B} ≤ 0.5 ≤ M{A2 |B}. If
M{A1 ∩ B} ≤ M{A2 ∩ B} ,
0.5 <
M{B} M{B}
then we have
M{A1 |B} = 1 −

M{Ac1 ∩ B} ∨ 0.5 ≤ 1 − M{Ac2 ∩ B} ∨ 0.5 = M{A |B}.
M{B} M{B} 2
This means that M{·|B} satisfies the monotonicity axiom. For any event A,
if
M{A ∩ B} ≥ 0.5, M{Ac ∩ B} ≥ 0.5,
M{B} M{B}
then we have M{A|B} + M{A |B} = 0.5 + 0.5 = 1 immediately. Otherwise,
c

M{A ∩ B} < 0.5 < M{Ac ∩ B} ,
M{B} M{B}
then we have
c M
M{A|B} + M{A |B} = M{B} + 1 − M{B} = 1.
{A ∩ B}

M {A ∩ B}

That is, M{·|B} satisfies the self-duality axiom. Finally, for any countable
sequence {Ai } of events, if M{Ai |B} < 0.5 for all i, it follows from the
countable subadditivity axiom that
(∞ ) ∞
) M M{Ai ∩ B} X
[ X
(∞ Ai ∩ B ∞
M M{Ai |B}.
[ i=1 i=1
i=1
Ai ∩ B ≤
M{B} ≤
M{B} =
i=1
Suppose there is one term greater than 0.5, say
M{A1 |B} ≥ 0.5, M{Ai |B} < 0.5, i = 2, 3, · · ·

If M{∪i Ai |B} = 0.5, then we immediately have

(∞ ) ∞
M M{Ai |B}.
[ X
Ai ∩ B ≤
i=1 i=1
If M{∪i Ai |B} > 0.5, we may prove the above inequality by the following
facts:
∞ ∞
!
[ \
c c
A1 ∩ B ⊂ (Ai ∩ B) ∪ Ai ∩ B ,
i=2 i=1
∞
(∞ )
M M{Ai ∩ B} + M
X \
{Ac1 ∩ B} ≤ Aci ∩B ,
i=2 i=1
(∞ )
M
\
(∞ ) Aci ∩B
M
[ i=1
i=1
Ai |B =1−
M{B} ,
∞
M{Ai ∩ B}
X
∞
X M {Ac1 ∩ B}
M{Ai |B} ≥ 1 − M{B} + M{B} .
i=2
i=1
If there are at least two terms greater than 0.5, then the countable subad-
ditivity is clearly true. Thus M{·|B} satisfies the countable subadditivity
axiom. Hence M{·|B} is an uncertain measure. Furthermore, (Γ, L, M{·|B})
is an uncertainty space.
Example 4.50: Let ξ and η be two uncertain variables. Then we have


M{ξ = x, η = y} , M{ξ = x, η = y} < 0.5
M{η = y} if
M{η = y}




M{ξ 6= x, η = y} , M{ξ 6= x, η = y} < 0.5

M {ξ = x|η = y} = 





1−
M{η = y} if
M{η = y}

0.5, otherwise
provided that M{η = y} > 0.

Definition 4.34 (Liu [138]) Let ξ be an uncertain variable on (Γ, L, M). A
conditional uncertain variable of ξ given B is a measurable function ξ|B from
the conditional uncertainty space (Γ, L, M{·|B}) to the set of real numbers
such that
ξ|B (γ) ≡ ξ(γ), ∀γ ∈ Γ. (4.90)
Definition 4.35 (Liu [138]) The conditional uncertainty distribution Φ: < →

[0, 1] of an uncertain variable ξ given B is defined by
Φ(x|B) = M {ξ ≤ x|B} (4.91)
provided that M{B} > 0.

Example 4.51: Let ξ and η be uncertain variables. Then the conditional
uncertainty distribution of ξ given η = y is

M{ξ ≤ x, η = y} , M{ξ ≤ x, η = y} < 0.5
M{η = y} if
M{η = y}




M M


Φ(x|η = y) = {ξ > x, η = y} {ξ > x, η = y}





1−
M{η = y} , if
M{η = y} < 0.5

0.5, otherwise
provided that M{η = y} > 0.

Definition 4.36 (Liu [138]) The conditional uncertainty density function φ
of an uncertain variable ξ given B is a nonnegative function such that
Z x
Φ(x|B) = φ(y|B)dy, ∀x ∈ <, (4.92)
−∞
Z +∞
φ(y|B)dy = 1 (4.93)
−∞
where Φ(x|B) is the conditional uncertainty distribution of ξ given B.
Definition 4.37 (Liu [138]) Let ξ be an uncertain variable. Then the con-
ditional expected value of ξ given B is defined by
Z +∞ Z 0
E[ξ|B] = M{ξ ≥ r|B}dr − M{ξ ≤ r|B}dr (4.94)
0 −∞
Following conditional uncertain measure and conditional expected value,

we also have conditional variance, conditional moments, conditional critical
values, conditional entropy as well as conditional convergence.
Chapter 5
Uncertain Process
An uncertain process is essentially a sequence of uncertain variables indexed

by time or space. The study of uncertain process was started by Liu [139]
in 2008. This chapter introduces the basic concepts of uncertain process,
including renewal process and stationary process.
5.1 Basic Definitions

Definition 5.1 (Liu [139]) Let T be an index set and let (Γ, L, M) be an
uncertainty space. An uncertain process is a measurable function from T ×
(Γ, L, M) to the set of real numbers, i.e., for each t ∈ T and any Borel set B
of real numbers, the set

{Xt ∈ B} = {γ ∈ Γ Xt (γ) ∈ B} (5.1)
is an event.
That is, an uncertain process Xt (γ) is a function of two variables such

that the function Xt∗ (γ) is an uncertain variable for each t∗ .
Definition 5.2 An uncertain process Xt is said to have independent incre-

ments if
Xt1 − Xt0 , Xt2 − Xt1 , · · · , Xtk − Xtk−1 (5.2)
are independent uncertain variables for any times t0 < t1 < · · · < tk . An
uncertain process Xt is said to have stationary increments if, for any given
t > 0, the increments Xs+t −Xs are identically distributed uncertain variables
for all s > 0.
Definition 5.3 For any partition of closed interval [0, t] with 0 = t1 < t2 <
· · · < tk+1 = t, the mesh is written as
∆ = max |ti+1 − ti |.
1≤i≤k
202 Chapter 5 - Uncertain Process
Let m > 0 be a real number. Then the m-variation of uncertain process Xt

is
Xk
kXkm
t = lim |Xti − Xti |m (5.3)
∆→0
i=1
provided that the limit exists almost surely and is an uncertain process. Espe-
cially, the m-variation is called total variation if m = 1; and squared variation
if m = 2.
5.2 Continuity Concepts

Definition 5.4 For each fixed γ ∗ , the function Xt (γ ∗ ) is called a sample
path of the uncertain process Xt .
Definition 5.5 An uncertain process Xt is said to be sample-continuous if

almost all sample paths are continuous with respect to t.
Definition 5.6 An uncertain process Xt is said to be continuous in mean if

for every t we have
lim E[|Xs − Xt |] = 0. (5.4)
s→t
Definition 5.7 An uncertain process Xt is said to be continuous in measure

if for every t and ε > 0 we have
lim M[|Xs − Xt | > ε] = 0. (5.5)

s→t
Definition 5.8 An uncertain process Xt is said to be continuous almost

surely if for every t we have
M lim Xs = Xt = 1.
h i
(5.6)
s→t
5.3 Renewal Process

Definition 5.9 (Liu [139]) Let ξ1 , ξ2 , · · · be iid positive uncertain variables.
Define S0 = 0 and Sn = ξ1 + ξ2 + · · · + ξn for n ≥ 1. Then the uncertain
process
Nt = max n Sn ≤ t (5.7)
n≥0
is called a renewal process.
If ξ1 , ξ2 , · · · denote the interarrival times of successive events. Then Sn

can be regarded as the waiting time until the occurrence of the nth event,
and Nt is the number of renewals in (0, t]. Each sample path of Nt is a
right-continuous and increasing step function taking only nonnegative integer
values. Furthermore, the size of each jump of Nt is always 1. In other words,
Section 5.3 - Renewal Process 203
N. t
...
..........
...
..
...........
4 ...
...
..............................
..
..
... ..
.......... .........................................................
3 ...
....
.. ..
..
..
.. .. ..
.......... ....................................... ..
2 ...
... .. ..
..
..
..
... .. ..
.. ..
... .. ..
.........................................................
1 .........
.... .. ..
..
..
..
..
..
... .. .. ..
.. .. .. .
.
......................................................................................................................................................................................................................................
0 ... ... ... ... ... t
ξ ...
... 1 ... .... ξ 2 ξ ....
... 3 ... ξ
....
4 ....
...
.. .. .. .. ..
S0 S1 S2 S3 S4
Figure 5.1: A Sample Path of Renewal Process
Nt has at most one renewal at each time. In particular, Nt does not jump at
time 0. Since Nt ≥ n is equivalent to Sn ≤ t, we immediately have
M{Nt ≥ n} = M{Sn ≤ t}. (5.8)
Theorem 5.1 Let Nt be a renewal process with uncertain interarrival times.

Then we have
∞
M{Sn ≤ t}.
X
E[Nt ] = (5.9)
n=1
Proof: Since Nt takes only nonnegative integer values, we have

∞ ∞ Z n
M{Nt ≥ r}dr = M{Nt ≥ r}dr
Z X
E[Nt ] =
0 n=1 n−1
∞ ∞
M{Nt ≥ n} = M{Sn ≤ t}.
X X
=
n=1 n=1
Example 5.1: A renewal process Nt is called an rectangular renewal process

if ξ1 , ξ2 , · · · are iid rectangular uncertain variables (a, b) with a > 0. Then
for each nonnegative integer n, we have

 0, if t < na

M{Nt ≥ n} =  0.5, if na ≤ t < nb (5.10)
1, if t ≥ nb,


1 t t
E[Nt ] = + (5.11)
2 a b
where bxc represents the maximal integer less than or equal to x.
204 Chapter 5 - Uncertain Process
Example 5.2: A renewal process Nt is called a triangular renewal process

if ξ1 , ξ2 , · · · are iid triangular uncertain variables (a, b, c) with a > 0. Then
for each nonnegative integer n, we have


 0, if t ≤ na


 t − na
 2n(b − a) , if na ≤ t ≤ nb



M{Nt ≥ n} =  nc − 2nb + t (5.12)

 , if nb ≤ t ≤ nc
2n(c − b)





if t ≥ nc.

1,
Theorem 5.2 (Renewal Theorem) Let Nt be a renewal process with uncer-

tain interarrival times ξ1 , ξ2 , · · · Then

E[Nt ] 1
lim =E . (5.13)
t→∞ t ξ1
Proof: Since ξ1 is a positive uncertain variable and Nt is a nonnegative
uncertain variable, we have
Z ∞ Z ∞
M t ≥ r dr, E ξ = M ξ ≥ r dr.

E[Nt ] Nt 1 1
=
t 0 1 0 1
It is easy to verify that
M ≤M

Nt 1
≥r ≥r
t ξ1
and
M =M

Nt 1
lim ≥r ≥r
t→∞ t ξ1
for any real numbers t > 0 and r > 0. It follows from Lebesgue dominated
convergence theorem that
Z ∞ Z ∞
M M

Nt 1
lim ≥ r dr = ≥ r dr.
t→∞ 0 t 0 ξ1
Hence (5.13) holds.
5.4 Stationary Process

Definition 5.10 An uncertain process Xt is called stationary if for any pos-
itive integer k and any times t1 , t2 , · · · , tk and s, the uncertain vectors
(Xt1 , Xt2 , · · · , Xtk ) and (Xt1 +s , Xt2 +s , · · · , Xtk +s ) (5.14)
are identically distributed.
Chapter 6
Uncertain Calculus
Uncertain calculus, founded by Liu [139] in 2008 and revised by Liu [141] in
2009, is composed of canonical process, uncertain integral and chain rule.
6.1 Canonical Process

Definition 6.1 (Liu [139][141]) An uncertain process Ct is said to be a
canonical process if
(i) C0 = 0 and Ct is sample-continuous,
(ii) Ct has stationary and independent increments,
(iii) every increment Cs+t − Cs is a normal uncertain variable with expected
value 0 and variance t2 .
C. t
..
........
...
... ......
... .. ... ........ .
..... .. .... ............
... ...... .... ... ..... .....
... ..
. ......
... ... ......
... ........ ........
..
..
...
... ... .... . .. .
....... ...
... ..
. ... ....... ....... .... ... ...
. .
.
... ...... ....
. ..
. ........ .. ...... . ....
.. ... ... .. . .... .. ......
... .
. ..... ... ... ....... ............. .. .
... .. .
... .. .. . .
... .. .. ... . .. .. .. . . .
... .. ..... .....
... ............. ......... ..... ... ... ....... .....
... ... .. .. .....
... ... ... ...
... ..
.
... ............
... ...
......
...............................................................................................................................................................................................................................................................
t
Figure 6.1: A Sample Path of Canonical Process
Remark 6.1: In 1828 the botanist Brown observed irregular movement of

pollen suspended in liquid. A rigorous mathematical definition of Brownian
motion was given by Wiener in 1931. This movement is now known as Brow-
nian motion. Please mention that Brownian motion is not a special canonical
process.
206 Chapter 6 - Uncertain Calculus
Theorem 6.1 (Existence Theorem) There is a canonical process.
Proof: Without loss of generality, we only prove that there is a canonical

process on the range of t ∈ [0, 1]. Let

ξ(r) r represents rational numbers in [0, 1]
be a countable sequence of independently normal uncertain variables with

expected value zero and variance one. For each integer n, we define an
uncertain process

k
 1 i k
 X
ξ , if t = (k = 0, 1, · · · , n)

Xn (t) = n i=1 n n


linear, otherwise.

Since the limit

lim Xn (t)
n→∞
exists almost surely, we may verify that the limit meets the conditions of
canonical process. Hence there is a canonical process.
Let Ct be a canonical process, and dt an infinitesimal time interval. Then
dCt = Ct+dt − Ct is an uncertain process with E[dCt ] = 0 and
dt2
≤ E[dCt2 ] ≤ dt2 . (6.1)
2
Theorem 6.2 (Reflection Principle) Let Ct be a canonical process. Then
for any level x > 0 and any time t > 0, we have
M{Ct ≥ x} ≤ M max Cs ≥ x ≤ 2M{Ct ≥ x}.

(6.2)
0≤s≤t
For any level x < 0 and any time t > 0, we have
M{Ct ≤ x} ≤ M max Cs ≤ x ≤ 2M{Ct ≤ x}.

(6.3)
0≤s≤t
Arithmetic Canonical Process

Definition 6.2 Let Ct be a canonical process. Then for any real numbers e
and σ,
At = et + σCt (6.4)
is called an arithmetic canonical process, where e is called the drift and σ is
called the diffusion.
It is easy to verify that the expected value E[At ] = et and variance

V [At ] = σ 2 t2 at any time t.
Section 6.2 - Uncertain Integral 207
C. t
...
.......... .
... ....
.... .. .. ......
.. ... ... ... .... ........ ..
... . . . .. ..... ... ..... ............
... ......... .. .. ..... .... ... ....... .....
.
... . .. ... .. ... ...
... .. . ... .. ... ........ .
......
.. .. ...
x ...........................................................................................................................................................................................................................................................
.... ............ . ..
. ...
. . ...
... .... .
. ..
.. ...... ....... .... .. .
.
.
........ .. ....... ..
. . ..
... .... . ... .............. ...........
. ... .... .... ... ..... ..
.
... ................ .... .... ... ......... ..... . .. ...... .
... .. .. ... ... .. ... .. .. .. ..... ..
... .. ... .... .... ....... ............ .... ........... ... ... ..... ..
... ..... ... .. .. .. .. .. ......
............. ..........
. .. ... ... .......
. ......
... .
.. . .. ...... .
... ..... ....... ........ .
... ..
... ...
... ..... ....
... ... ....
... ...
......
..............................................................................................................................................................................................................................................................
t
Figure 6.2: Reflection Principle of Canonical Process
Geometric Canonical Process

Definition 6.3 Let Ct be a canonical process. Then et+σCt is an arithmetic
canonical process, and the uncertain process
Gt = exp(et + σCt ) (6.5)
is called a geometric canonical process, where e is called the log-drift and σ

is called the log-diffusion.
6.2 Uncertain Integral

Definition 6.4 (Liu [141]) Let Xt be an uncertain process and let Ct be a
canonical process. For any partition of closed interval [a, b] with a = t1 <
t2 < · · · < tk+1 = b, the mesh is written as
∆ = max |ti+1 − ti |.
1≤i≤k
Then the uncertain integral of Xt with respect to Ct is

Z b k
X
Xt dCt = lim Xti · (Cti+1 − Cti ) (6.6)
a ∆→0
i=1
provided that the limit exists almost surely and is an uncertain variable.
Example 6.1: Let Ct be a canonical process. Then for any partition 0 =

t1 < t2 < · · · < tk+1 = s, we have
Z s k
X
dCt = lim (Cti+1 − Cti ) ≡ Cs − C0 = Cs .
0 ∆→0
i=1

t1 < t2 < · · · < tk+1 = s, we have
k
X
sCs = ti+1 Cti+1 − ti Cti
i=1
k
X k
X
= ti (Cti+1 − Cti ) + Cti+1 (ti+1 − ti )
i=1 i=1
Z s Z s
→ tdCt + Ct dt
0 0
as ∆ → 0. It follows that
Z s Z s
tdCt = sCs − Ct dt.
0 0

t1 < t2 < · · · < tk+1 = s, we have
k
X
Cs2 = Ct2i+1 − Ct2i
i=1
k k
X 2 X
= Cti+1 − Cti +2 Cti Cti+1 − Cti
i=1 i=1
Z s
→0+2 Ct dCt
0
as ∆ → 0. That is, Z s
1 2
Ct dCt = C .
0 2 s
Example 6.4: Let Ct be a canonical process. Then for any number α

(0 < α < 1), the uncertain process
Z t
Ft = (t − s)α dCs (6.7)
0
is called a fractional canonical process with index α.
6.3 Chain Rule

Theorem 6.3 (Liu [141]) Let Ct be a canonical process, and let h(t, c) be a
continuously differentiable function. Define Xt = h(t, Ct ). Then we have the
following chain rule
∂h ∂h
dXt = (t, Ct )dt + (t, Ct )dCt . (6.8)
∂t ∂c
Section 6.3 - Chain Rule 209
Proof: Since the function h is continuously differentiable, by using Taylor

series expansion, the infinitesimal increment of Xt has a first-order approxi-
mation
∂h ∂h
∆Xt = (t, Ct )∆t + (t, Ct )∆Ct .
∂t ∂c
Hence we obtain the chain rule because it makes
Z s Z s
∂h ∂h
Xs = X0 + (t, Ct )dt + (t, Ct )dCt
0 ∂t 0 ∂c
for any s ≥ 0.
Remark 6.2: The infinitesimal increment dCt in (6.8) may be replaced with
the derived canonical process
dYt = ut dt + vt dCt (6.9)
where ut and vt are absolutely integrable uncertain processes, thus producing

∂h ∂h
dh(t, Yt ) = (t, Yt )dt + (t, Yt )dYt . (6.10)
∂t ∂c
Example 6.5: Applying the chain rule, we obtain the following formula
d(tCt ) = Ct dt + tdCt .
Hence we have
Z s Z s Z s
sCs = d(tCt ) = Ct dt + tdCt .
0 0 0
That is, Z s Z s
0 0
d(Ct2 ) = 2Ct dCt .
Then we have Z s Z s
Cs2 = d(Ct2 ) =2 Ct dCt .
0 0
It follows that Z s
1 2
Ct dCt = C .
0 2 s
d(Ct3 ) = 3Ct2 dCt .

Thus we get Z s Z s
Cs3 = d(Ct3 ) =3 Ct2 dCt .
0 0
That is Z s
1 3
Ct2 dCt = C .
0 3 s
6.4 Integration by Parts

Theorem 6.4 (Liu [139], Integration by Parts) Suppose that Ct is a canon-
ical process and F (t) is an absolutely continuous function. Then
Z s Z s
F (t)dCt = F (s)Cs − Ct dF (t). (6.11)
0 0
Proof: By defining h(t, Ct ) = F (t)Ct and using the chain rule, we get
d(F (t)Ct ) = Ct dF (t) + F (t)dCt .
Thus Z s Z s Z s
F (s)Cs = d(F (t)Ct ) = Ct dF (t) + F (t)dCt
0 0 0
which is just (6.11).
Example 6.8: Assume F (t) ≡ 1. Then by using the integration by parts,

we immediately obtain Z s
dCt = Cs .
0
Example 6.9: Assume F (t) = t. Then by using the integration by parts,

we immediately obtain
Z s Z s
0 0
Example 6.10: Assume F (t) = t2 . Then by using the integration by parts,

we obtain
Z s Z s Z s
t2 dCt = s2 Cs − Ct dt2 = s2 Cs − 2 tCt dt
0 0 0
Z s
= (s2 − 2s)Cs + 2 Ct dt.
0
Example 6.11: Assume F (t) = sin t. Then by using the integration by

parts, we obtain
Z s Z s Z s
sin tdCt = sin s · Cs − Ct d sin t = sin s · Cs − cos t · Ct dt.
0 0 0
Chapter 7
Uncertain Differential
Equation
Uncertain differential equation, proposed by Liu [139] in 2008, is a type of

differential equation driven by canonical process. This chapter will discuss
the existence, uniqueness and stability of solutions of uncertain differential
equations. This chapter also presents some applications of uncertain differen-
tial equation in uncertain finance, uncertain control, uncertain filtering, and
uncertain dynamical system.
7.1 Uncertain Differential Equation

Definition 7.1 (Liu [139]) Suppose Ct is a canonical process, and f and g
are some given functions. Then
dXt = f (t, Xt )dt + g(t, Xt )dCt (7.1)
is called an uncertain differential equation. A solution is an uncertain process

Xt that satisfies (7.1) identically in t.
Remark 7.1: Note that there is no precise definition for the terms dXt ,
dt and dCt in the uncertain differential equation (7.1). The mathematically
meaningful form is the uncertain integral equation
Z s Z s
Xs = X0 + f (t, Xt )dt + g(t, Xt )dCt . (7.2)
0 0
However, the differential form is more convenient for us. This is the main
reason why we accept the differential form.
212 Chapter 7 - Uncertain Differential Equation
Example 7.1: Let Ct be a canonical process. Then the uncertain differential

equation
dXt = adt + bdCt
has a solution Xt = at + bCt which is just an arithmetic canonical process.

equation
dXt = aXt dt + bXt dCt
has a solution
Xt = exp (at + bCt )
which is just a geometric canonical process.

equation
dXt = (aXt + b)dt + cdCt
has a solution
Z t
b
Xt = exp(at) X0 + (1 − exp(−at)) + c exp(−as)dCs
a 0
provided that a 6= 0.
7.2 Existence and Uniqueness Theorem

Theorem 7.1 (Existence and Uniqueness Theorem) The uncertain differen-
tial equation (7.1) has a unique solution if the coefficients f (x, t) and g(x, t)
satisfy the Lipschitz condition
|f (x, t) − f (y, t)| + |g(x, t) − g(y, t)| ≤ L|x − y|, ∀x, y ∈ <, t ≥ 0 (7.3)
and linear growth condition
|f (x, t)| + |g(x, t)| ≤ L(1 + |x|), ∀x ∈ <, t ≥ 0 (7.4)
for some constant L. Moreover, the solution is sample-continuous.
7.3 Stability
Definition 7.2 (Liu [141]) An uncertain differential equation is said to be
stable if for any given > 0 and ε > 0, there exists δ > 0 such that for any
solutions Xt and Yt , we have
M{|Xt − Yt | > } < ε, ∀t > 0 (7.5)
whenever |X0 − Y0 | < δ.

Section 7.4 - Uncertain Finance 213
Example 7.4: The uncertain differential equation dXt = adt+bdCt is stable

since for any given > 0 and ε > 0, we may take δ = and have
M{|Xt − Yt | > } = M{|X0 − Y0 | > } = M{∅} = 0 < ε

for any t > 0 whenever |X0 − Y0 | < δ.
Example 7.5: The uncertain differential equation dXt = Xt dt + bdCt is

unstable since for any given > 0 and any different initial solutions X0 and
Y0 , we have
M{|Xt − Yt | > } = M{exp(t)|X0 − Y0 | > } = 1

provided that t is sufficiently large.
7.4 Uncertain Finance
If we assume that the stock price follows some uncertain differential equation,
then we may produce a new topic of uncertain finance. As an example, let
stock price follow geometric canonical process. Then we have a stock model
(Liu [141]) in which the bond price Xt and the stock price Yt are determined
by
(
dXt = rXt dt
(7.6)
dYt = eYt dt + σYt dCt
where r is the riskless interest rate, e is the stock drift, σ is the stock diffusion,
and Ct is a canonical process.
European Call Option Price
A European call option gives the holder the right to buy a stock at a speci-
fied time for specified price. Assume that the option has strike price K and
expiration time s. Then the payoff from such an option is (Ys − K)+ . Con-
sidering the time value of money resulted from the bond, the present value
of this payoff is exp(−rs)(Ys − K)+ . Hence the European call option price
should be
fc = exp(−rs)E[(Ys − K)+ ]. (7.7)
It is clear that the option price is a decreasing function of interest rate r.

That is, the European call option will devaluate if the interest rate is raised;
and the European call option will appreciate in value if the interest rate is
reduced. In addition, the option price is also a decreasing function of strike
price K.
Let us consider the financial market described by stock model (7.6). The
European call option price is
fc = exp(−rs)E[(Y0 exp(es + σCs ) − K)+ ]

Z +∞
= exp(−rs) M{Y0 exp(es + σCs ) − K ≥ x}dx
0
+∞
M{exp(es + σCs ) ≥ y}dy
Z
= exp(−rs)Y0
K/Y0
Z +∞
= exp(−rs)Y0 M{es + σCs ≥ ln y}dy
K/Y0
+∞ −1
π(ln y − es)
Z
= exp(−rs)Y0 1 + exp √ dy
K/Y0 3σs
which produces a European call option price formula (Liu [141])

+∞ −1
π(ln y − es)
Z
fc = exp(−rs)Y0 1 + exp √ dy. (7.8)
K/Y0 3σs
European Put Option Price

A European put option gives the holder the right to sell a stock at a speci-
fied time for specified price. Assume that the option has strike price K and
expiration time s. Then the payoff from such an option is (K − Ys )+ . Con-
sidering the time value of money resulted from the bond, the present value
of this payoff is exp(−rs)(K − Ys )+ . Hence the European put option price
should be
fp = exp(−rs)E[(K − Ys )+ ]. (7.9)
It is easy to verify that the option price is a decreasing function of interest
rate r, and is an increasing function of strike price K.
Let us consider the financial market described by stock model (7.6). The
European put option price is
fp = exp(−rs)E[(K − Y0 exp(es + σCs ))+ ]

Z +∞
= exp(−rs) M{K − Y0 exp(es + σCs ) ≥ x}dx
0
+∞
M{exp(es + σCs ) ≤ y}dy
Z
= exp(−rs)Y0
K/Y0
K/Y0
M{es + σCs ≤ ln y}dy
Z
= exp(−rs)Y0
0
K/Y0 −1
π(es − ln y)
Z
= exp(−rs)Y0 1 + exp √ dy
0 3σs
Section 7.4 - Uncertain Finance 215
which produces a European put option price formula
K/Y0 −1
π(es − ln y)
Z
fp = exp(−rs)Y0 1 + exp √ dy. (7.10)
0 3σs
Multi-factor Stock Model
Now we assume that there are multiple stocks whose prices are determined
by multiple canonical processes. For this case, we have a multi-factor stock
model (Liu [141]) in which the bond price Xt and the stock prices Yit are
determined by


 dXt = rXt dt

n
X (7.11)
 dYit = ei Yit dt +

 σij Yit dCjt , i = 1, 2, · · · , m
j=1
where r is the riskless interest rate, ei are the stock drift coefficients, σij
are the stock diffusion coefficients, Cit are independent canonical processes,
i = 1, 2, · · · , m, j = 1, 2, · · · , n.
Portfolio Selection
For the stock model (7.11), we have the choice of m + 1 different investments.
At each instant t we may choose a portfolio (βt , β1t , · · · , βmt ) (i.e., the in-
vestment fractions meeting βt + β1t + · · · + βmt = 1). Then the wealth Zt at
time t should follow uncertain differential equation
m
X m X
X n
dZt = rβt Zt dt + ei βit Zt dt + σij βit Zt dCjt . (7.12)
i=1 i=1 j=1
Portfolio selection problem is to find an optimal portfolio (βt , β1t , · · · , βmt )

such that the expected wealth E[Zs ] is maximized and variance V [Zs ] is
minimized at terminal time s. In order to balance the two objectives, we may,
for example, maximize the expected wealth subject to a variance constraint,
i.e.,

 max E[Zs ]

subject to: (7.13)

V [Zs ] ≤ V

where V is a given level. This is just the so-called mean-variance model in

finance.
No-Arbitrage
The stock model (7.11) is said to be no-arbitrage if there is no portfolio
(βt , β1t , · · · , βmt ) such that for some time s > 0, we have
M{exp(−rs)Zs ≥ Z0 } = 1 (7.14)
and
M{exp(−rs)Zs > Z0 } > 0 (7.15)
where Zt is determined by (7.12) and represents the wealth at time t. We may
prove that the stock model (7.11) is no-arbitrage if and only if its diffusion
matrix  
σ11 σ12 · · · σ1n
 σ21 σ22 · · · σ2n 
 
 .. .. .. .. 
 . . . . 
σm1 σm2 ··· σmn
has rank m, i.e., the row vectors are linearly independent.
7.5 Uncertain Control

A control system is assumed to follow the uncertain differential equation
dXt = f (t, Xt , Zt )dt + g(t, Xt , Zt )dCt (7.16)
where Xt is the state and Zt is a control. An uncertain control theory is
needed to deal with this type of uncertain control system.
Assume that R is the return function and T is the function of terminal
reward. If we want to maximize the expected return on [0, s] by using an
optimal control, then we have the following control model,
 Z s


 max E R(t, Xt , Z t )dt + T (s, X s )
 Zt 0
(7.17)

 subject to:


dXt = f (t, Xt , Zt )dt + g(t, Xt , Zt )dCt .
Hamilton-Jacobi-Bellman equation provides a necessary condition for ex-
tremum of stochastic control model, and Zhu’s equation [279] provides a
necessary condition for extremum of fuzzy control model. What is the nec-
essary condition for extremum of general control model? How do we find the
optimal control?
7.6 Uncertain Filtering

Suppose an uncertain system Xt is described by an uncertain differential
equation
dXt = f (t, Xt )dt + g(t, Xt )dCt (7.18)
Section 7.7 - Uncertain Dynamical System 217
where Ct is a canonical process. We also have an observation X

bt with
dX
bt = fb(t, Xt )dt + gb(t, Xt )dC
bt (7.19)
where C bt is another canonical process that is independent of Ct . The uncer-

tain filtering problem is to find the best estimate of Xt based on the obser-
vation X bt .
One outstanding contribution to stochastic filtering problem is Kalman-
Bucy filter. How do we filter the noise away from the observation for general
uncertain process?
7.7 Uncertain Dynamical System

Usually a stochastic dynamical system is described by a stochastic differential
equation, and a fuzzy dynamical system is described by a fuzzy differential
equation. Here we define an uncertain dynamical system as an uncertain
differential equation. Especially, a first-order uncertain system is a first-order
uncertain differential equation
Ẋt = f (t, Xt ) + g(t, Xt )Ċt , (7.20)
and a second-order uncertain system is a second-order uncertain differential

equation
Ẍt = f (t, Xt , Ẋt ) + g(t, Xt , Ẋt )Ċt , (7.21)
where
dXt dẊt dCt
Ẋt = , Ẍt = , Ċt = . (7.22)
dt dt dt
We should develop a new theory for uncertain dynamical systems.
Chapter 8
Uncertain Logic
Uncertain logic was designed by Li and Liu [108] in 2009 as a generalization

of logic for dealing with uncertain knowledge. It is essentially the uncertainty
theory in which uncertain variables take values either 0 or 1. A key point in
uncertain logic is that the truth value of an uncertain proposition is defined
as the uncertain measure that the proposition is true. One advantage of
uncertain logic is the well consistency with classical logic!
8.1 Uncertain Proposition
Definition 8.1 An uncertain proposition is a statement with truth value be-

longing to [0, 1].
Example 8.1: “Tom is tall with truth value 0.8” is an uncertain proposition,
where “Tom is tall” is a statement, and its truth value is 0.8 in uncertain
measure.
Generally, we use ξ to express the uncertain proposition and use c to
express its uncertain measure value. In fact, the uncertain proposition ξ is
essentially an uncertain variable
(
1 with uncertain measure c
ξ= (8.1)
0 with uncertain measure 1 − c
where ξ = 1 means ξ is true and ξ = 0 means ξ is false.
Definition 8.2 Uncertain propositions are called independent if they are in-
dependent uncertain variables.
220 Chapter 8 - Uncertain Logic
Example 8.2: Two uncertain propositions ξ and η are independent if and

only if they are independent uncertain variables, i.e.,
M{(ξ = 1) ∩ (η = 1)} = M{ξ = 1} ∧ M{η = 1},

M{(ξ = 1) ∩ (η = 0)} = M{ξ = 1} ∧ M{η = 0},
M{(ξ = 0) ∩ (η = 1)} = M{ξ = 0} ∧ M{η = 1},
M{(ξ = 0) ∩ (η = 0)} = M{ξ = 0} ∧ M{η = 0}.
Or equivalently,
M{(ξ = 1) ∪ (η = 1)} = M{ξ = 1} ∨ M{η = 1},

M{(ξ = 1) ∪ (η = 0)} = M{ξ = 1} ∨ M{η = 0},
M{(ξ = 0) ∪ (η = 1)} = M{ξ = 0} ∨ M{η = 1},
M{(ξ = 0) ∪ (η = 0)} = M{ξ = 0} ∨ M{η = 0}.
8.2 Uncertain Formula
An uncertain formula is a finite sequence of uncertain propositions and con-
nective symbols that must make sense. For example, let ξ, η, τ be uncertain
propositions. Then
X = ¬ξ, X = ξ ∧ η, X = (ξ ∨ η) → τ
are all uncertain formulas. Note that an uncertain formula X is essentially

an uncertain variable taking values 0 or 1. If X = 1, then X is true; if X = 0,
then X is false.
Definition 8.3 Uncertain formulas are called independent if they are inde-
pendent uncertain variables.
8.3 Truth Value

Truth value is a key concept in uncertain logic, and is defined as the uncertain
measure that the uncertain formula is true.
Definition 8.4 (Li and Liu [108]) Let X be an uncertain formula. Then
the truth value of X is defined as the uncertain measure that the uncertain
formula X is true, i.e.,
T (X) = M{X = 1}. (8.2)
The truth value T (X) = 1 means X is certainly true, T (X) = 0 means X

is certainly false, and T (X) = 0.5 means X is totally uncertain. The higher
the truth value is, the more true the uncertain formula is.
Section 8.4 - Some Laws 221
Example 8.3: Let ξ and η be two independent uncertain propositions with

uncertain measure values a and b, respectively. Then
T (ξ) = M{ξ = 1} = a, (8.3)
T (¬ξ) = M{ξ = 0} = 1 − a, (8.4)

T (ξ ∨ η) = M{ξ ∨ η = 1} = M{(ξ = 1) ∪ (η = 1)} = a ∨ b, (8.5)
T (ξ ∧ η) = M{ξ ∧ η = 1} = M{(ξ = 1) ∩ (η = 1)} = a ∧ b, (8.6)
T (ξ → η) = T (¬ξ ∨ η) = (1 − a) ∨ b. (8.7)
Theorem 8.1 (Chen [21]) Let X be an uncertain formula containing inde-

pendent uncertain propositions ξ1 , ξ2 , · · · , ξn whose truth function is f . Then
the truth value of X is

 sup min νi (xi ),
 f (x1 ,x2 ,··· ,xn )=1 1≤i≤n






 if sup min νi (xi ) < 0.5
f (x1 ,x2 ,··· ,xn )=1 1≤i≤n

T (X) = (8.8)

 1− sup min νi (xi ),

 f (x1 ,x2 ,··· ,xn )=0 1≤i≤n


if sup min νi (xi ) ≥ 0.5



1≤i≤n

f (x1 ,x2 ,··· ,xn )=1
where νi (xi ) are identification functions (no matter what types they are) of
ξi , i = 1, 2, · · · , n, respectively.
Proof: Since X = 1 if and only if f (ξ1 , ξ2 , · · · , ξn ) = 1, the formula of truth

value follows from the operational law immediately.
8.4 Some Laws

Theorem 8.2 (Law of Contradiction) For any uncertain formula X, we
have
T (X ∧ ¬X) = 0. (8.9)
Proof: It follows from the definition of truth value and property of uncertain
measure that
T (X ∧ ¬X) = M{X ∧ ¬X = 1} = M{(X = 1) ∩ (X = 0)} = M{∅} = 0.
Theorem 8.3 (Law of Excluded Middle) For any uncertain formula X, we

have
T (X ∨ ¬X) = 1. (8.10)
Proof: It follows from the definition of truth value and property of uncertain
measure that
T (X ∨ ¬X) = M{X ∨ ¬X = 1} = M{(X = 1) ∪ (X = 0)} = 1.
Theorem 8.4 (Law of Truth Conservation) For any uncertain formula X,

we have
T (X) + T (¬X) = 1. (8.11)
Proof: It follows from the self-duality of uncertain measure that
T (¬X) = M{¬X = 1} = M{X = 0} = 1 − M{X = 1} = 1 − T (X).
Theorem 8.5 (De Morgan’s Law) For any uncertain formulas X and Y , we
have
T (¬(X ∧ Y )) = T ((¬X) ∨ (¬Y )), (8.12)
T (¬(X ∨ Y )) = T ((¬X) ∧ (¬Y )). (8.13)
Proof: It follows from the basic properties of uncertain measure that
T (¬(X ∧ Y )) = M{X ∧ Y = 0} = M{(X = 0) ∩ (Y = 0)}

= M{(¬X) ∨ (¬Y ) = 1} = T ((¬X) ∨ (¬Y ))
which proves the first equality. A similar way may verify the second equality.
Theorem 8.6 (Law of Contraposition) For any uncertain formulas X and

Y , we have
T (X → Y ) = T (¬Y → ¬X). (8.14)
Proof: It follows from the basic properties of uncertain measure that
T (X → Y ) = M{(¬X) ∨ Y = 1} = M{(X = 0) ∪ (Y = 1)}

= M{Y ∨ (¬X) = 1} = T (¬Y → ¬X).
Theorem 8.7 (Monotonicity and Subadditivity) For any uncertain formulas

X and Y , we have
T (X) ∨ T (Y ) ≤ T (X ∨ Y ) ≤ T (X) + T (Y ). (8.15)

Section 8.4 - Some Laws 223
Proof: It follows from the monotonicity of uncertain measure that
T (X ∨ Y ) = M{X ∨ Y = 1} = M{(X = 1) ∪ (Y = 1)}

≥ M{X = 1} ∨ M{Y = 1} = T (X) ∨ T (Y ).
It follows from the subadditivity of uncertain measure that
T (X ∨ Y ) = M{X ∨ Y = 1} = M{(X = 1) ∪ (Y = 1)}

≤ M{X = 1} + M{Y = 1} = T (X) + T (Y ).
The theorem is verified.
Theorem 8.8 For any uncertain formulas X and Y , we have
T (X) + T (Y ) − 1 ≤ T (X ∧ Y ) ≤ T (X) ∧ T (Y ). (8.16)
Proof: It follows from the monotonicity of truth value that
T (X ∧ Y ) = 1 − T (¬X ∨ ¬Y ) ≤ 1 − T (¬X) ∨ T (¬Y )

= (1 − T (¬X)) ∧ (1 − T (¬Y )) = T (X) ∧ T (Y ).
It follows from the subadditivity of truth value that
T (X ∧ Y ) = 1 − T (¬X ∨ ¬Y ) ≥ 1 − (T (¬X) + T (¬Y ))

= 1 − (1 − T (X)) − (1 − T (Y )) = T (X) + T (Y ) − 1.
Theorem 8.9 For any uncertain formula X, we have
T (X → X) = 1, T (X → ¬X) = 1 − T (X). (8.17)
Proof: It follows from the law of excluded middle and law of truth conser-
vation that
T (X → X) = T (¬X ∨ X) = 1,
T (X → ¬X) = T (¬X ∨ ¬X) = T (¬X) = 1 − T (X).
Independence Case
Theorem 8.10 If uncertain formulas X and Y are independent, then we
have
T (X ∨ Y ) = T (X) ∨ T (Y ), T (X ∧ Y ) = T (X) ∧ T (Y ). (8.18)

Proof: Since X and Y are independent uncertain formulas, they are inde-
pendent uncertain variables. Hence
T (X ∨ Y ) = M{X ∨ Y = 1} = M{X = 1} ∨ M{Y = 1} = T (X) ∨ T (Y ),

T (X ∧ Y ) = M{X ∧ Y = 1} = M{X = 1} ∧ M{Y = 1} = T (X) ∧ T (Y ).
Theorem 8.11 If uncertain formulas X and Y are independent, then we

have
T (X → Y ) = (1 − T (X)) ∨ T (Y ). (8.19)
Proof: Since X and Y are independent, the ¬X and Y are also independent.
It follows that
T (X → Y ) = T (¬X ∨ Y ) = T (¬X) ∨ T (Y ) = (1 − T (X)) ∨ T (Y )
which proves the theorem.

Chapter 9
Uncertain Inference
Uncertain inference was proposed by Liu [141] in 2009. How do we infer

Y from the uncertain knowledge X and X → Y? This is the main issue of
uncertain inference. A key point is that the inference rules use the tool of
conditional uncertainty.
9.1 Uncertain Sets

Before introducing uncertain inference, let us recall the concept of uncertain
set proposed by Liu [138] in 2007.
Definition 9.1 (Liu [138]) An uncertain set is a measurable function ξ from

an uncertainty space (Γ, L, M) to a collection of sets of real numbers, i.e., for
any Borel set B, the set

{γ ∈ Γ ξ(γ) ∩ B 6= ∅} (9.1)
is an event.
Let ξ and ξ ∗ be two uncertain sets. In order to define the degree that ξ ∗
matches ξ, we first introduce the following symbols:
{ξ ∗ ⊂ ξ} = {γ ∈ Γ ξ ∗ (γ) ⊂ ξ(γ)},

(9.2)
{ξ ∗ ⊂ ξ c } = {γ ∈ Γ ξ ∗ (γ) ⊂ ξ(γ)c },

(9.3)
{ξ ∗ 6⊂ ξ c } = {γ ∈ Γ ξ ∗ (γ) ∩ ξ(γ) 6= ∅}.

(9.4)
What are concerned is to know the degree that ξ ∗ belongs to ξ. Note that
we cannot use the symbol “⊂” because it has certain meaning in classical set
theory and is not applicable to uncertain sets. Thus we need a new symbol
“ξ ∗ B ξ” to represent the belongingness that reads as “ξ ∗ matches ξ”.
226 Chapter 9 - Uncertain Inference
Definition 9.2 Let ξ and ξ ∗ be two uncertain sets. Then the event ξ ∗ B ξ
(i.e., ξ ∗ matches ξ) is defined as
{ξ ∗ ⊂ ξ}, if M{ξ ∗ ⊂ ξ} > M{ξ ∗ ⊂ ξ c }

(
∗
{ξ B ξ} = (9.5)
{ξ ∗ 6⊂ ξ c }, if M{ξ ∗ ⊂ ξ} ≤ M{ξ ∗ ⊂ ξ c }.
That is, we take the event of {ξ ∗ ⊂ ξ} or {ξ ∗ 6⊂ ξ c } such that its uncertain

measure is as closed to 0.5 as possible. This idea coincides with the maximum
uncertainty principle.
Definition 9.3 Let ξ and ξ ∗ be two uncertain sets. Then the matching degree
of ξ ∗ B ξ is defined as
M{ξ ∗ ⊂ ξ}, if M{ξ ∗ ⊂ ξ} > M{ξ ∗ ⊂ ξ c }

(
M{ξ B ξ} =
∗
(9.6)
M{ξ ∗ 6⊂ ξ c }, if M{ξ ∗ ⊂ ξ} ≤ M{ξ ∗ ⊂ ξ c }.
9.2 Inference Rule
Let X and Y be two concepts. It is assumed that the perfect relation between
X and Y is not known. Here we suppose that we only have a rule “if X is ξ
then Y is η” where ξ and η are two uncertain sets (not uncertain variable!).
From X = ξ ∗ we infer that Y is
η ∗ = η|ξ∗ Bξ (9.7)
which is the conditional uncertain set of η given ξ ∗ B ξ. Thus we have the

following inference rule (Liu [141]):
Rule: If X is ξ then Y is η
From: X is ξ ∗ (9.8)
Infer: Y is η ∗ = η|ξ∗ Bξ
Theorem 9.1 Assume the if-then rule “if X is ξ then Y is η”. From M{ξ ∗ B
ξ} = 1 we infer η ∗ = η.
Proof: It follows from M{ξ ∗ B ξ} = 1 and the definition of conditional

uncertainty that η ∗ = η.
Example 9.1: Suppose ξ and ξ ∗ are independent uncertain sets with first
identification functions λ and λ∗ , and η has a first identification function ν
that is independent of (ξ, ξ ∗ ). If λ∗ (x) ≤ λ(x) for any x ∈ <, then M{ξ ∗ Bξ} =
1 and ν ∗ (y) = ν(y). This means that if ξ ∗ matches ξ, then η ∗ matches η.
Example 9.2: If the supports of ξ and ξ ∗ are disjoint, then M{ξ ∗ B ξ} = 0.

For this case, we set ν ∗ (y) ≡ 0.5. This means that “if nothing is known then
anything is possible”.
Section 9.5 - General Inference Rule 227
9.3 Inference Rule with Multiple Antecedents

Assume that we have a rule “if X is ξ and Y is η then Z is τ ”. From X = ξ ∗
and Y = η ∗ we infer that Z is
τ ∗ = τ |(ξ∗ Bξ)∩(η∗ Bη) (9.9)
which is the conditional uncertain set of τ given ξ ∗ B ξ and η ∗ B η. Thus we

have the following inference rule:
Rule: If X is ξ and Y is η then Z is τ

From: X is ξ ∗ and Y is η ∗ (9.10)
Infer: Z is τ ∗ = τ |(ξ∗ Bξ)∩(η∗ Bη)
Theorem 9.2 Assume the if-then rule “if X is ξ and Y is η then Z is τ ”.

From M{ξ ∗ B ξ} = 1 and M{η ∗ B η} = 1 we infer τ ∗ = τ .
Proof: If M{ξ ∗ Bξ} = 1 and M{η ∗ Bη} = 1, then M{(ξ ∗ Bξ)∩(η ∗ Bη)} = 1.
It follows from the definition of conditional uncertainty that τ ∗ = τ .
9.4 Inference Rule with Multiple If-Then Rules

Assume that we have two rules “if X is ξ1 then Y is η1 ” and “if X is ξ2 then
Y is η2 ”. From the rule “if X is ξ1 then Y is η1 ” and X = ξ ∗ we infer that Y
is
η1∗ = η1 |ξ1∗ Bξ .
From the rule “if X is ξ2 then Y is η2 ” and X = ξ ∗ we infer that Y is
η2∗ = η2 |ξ2∗ Bξ .
The uncertain sets η1∗ and η2∗ will aggregate a new η ∗ that is determined by
M{η1∗ ⊂ B} ∧ M{η2∗ ⊂ B},




if M{η1∗ ⊂ B} + M{η2∗ ⊂ B} < 1



M{η ⊂ B} = 

∗
(9.11)
 M {η1∗ ⊂ B} ∨ M{η2∗ ⊂ B},
if M{η1∗ ⊂ B} + M{η2∗ ⊂ B} ≥ 1.




for any Borel set B. That is, we assign it the value of M{η1∗ ⊂ B} or
M{η2∗ ⊂ B} such that it is as far from 0.5 as possible. Thus we infer Y = η∗ .
Hence we have the following inference rule:
Rule 1: If X is ξ1 then Y is η1
Rule 2: If X is ξ2 then Y is η2
(9.12)
From: X is ξ ∗
Infer: Y is η ∗ determined by (9.11)
228 Chapter 9 - Uncertain Inference
9.5 General Inference Rule

As an exercise, please the reader gives an inference rule for the following
problem:
Rule 1: If X1 is ξ11 and X2 is ξ12 and · · · and Xm is ξ1m then Y is η1

Rule 2: If X1 is ξ21 and X2 is ξ22 and · · · and Xm is ξ2m then Y is η2
···
Rule n: If X1 is ξn1 and X2 is ξn2 and · · · and Xm is ξnm then Y is ηn
From: X1 is ξ1∗ and X2 is ξ2∗ and · · · and Xm is ξm
∗
∗
Infer: Y is η
What is the uncertain set η ∗ ?

Appendix A
Measure and Integral
A.1 Measurable Sets

Algebra and σ-algebra are very important and fundamental concepts in mea-
sure theory.
Definition A.1 Let Ω be a nonempty set. A collection A is called an algebra
over Ω if the following conditions hold:
(a) Ω ∈ A;
(b) if A ∈ A, then Ac ∈ A;
(c) if Ai ∈ A for i = 1, 2, · · · , n, then ∪ni=1 Ai ∈ A.
If the condition (c) is replaced with closure under countable union, then A is
called a σ-algebra over Ω.
Example A.1: Assume that Ω is a nonempty set. Then {∅, Ω} is the
smallest σ-algebra over Ω, and the power set P (all subsets of Ω) is the
largest σ-algebra over Ω.
Example A.2: Let A be the set of all finite disjoint unions of all intervals
of the form (−∞, a], (a, b], (b, ∞) and ∅. Then A is an algebra over <, but
not a σ-algebra because Ai = (0, (i − 1)/i] ∈ A for all i but
∞
Ai = (0, 1) 6∈ A.
[
i=1
Theorem A.1 A σ-algebra A is closed under difference, countable union,

countable intersection, upper limit, lower limit, and limit. That is,
∞ ∞
A2 \ A1 ∈ A; Ai ∈ A; Ai ∈ A;
[ \
(A.1)
i=1 i=1
∞ [
∞
Ai ∈ A;
\
lim sup Ai = (A.2)
i→∞
k=1 i=k
230 Appendix A - Measure and Integral
∞ \
∞
Ai ∈ A;
[
lim inf Ai = (A.3)
i→∞
k=1 i=k
lim Ai ∈ A. (A.4)
i→∞
Theorem A.2 The intersection of any collection of σ-algebras is a σ-algebra.

Furthermore, for any nonempty class C, there is a unique minimal σ-algebra
containing C.
Theorem A.3 (Monotone Class Theorem) Assume that A0 is an algebra
over Ω, and C is a monotone class of subsets of Ω (if Ai ∈ C and Ai ↑ A
or Ai ↓ A, then A ∈ C). If A0 ⊂ C and σ(A0 ) is the smallest σ-algebra
containing A0 , then σ(A0 ) ⊂ C.
Definition A.2 Let Ω be a nonempty set, and A a σ-algebra over Ω. Then
(Ω, A) is called a measurable space, and any elements in A are called mea-
surable sets.
Theorem A.4 The smallest σ-algebra B containing all open intervals is
called the Borel algebra of <, any elements in B are called a Borel set.
Borel algebra B is a special σ-algebra over <. It follows from Theorem A.2
that Borel algebra is unique. It has been proved that open set, closed set,
the set of rational numbers, the set of irrational numbers, and countable set
of real numbers are all Borel sets.
Example A.3: We divide the interval [0, 1] into three equal open intervals
from which we choose the middle one, i.e., (1/3, 2/3). Then we divide each
of the remaining two intervals into three equal open intervals, and choose
the middle one in each case, i.e., (1/9, 2/9) and (7/9, 8/9). We perform this
process and obtain Dij for j = 1, 2, · · · , 2i−1 and i = 1, 2, · · · Note that {Dij }
is a sequence of mutually disjoint open intervals. Without loss of generality,
suppose Di1 < Di2 < · · · < Di,2i−1 for i = 1, 2, · · · Define the set
i−1
∞ 2[
[
D= Dij . (A.5)
i=1 j=1
Then C = [0, 1] \ D is called the Cantor set. In other words, x ∈ C if and

only if x can be expressed in ternary form using only digits 0 and 2, i.e.,
∞
X ai
x= (A.6)
i=1
3i
where ai = 0 or 2 for i = 1, 2, · · · The Cantor set is closed, uncountable, and

a Borel set.
Let Ω1 , Ω2 , · · · , Ωn be any sets (not necessarily subsets of the same space).
The product Ω = Ω1 × Ω2 × · · · × Ωn is the set of all ordered n-tuples of the
form (x1 , x2 , · · · , xn ), where xi ∈ Ωi for i = 1, 2, · · · , n.
Section 1.2 - Classical Measures 231
Definition A.3 Let Ai be σ-algebras over Ωi , i = 1, 2, · · · , n, respectively.

Write Ω = Ω1 × Ω2 × · · · × Ωn . A measurable rectangle in Ω is a set A =
A1 × A2 × · · · × An , where Ai ∈ Ai for i = 1, 2, · · · , n. The smallest σ-
algebra containing all measurable rectangles of Ω is called the product σ-
algebra, denoted by A = A1 × A2 × · · · × An .
Note that the product σ-algebra A is the smallest σ-algebra containing

measurable rectangles, rather than the product of A1 , A2 , · · · , An .
Product σ-algebra may be easily extended to countably infinite case by
defining a measurable rectangle as a set of the form
A = A1 × A2 × · · ·
where Ai ∈ Ai for all i and Ai = Ωi for all but finitely many i. The smallest
σ-algebra containing all measurable rectangles of Ω = Ω1 × Ω2 × · · · is called
the product σ-algebra, denoted by
A = A1 × A2 × · · ·
A.2 Classical Measures
This appendix introduces the concepts of classical measure, measure space,
Lebesgue measure, and product measure.
Definition A.4 Let Ω be a nonempty set, and A a σ-algebra over Ω. A

classical measure π is a set function on A satisfying
Axiom 1. (Nonnegativity) π{A} ≥ 0 for any A ∈ A;
Axiom 2. (Countable Additivity) For every countable sequence of mutually
disjoint measurable sets {Ai }, we have
(∞ ) ∞
[ X
π Ai = π{Ai }. (A.7)
i=1 i=1
Example A.4: Length, area and volume are instances of measure concept.
Definition A.5 Let Ω be a nonempty set, A a σ-algebra over Ω, and π a

measure on A. Then the triplet (Ω, A, π) is called a measure space.
It has been proved that there is a unique measure π on the Borel algebra
of < such that π{(a, b]} = b − a for any interval (a, b] of <.
Definition A.6 The measure π on the Borel algebra of < satisfying
π{(a, b]} = b − a, ∀(a, b]
is called the Lebesgue measure.

Theorem A.5 (Product Measure Theorem) Let (Ωi , Ai , πi ), i = 1, 2, · · · , n

be measure spaces such that πi {Ωi } < ∞ for i = 1, 2, · · · , n. Write
A = A1 × A2 × · · · × An .
Ω = Ω 1 × Ω 2 × · · · × Ωn ,
Then there is a unique measure π on A such that
π{A1 × A2 × · · · × An } = π1 {A1 } × π2 {A2 } × · · · × πn {An } (A.8)
for every measurable rectangle A1 × A2 × · · · × An . The measure π is called

the product of π1 , π2 , · · · , πn , denoted by π = π1 × π2 × · · · × πn . The triplet
(Ω, A, π) is called the product measure space.
Theorem A.6 (Infinite Product Measure Theorem) Assume that (Ωi , Ai , πi )

are measure spaces such that πi {Ωi } = 1 for i = 1, 2, · · · Write
A = A1 × A2 × · · ·
Ω = Ω1 × Ω2 × · · ·
Then there is a unique measure π on A such that
π{A1 ×· · ·×An ×Ωn+1 ×Ωn+2 ×· · · } = π1 {A1 }×π2 {A2 }×· · ·×πn {An } (A.9)
for any measurable rectangle A1 × · · · × An × Ωn+1 × Ωn+2 × · · · and all

n = 1, 2, · · · The measure π is called the infinite product, denoted by π =
π1 × π2 × · · · The triplet (Ω, A, π) is called the infinite product measure space.
A.3 Measurable Functions

This section introduces the concepts of measurable function, simple function,
step function, absolutely continuous function, singular function, and Cantor
function.
Definition A.7 A function f from (Ω, A) to the set of real numbers is said
to be measurable if
f −1 (B) = {ω ∈ Ω|f (ω) ∈ B} ∈ A (A.10)
for any Borel set B of real numbers. If Ω is a Borel set, then A is always
assumed to be the Borel algebra over Ω.
Theorem A.7 The function f is measurable from (Ω, A) to < if and only if
f −1 (I) ∈ A for any open interval I of <.
Definition A.8 A function f : < → < is said to be continuous if for any

given x ∈ < and ε > 0, there exists a δ > 0 such that |f (y) − f (x)| < ε
whenever |y − x| < δ.
Section 1.3 - Measurable Functions 233
Example A.5: Any continuous function f is measurable, because f −1 (I) is

an open set (not necessarily interval) of < for any open interval I ∈ <.
Example A.6: A monotone function f from < to < is measurable because

f −1 (I) is an interval for any interval I.
Example A.7: A function is said to be simple if it takes a finite set of

values. A function is said to be step if it takes a countably infinite set of
values. Generally speaking, a step (or simple) function is not necessarily
measurable except that it can be written as f (x) = ai if x ∈ Ai , where Ai
are measurable sets, i = 1, 2, · · · , respectively.
Example A.8: Let f be a measurable function from (Ω, A) to <. Then its
positive part and negative part
( (
+ f (ω), if f (ω) ≥ 0 − −f (ω), if f (ω) ≤ 0
f (ω) = f (ω) =
0, otherwise, 0, otherwise
are measurable functions, because
+
ω f (ω) > t = ω f (ω) > t ∪ ω f (ω) ≤ 0 if t < 0 ,
−
ω f (ω) > t = ω f (ω) < −t ∪ ω f (ω) ≥ 0 if t < 0 .
Example A.9: Let f1 and f2 be measurable functions from (Ω, A) to <.

Then f1 ∨ f2 and f1 ∧ f2 are measurable functions, because

ω f1 (ω) ∨ f2 (ω) > t = ω f1 (ω) > t ∪ ω f2 (ω) > t ,

ω f1 (ω) ∧ f2 (ω) > t = ω f1 (ω) > t ∩ ω f2 (ω) > t .
Theorem A.8 Let {fi } be a sequence of measurable functions from (Ω, A)
to <. Then the following functions are measurable:
sup fi (ω); inf fi (ω); lim sup fi (ω); lim inf fi (ω). (A.11)
1≤i<∞ 1≤i<∞ i→∞ i→∞
Especially, if limi→∞ fi (ω) exists, then it is also a measurable function.

Theorem A.9 Let f be a nonnegative measurable function from (Ω, A) to
<. Then there exists an increasing sequence {hi } of nonnegative simple mea-
surable functions such that
lim hi (ω) = f (ω), ∀ω ∈ Ω. (A.12)
i→∞
Furthermore, the functions hi may be defined as follows,


 k − 1 , if k − 1 ≤ f (ω) < k , k = 1, 2, · · · , i2i

hi (ω) = 2i 2i 2i (A.13)

 i, if i ≤ f (ω)
for i = 1, 2, · · ·
Definition A.9 A function f : < → < is said to be Lipschitz continuous if

there is a positive number K such that
|f (y) − f (x)| ≤ K|y − x|, ∀x, y ∈ <. (A.14)
Definition A.10 A function f : < → < is said to be absolutely continuous
if for any given ε > 0, there exists a small number δ > 0 such that
m
X
|f (yi ) − f (xi )| < ε (A.15)
i=1
for every finite disjoint class {(xi , yi ), i = 1, 2, · · · , m} of bounded open in-

tervals for which
Xm
|yi − xi | < δ. (A.16)
i=1
Definition A.11 A continuous and increasing function f : < → < is said to

be singular if f is not a constant and its derivative f 0 = 0 almost everywhere.
Example A.10: Let C be the Cantor set. We define a function g on C as

follows,
∞ ∞
!
X ai X ai
g i
= i+1
(A.17)
i=1
3 i=1
2
where ai = 0 or 2 for i = 1, 2, · · · Then g(x) is an increasing function and
g(C) = [0, 1]. The Cantor function is defined on [0, 1] as follows,

f (x) = sup g(y) y ∈ C, y ≤ x . (A.18)
It is clear that the Cantor function f (x) is increasing such that
f (0) = 0, f (1) = 1, f (x) = g(x), ∀x ∈ C.
Moreover, f (x) is a continuous function and f 0 (x) = 0 almost everywhere.
Thus the Cantor function f is a singular function.
A.4 Lebesgue Integral

This section introduces Lebesgue integral, Lebesgue-Stieltjes integral, mono-
tone convergence theorem, Fatou’s lemma, Lebesgue dominated convergence
theorem, and Fubini theorem.
Definition A.12 Let h(x) be a nonnegative simple measurable function de-
fined by 

 c1 , if x ∈ A1
c2 , if x ∈ A2

h(x) =

 · ··
cm , if x ∈ Am

Section 1.4 - Lebesgue Integral 235
where A1 , A2 , · · · , Am are Borel sets. Then the Lebesgue integral of h on a

Borel set A is Z Xm
h(x)dx = ci π{A ∩ Ai }. (A.19)
A i=1
Definition A.13 Let f (x) be a nonnegative measurable function on the Borel

set A, and {hi (x)} a sequence of nonnegative simple measurable functions
such that hi (x) ↑ f (x) as i → ∞. Then the Lebesgue integral of f on A is
Z Z
f (x)dx = lim hi (x)dx. (A.20)
A i→∞ A
Definition A.14 Let f (x) be a measurable function on the Borel set A, and
define
( (
+ f (x), if f (x) > 0 − −f (x), if f (x) < 0
f (x) = f (x) =
0, otherwise, 0, otherwise.
Then the Lebesgue integral of f on A is

Z Z Z
f (x)dx = f + (x)dx − f − (x)dx (A.21)
A A A
f + (x)dx and f − (x)dx is finite.

R R
provided that at least one of A A
Definition A.15 Let f (x)

R be a measurable function on the Borel set A. If
both of A f + (x)dx and A f − (x)dx are finite, then the function f is said to
R
be integrable on A.
Theorem A.10 (Monotone Convergence Theorem) Let {fi } be an increas-

ing sequence of measurable functions on A. If there is an integrable function
g such that fi (x) ≥ g(x) for all i, then we have
Z Z
lim fi (x)dx = lim fi (x)dx. (A.22)
A i→∞ i→∞ A
Example A.11: The condition fi ≥ g cannot be removed in the monotone

convergence theorem. For example, let fi (x) = 0 if x ≤ i and −1 otherwise.
Then fi (x) ↑ 0 everywhere on A. However,
Z Z
lim fi (x)dx = 0 6= −∞ = lim fi (x)dx.
< i→∞ i→∞ <
Theorem A.11 (Fatou’s Lemma) Assume that {fi } is a sequence of mea-

surable functions on A. (a) If there is an integrable function g such that
fi ≥ g for all i, then
Z Z
lim inf fi (x)dx ≤ lim inf fi (x)dx. (A.23)
A i→∞ i→∞ A
(b) If there is an integrable function g such that fi ≤ g for all i, then

Z Z
lim sup fi (x)dx ≥ lim sup fi (x)dx. (A.24)
A i→∞ i→∞ A
Theorem A.12 (Lebesgue Dominated Convergence Theorem) Let {fi } be a

sequence of measurable functions on A whose limitation limi→∞ fi (x) exists
a.s. If there is an integrable function g such that |fi (x)| ≤ g(x) for any i,
then we have Z Z
lim fi (x)dx = lim fi (x)dx. (A.25)
A i→∞ i→∞ A
Example A.12: The condition |fi | ≤ g in the Lebesgue dominated conver-

gence theorem cannot be removed. Let A = (0, 1), fi (x) = i if x ∈ (0, 1/i)
and 0 otherwise. Then fi (x) → 0 everywhere on A. However,
Z Z
lim fi (x)dx = 0 6= 1 = lim fi (x)dx.
A i→∞ i→∞ A
Theorem A.13 (Fubini Theorem) Let f (x, y) be an integrable function on

<2 . Then we have
(a) f (x, y) is an integrable function of x for almost all y;
(b) f (x, y) is an integrable function of y for almost all x;
Z Z Z Z Z
(c) f (x, y)dxdy = f (x, y)dy dx = f (x, y)dx dy.
<2 < < < <
Theorem A.14 Let Φ be an increasing and right-continuous function on <.

Then there exists a unique measure π on the Borel algebra of < such that
π{(a, b]} = Φ(b) − Φ(a) (A.26)
for all a and b with a < b. Such a measure is called the Lebesgue-Stieltjes
measure corresponding to Φ.
Definition A.16 Let Φ(x) be an increasing and right-continuous function

on <, and let h(x) be a nonnegative simple measurable function, i.e.,


 c1 , if x ∈ A1
 c2 , if x ∈ A2

h(x) = ..


 .
cm , if x ∈ Am .

Then the Lebesgue-Stieltjes integral of h on the Borel set A is

Z Xm
h(x)dΦ(x) = ci π{A ∩ Ai } (A.27)
A i=1
where π is the Lebesgue-Stieltjes measure corresponding to Φ.

Section 1.4 - Lebesgue Integral 237
Definition A.17 Let f (x) be a nonnegative measurable function on the Borel

set A, and let {hi (x)} be a sequence of nonnegative simple measurable func-
tions such that hi (x) ↑ f (x) as i → ∞. Then the Lebesgue-Stieltjes integral
of f on A is Z Z
f (x)dΦ(x) = lim hi (x)dΦ(x). (A.28)
A i→∞ A
Definition A.18 Let f (x) be a measurable function on the Borel set A, and
define
( (
+ f (x), if f (x) > 0 − −f (x), if f (x) < 0
f (x) = f (x) =
0, otherwise, 0, otherwise.
Then the Lebesgue-Stieltjes integral of f on A is

Z Z Z
f (x)dΦ(x) = f + (x)dΦ(x) − f − (x)dΦ(x) (A.29)
A A A
f + (x)dΦ(x) and f − (x)dΦ(x) is finite.

R R
provided that at least one of A A
Appendix B
Euler-Lagrange Equation
Let φ(x) be an unknown function. We consider the functional of φ as follows,

Z b
L(φ) = F (x, φ(x), φ0 (x))dx (B.1)
a
where F is a known function with continuous first and second partial deriva-
tives. If L has an extremum (maximum or minimum) at φ(x), then

∂F d ∂F
− =0 (B.2)
∂φ(x) dx ∂φ0 (x)
which is called the Euler-Lagrange equation. If φ0 (x) is not involved, then

the Euler-Lagrange equation reduces to
∂F
= 0. (B.3)
∂φ(x)
Note that the Euler-Lagrange equation is only a necessary condition for the
existence of an extremum. However, if the existence of an extremum is clear
and there exists only one solution to the Euler-Lagrange equation, then this
solution must be the curve for which the extremum is achieved.
Appendix C
Logic
Propositions, connective symbols, formulas, truth functions, truth value are

basic concepts in logic. This appendix also introduces the probabilistic logic,
credibilistic logic and hybrid logic.
C.1 Traditional Logic
Propositions
A proposition is a statement with truth value belonging to {0, 1}.
Connective Symbols
In addition to the proposition symbols ξ and η, we also need the negation

symbol ¬, conjunction symbol ∧, disjunction symbol ∨, and implication sym-
bol →. Note that {¬, ∨} is complete, and
ξ ∧ η = ¬(¬ξ ∨ ¬η), ξ → η = (¬ξ) ∨ η. (C.1)
Formulas
A formula is a finite sequence of proposition and connective symbols that

must make sense. In other words, the nonsensical sequence must be excluded.
It is clear that any propositions ξ, η and τ are formulas, and so are
¬ξ, ξ ∨ η, ξ ∧ η, ξ → η, (ξ ∨ η) → τ.
However, ¬ ∨ ξ, ξ → ∨ and ξη → τ are not formulas. For any formula X, if

X = 1, then X is true; if X = 0, then X is false.
240 Appendix C - Logic
Truth Functions
Assume X is a formula containing propositions ξ1 , ξ2 , · · · , ξn . It is well-known
that there is a function f : {0, 1}n → {0, 1} such that X = 1 if and only if
f (ξ1 , ξ2 , · · · , ξn ) = 1. Such a Boolean function f is called the truth function
of X. For example, the truth function of formula ξ1 ∨ ξ2 → ξ3 is
f (1, 1, 1) = 1, f (1, 0, 1) = 1, f (0, 1, 1) = 1, f (0, 0, 1) = 1,

f (1, 1, 0) = 0, f (1, 0, 0) = 0, f (0, 1, 0) = 0, f (0, 0, 0) = 1.
C.2 Probabilistic Logic

Probabilistic logic was proposed by Nilsson [186] as a generalization of logic
for dealing with random knowledge. A random proposition is a statement with
probability value belonging to [0, 1]. A random formula is a finite sequence of
random propositions and connective symbols that must make sense. Let X
be a random formula. Then the truth value of X is defined as the probability
that the random formula X is true, i.e.,
T (X) = Pr{X = 1}. (C.2)
If X is a random formula containing random propositions ξ1 , ξ2 , · · · , ξn whose

truth function is f (x1 , x2 , · · · , xn ), then the truth value of X is
X
T (X) = p(x1 , x2 , · · · , xn ) (C.3)
f (x1 ,x2 ,··· ,xn )=1
where p(x1 , x2 , · · · , xn ) is the joint probability mass function of (ξ1 , ξ2 , · · · , ξn ).
Example C.1: Let ξ and η be independent random propositions with prob-

ability values a and b, respectively. Then we have
(a) T (ξ) = Pr{ξ = 1} = a;
(b) T (¬ξ) = Pr{ξ = 0} = 1 − a;
(c) T (ξ ∧ η) = Pr{ξ ∧ η = 1} = Pr{(ξ = 1) ∩ (η = 1)} = a · b;
(d) T (ξ ∨ η) = Pr{ξ ∨ η = 1} = Pr{(ξ = 1) ∪ (η = 1)} = a + b − a · b;
(e) T (ξ → η) = Pr{(ξ → η) = 1} = Pr{(ξ = 0) ∪ (η = 1)} = 1 − a + a · b.
Remark C.1: Probabilistic logic is consistent with the law of contradiction

and law of excluded middle, i.e.,
T (X ∧ ¬X) = 0, T (X ∨ ¬X) = 1, T (X) + T (¬X) = 1. (C.4)
C.3 Credibilistic Logic

Credibilistic logic was designed by Li and Liu [109] as a generalization of
logic for dealing with fuzzy knowledge. A fuzzy proposition is a statement
with credibility value belonging to [0, 1]. A fuzzy formula is a finite sequence
Section 3.4 - Hybrid Logic 241
of fuzzy propositions and connective symbols that must make sense. Let X
be a fuzzy formula. Then the truth value of X is defined as the credibility
that the fuzzy formula X is true, i.e.,
T (X) = Cr{X = 1}. (C.5)
If X is a fuzzy formula containing fuzzy propositions ξ1 , ξ2 , · · · , ξn whose

truth function is f (x1 , x2 , · · · , xn ), then the truth value of X is

1
T (X) = max µ(x) + 1 − max µ(x) (C.6)
2 f (x)=1 f (x)=0
where x = (x1 , x2 , · · · , xn ), and µ(x) is the joint membership function of

(ξ1 , ξ2 , · · · , ξn ).
Example C.2: Let ξ and η be (not necessarily independent) fuzzy proposi-

tions with credibility values a and b, respectively. Then we have
(a) T (ξ) = Cr{ξ = 1} = a;
(b) T (¬ξ) = Cr{ξ = 0} = 1 − a;
(c) T (ξ ∧ η) = Cr{(ξ = 1) ∩ (η = 1)} = a ∧ b provided that a + b > 1;
(d) T (ξ ∨ η) = Cr{(ξ = 1) ∪ (η = 1)} = a ∨ b provided that a + b < 1;
(e) T (ξ → η) = Cr{(ξ = 0) ∪ (η = 1)} = (1 − a) ∨ b provided that a > b.
Example C.3: Let ξ and η be independent fuzzy propositions with credi-

bility values a and b, respectively. Then we have
(a) T (ξ ∧ η) = Cr{(ξ = 1) ∩ (η = 1)} = a ∧ b;
(b) T (ξ ∨ η) = Cr{(ξ = 1) ∪ (η = 1)} = a ∨ b;
(c) T (ξ → η) = Cr{(ξ = 0) ∪ (η = 1)} = (1 − a) ∨ b.
Remark C.2: Credibilistic logic is consistent with the law of contradiction

T (X ∧ ¬X) = 0, T (X ∨ ¬X) = 1, T (X) + T (¬X) = 1. (C.7)
This is the key point that credibilistic logic is different from possibilistic logic.
C.4 Hybrid Logic

Hybrid logic was designed by Li and Liu [108] in 2009 as a generalization
of logic for dealing with fuzzy knowledge and random knowledge simultane-
ously. A hybrid formula is a finite sequence of fuzzy propositions, random
propositions and connective symbols that must make sense. Let X be a hy-
brid formula. Then the truth value of X is defined as the chance that the
hybrid formula X is true, i.e.,
T (X) = Ch{X = 1}. (C.8)

242 Appendix C - Logic
If X is a hybrid formula containing fuzzy propositions ξ1 , ξ2 , · · · , ξn and ran-

dom propositions η1 , η2 , · · · , ηm whose truth function is f , then the truth
value of X is
  
 sup  µ(x) ∧

 X
φ(y) ,

2

x



 f (x,y)=1

  



 µ(x) X

 if sup  ∧ φ(y) < 0.5
2


 x
f (x,y)=1
T (X) =   (C.9)
µ(x)

 X
1 − sup  ∧


 φ(y) ,

 x 2
f (x,y)=0



  


µ(x)

 X
if sup  ∧ φ(y) ≥ 0.5




 x 2
f (x,y)=1
where x = (x1 , x2 , · · · , xn ), y = (y1 , y2 , · · · , ym ), µ(x) is the joint mem-

bership function of (ξ1 , ξ2 , · · · , ξn ), and φ(y) is the joint probability mass
function of (η1 , η2 , · · · , ηm ).
Example C.4: Let ξ, κ be independent fuzzy propositions with credibility

values a, c, and let η, τ be independent random propositions with probability
values b, d, respectively. Then we have
(a) T (ξ ∧ η) = Ch{(ξ = 1) ∩ (η = 1)} = a ∧ b;
(b) T (ξ ∨ η) = Ch{(ξ = 1) ∪ (η = 1)} = a ∨ b;
(c) T (ξ → τ ) = Ch{(ξ = 0) ∪ (τ = 1)} = (1 − a) ∨ d;
(d) T (η → κ) = Ch{(η = 0) ∪ (κ = 1)} = (1 − b) ∨ c.
Furthermore, we have
T (ξ ∧ η → κ) = Ch{(ξ ∧ η = 0) ∪ (κ = 1)}
= 1 − Ch{(ξ = 1) ∩ (η = 1) ∩ (κ = 0)}
= 1 − Cr{(ξ = 1) ∩ (κ = 0)} ∧ Pr{η = 1}
= 1 − a ∧ (1 − c) ∧ b;
T (ξ ∧ η → τ ) = Ch{(ξ ∧ η = 0) ∪ (τ = 1)}
= 1 − Ch{(ξ = 1) ∩ (η = 1) ∩ (τ = 0)}
= 1 − Cr{ξ = 1} ∧ Pr{(η = 1) ∩ (τ = 0)}
= 1 − a ∧ (b − bd).
Remark C.3: Hybrid logic is consistent with the law of contradiction and
law of excluded middle, i.e.,
T (X ∧ ¬X) = 0, T (X ∨ ¬X) = 1, T (X) + T (¬X) = 1. (C.10)

Section 3.5 - Uncertain Logic 243
C.5 Uncertain Logic

Uncertain logic was designed by Li and Liu [108] in 2009 as a generalization
of logic for dealing with uncertain knowledge. An uncertain proposition is
a statement with truth value belonging to [0, 1]. An uncertain formula is a
finite sequence of uncertain propositions and connective symbols that must
make sense. Let X be an uncertain formula. Then the truth value of X is
defined as the uncertain measure that the uncertain formula X is true, i.e.,
T (X) = M{X = 1}. (C.11)
If X is an uncertain formula containing independent uncertain propositions

ξ1 , ξ2 , · · · , ξn whose truth function is f , then the truth value of X is

 sup min νi (xi ),
 f (x1 ,x2 ,··· ,xn )=1 1≤i≤n






 if sup min νi (xi ) < 0.5
f (x1 ,x2 ,··· ,xn )=1 1≤i≤n

T (X) = (C.12)

 1− sup min νi (xi ),

 f (x1 ,x2 ,··· ,xn )=0 1≤i≤n


if sup min νi (xi ) ≥ 0.5



f (x1 ,x2 ,··· ,xn )=1 1≤i≤n

where νi (xi ) are identification functions (no matter what types they are) of
ξi , i = 1, 2, · · · , n, respectively.
Remark C.4: Uncertain logic is consistent with the law of contradiction

T (X ∧ ¬X) = 0, T (X ∨ ¬X) = 1, T (X) + T (¬X) = 1. (C.13)

Appendix D
Inference
This appendix will introduce traditional inference, fuzzy inference and un-
certain inference.
D.1 Traditional Inference

As traditional rules of inference, the modus ponens, modus tollens and hypo-
thetical syllogism play an important role in classical logic.
Modus Ponens: From formulas X and X → Y we may infer Y . That is, if
X and X → Y are true, then Y is true.
Modus Tollens: From formulas ¬Y and X → Y we may infer ¬X. That
is, if Y is false and X → Y is true, then X is false.
Hypothetical Syllogism: From X → Y and Y → Z we infer X → Z.
That is, if both X → Y and Y → Z are true, then X → Z is true.
D.2 Fuzzy Inference

Before introducing fuzzy inference, let us recall the concept of fuzzy set that
was given by L.A. Zadeh in 1965 via membership function. For our purpose,
we accept the following definition of fuzzy set.
Definition D.1 (Liu [138]) A fuzzy set is a function ξ from a credibility

space (Θ, P, Cr) to a collection of sets of real numbers.
Let ξ and ξ ∗ be two fuzzy sets. In order to define the degree that ξ ∗
matches ξ, we first introduce the following symbols:
{ξ ∗ ⊂ ξ} = {θ ∈ Θ ξ ∗ (θ) ⊂ ξ(θ)},

(D.1)
{ξ ∗ ⊂ ξ c } = {θ ∈ Θ ξ ∗ (θ) ⊂ ξ(θ)c },

(D.2)
Section 4.3 - Hybrid Inference 245
{ξ ∗ 6⊂ ξ c } = {θ ∈ Θ ξ ∗ (θ) ∩ ξ(θ) 6= ∅}.

(D.3)
Definition D.2 Let ξ and ξ ∗ be two fuzzy sets. Then the event ξ ∗ B ξ (i.e.,
ξ ∗ matches ξ) is defined as
{ξ ∗ ⊂ ξ}, if Cr{ξ ∗ ⊂ ξ} > Cr{ξ ∗ ⊂ ξ c }

(
∗
{ξ B ξ} = (D.4)
{ξ ∗ 6⊂ ξ c }, if Cr{ξ ∗ ⊂ ξ} ≤ Cr{ξ ∗ ⊂ ξ c }.
That is, we take the event of {ξ ∗ ⊂ ξ} or {ξ ∗ 6⊂ ξ c } such that its credibility

measure is as closed to 0.5 as possible. This idea coincides with the maximum
uncertainty principle.
Definition D.3 Let ξ and ξ ∗ be two fuzzy sets. Then the matching degree
Cr{ξ ∗ ⊂ ξ}, if Cr{ξ ∗ ⊂ ξ} > Cr{ξ ∗ ⊂ ξ c }

(
∗
Cr{ξ B ξ} = (D.5)
Cr{ξ ∗ 6⊂ ξ c }, if Cr{ξ ∗ ⊂ ξ} ≤ Cr{ξ ∗ ⊂ ξ c }.
Fuzzy inference has been studied by many scholars and a lot of inference
rules have been suggested. For example, Zadeh’s compositional rule of infer-
ence, Lukasiewicz’s inference rule, and Mamdani’s inference rule are widely
used in fuzzy inference systems.
However, instead of using the above inference rules, we will introduce an
inference rule proposed by Liu [141]. Let X and Y be two concepts. It is
assumed that we have a rule “if X is a fuzzy set ξ then Y is a fuzzy set η”.
From X is a fuzzy set ξ ∗ we infer that Y is a fuzzy set
η ∗ = η|ξ∗ Bξ (D.6)
which is the conditional fuzzy set of η given ξ ∗ B ξ.
D.3 Hybrid Inference

Definition D.4 (Liu [138]) A hybrid set is a function ξ from a chance space
space (Θ, P, Cr) × (Ω, A, Pr) to a collection of sets of real numbers, i.e., for
any Borel set B, the set

{(θ, ω) ∈ Θ × Ω ξ(θ, ω) ∩ B 6= ∅} (D.7)
is an event.
Definition D.5 Let ξ and ξ ∗ be two hybrid sets. Then the event ξ ∗ B ξ (i.e.,
ξ ∗ matches ξ) is defined as
{ξ ∗ ⊂ ξ}, if Ch{ξ ∗ ⊂ ξ} > Ch{ξ ∗ ⊂ ξ c }

(
∗
{ξ B ξ} = (D.8)
{ξ ∗ 6⊂ ξ c }, if Ch{ξ ∗ ⊂ ξ} ≤ Ch{ξ ∗ ⊂ ξ c }.
246 Appendix D - Inference
Definition D.6 Let ξ and ξ ∗ be two hybrid sets. Then the matching degree
Cr{ξ ∗ ⊂ ξ}, if Ch{ξ ∗ ⊂ ξ} > Ch{ξ ∗ ⊂ ξ c }

(
∗
Ch{ξ B ξ} = (D.9)
Ch{ξ ∗ 6⊂ ξ c }, if Ch{ξ ∗ ⊂ ξ} ≤ Ch{ξ ∗ ⊂ ξ c }.
This section will introduce an inference rule proposed by Liu [141]. Let
X and Y be two concepts. It is assumed that we have a rule “if X is a hybrid
set ξ then Y is a hybrid set η”. From X is a hybrid set ξ ∗ we infer that Y is
a hybrid set
η ∗ = η|ξ∗ Bξ (D.10)
which is the conditional hybrid set of η given ξ ∗ B ξ.
D.4 Uncertain Inference

Uncertain inference was proposed by Liu [141] in 2009. Let X and Y be two
concepts. It is assumed that we only have a rule “if X is an uncertain set ξ
then Y is an uncertain set η”. From X is an uncertain set ξ ∗ , we infer that
Y is an uncertain set
η ∗ = η|ξ∗ Bξ (D.11)
which is the conditional uncertain set of η given ξ ∗ B ξ. Thus we have the
following inference rule:
Rule: If X is ξ then Y is η
From: X is ξ ∗ (D.12)
Infer: Y is η ∗ = η|ξ∗ Bξ
Appendix E
Maximum Uncertainty
Principle
An event has no uncertainty if its measure is 1 (or 0) because we may believe

that the event occurs (or not). An event is the most uncertain if its measure
is 0.5 because the event and its complement may be regarded as “equally
likely”.
In practice, if there is no information about the measure of an event,
we should assign 0.5 to it. Sometimes, only partial information is available.
For this case, the value of measure may be specified in some range. What
value does the measure take? For the safety purpose, we should assign it the
value as close to 0.5 as possible. This is the maximum uncertainty principle
proposed by Liu [138].
Maximum Uncertainty Principle: For any event, if there are multiple
reasonable values that a measure may take, then the value as close to 0.5 as
possible is assigned to the event.
Perhaps the reader would like to ask what values are reasonable. The
answer is problem-dependent. At least, the values should ensure that all
axioms about the measure are satisfied, and should be consistent with the
given information.
Example E.1: Let Λ be an event. Based on some given information, the

measure value M{Λ} is on the interval [a, b]. By using the maximum uncer-
tainty principle, we should assign


 a, if 0.5 < a ≤ b
M{Λ} =  0.5, if a ≤ 0.5 ≤ b


b, if a ≤ b < 0.5.

Appendix F
Why Uncertain Measure?
Classical measure and probability measure belong to the class of “completely

additive measure”.
Capacity, belief measure, plausibility measure, fuzzy measure, possibility
measure and necessity measure belong to the class of “completely nonaddi-
tive measure”. Since this class of measures does not assume the self-duality
property, all of those measures are inconsistent with the law of contradiction
and law of excluded middle.
Uncertain measure is neither a completely additive measure nor a com-
pletely nonadditive measure. In fact, uncertain measure is a “partially ad-
ditive measure” because of its self-duality. Credibility measure and chance
measure are special types of uncertain measure. One advantage of this class
of measures is the well consistency with the law of contradiction and law of
excluded middle.
Completely Classical Measure (Borel and Lebesgue, 1900)

Additive Measures Probability Measure (Kolmogoroff, 1933)
Capacity (Choquet, 1954)
Belief Measure (Dempster, 1967; Shafer, 1976)
Completely Plausibility Measure (ibid)
Nonadditive Measures Fuzzy Measure (Sugeno, 1974)
Possibility Measure (Zadeh, 1978)
Necessity Measure (Zadeh, 1979)
Uncertain Measure Credibility Measure (Liu and Liu, 2002)
(Liu, 2007) Chance Measure (Li and Liu, 2009)
The essential differentia between fuzziness and uncertainty is that the
former assumes
M{A ∪ B} = M{A} ∨ M{B}
for any events A and B no matter if they are independent or not, and the
latter assumes the relation only for independent events. However, a lot of
Appendix F - Why Uncertain Measure? 249
surveys showed that
M{A ∪ B} =
6 M{A} ∨ M{B}
when A and B are not independent events. Fuzziness is an extremely special

but non-practical case of uncertainty. This is the main reason why we need
the uncertainty theory.
Bibliography
[1] Alefeld G, Herzberger J, Introduction to Interval Computations, Academic

Press, New York, 1983.
[2] Atanassov KT, Intuitionistic Fuzzy Sets: Theory and Applications, Physica-
Verlag, Heidelberg, 1999.
[3] Bamber D, Goodman IR, Nguyen HT, Extension of the concept of propo-
sitional deduction from classical logic to probability: An overview of
probability-selection approaches, Information Sciences, Vol.131, 195-250,
2001.
[4] Bedford T, and Cooke MR, Probabilistic Risk Analysis, Cambridge University
Press, 2001.
[5] Bandemer H, and Nather W, Fuzzy Data Analysis, Kluwer, Dordrecht, 1992.
[6] Bellman RE, and Zadeh LA, Decision making in a fuzzy environment, Man-
agement Science, Vol.17, 141-164, 1970.
[7] Bhandari D, and Pal NR, Some new information measures of fuzzy sets,
Information Sciences, Vol.67, 209-228, 1993.
[8] Black F, and Scholes M, The pricing of option and corporate liabilities, Jour-
nal of Political Economy, Vol.81, 637-654, 1973.
[9] Bouchon-Meunier B, Mesiar R, Ralescu DA, Linear non-additive set-
functions, International Journal of General Systems, Vol.33, No.1, 89-98,
2004.
[10] Buckley JJ, Possibility and necessity in optimization, Fuzzy Sets and Systems,
Vol.25, 1-13, 1988.
[11] Buckley JJ, Stochastic versus possibilistic programming, Fuzzy Sets and Sys-
tems, Vol.34, 173-177, 1990.
[12] Cadenas JM, and Verdegay JL, Using fuzzy numbers in linear programming,
IEEE Transactions on Systems, Man and Cybernetics–Part B, Vol.27, No.6,
1016-1022, 1997.
[13] Campos L, and González, A, A subjective approach for ranking fuzzy num-
bers, Fuzzy Sets and Systems, Vol.29, 145-153, 1989.
[14] Campos L, and Verdegay JL, Linear programming problems and ranking of
fuzzy numbers, Fuzzy Sets and Systems, Vol.32, 1-11, 1989.
[15] Campos FA, Villar J, and Jimenez M, Robust solutions using fuzzy chance
constraints, Engineering Optimization, Vol.38, No.6, 627-645, 2006.
252 Bibliography
[16] Carlsson C, Fuller R, and Majlender P, A possibilistic approach to selecting

portfolios with highest utility score, Fuzzy Sets and Systems, Vol.131, No.1,
13-21, 2002.
[17] Chanas S, and Kuchta D, Multiobjective programming in optimization of
interval objective functions—a generalized approach, European Journal of
Operational Research, Vol.94, 594-598, 1996.
[18] Chen A, and Ji ZW, Path finding under uncertainty, Journal of Advance
Transportation, Vol.39, No.1, 19-37, 2005.
[19] Chen SJ, and Hwang CL, Fuzzy Multiple Attribute Decision Making: Methods
and Applications, Springer-Verlag, Berlin, 1992.
[20] Chen XW, A new existence and uniqueness theorem for fuzzy differential
equations, http://orsc.edu.cn/process/080929.pdf.
[21] Chen XW, A note on truth value in uncertain logic.
[22] Chen Y, Fung RYK, Yang J, Fuzzy expected value modelling approach for
determining target values of engineering characteristics in QFD, International
Journal of Production Research, Vol.43, No.17, 3583-3604, 2005.
[23] Chen Y, Fung RYK, Tang JF, Rating technical attributes in fuzzy QFD by
integrating fuzzy weighted average method and fuzzy expected value operator,
European Journal of Operational Research, Vol.174, No.3, 1553-1566, 2006.
[24] Choquet G, Theory of capacities, Annals de l’Institute Fourier, Vol.5, 131-
295, 1954.
[25] Dai W, Reflection principle of Liu process, http://orsc.edu.cn/process/
071110.pdf.
[26] Dai W, Lipschitz continuity of Liu process, http://orsc.edu.cn/process/
080831.pdf.
[27] Das B, Maity K, Maiti A, A two warehouse supply-chain model under possi-
bility/necessity/credibility measures, Mathematical and Computer Modelling,
Vol.46, No.3-4, 398-409, 2007.
[28] De Cooman G, Possibility theory I-III, International Journal of General Sys-
tems, Vol.25, 291-371, 1997.
[29] De Luca A, and Termini S, A definition of nonprobabilistic entropy in the
setting of fuzzy sets theory, Information and Control, Vol.20, 301-312, 1972.
[30] Dempster AP, Upper and lower probabilities induced by a multivalued map-
ping, Ann. Math. Stat., Vol.38, No.2, 325-339, 1967.
[31] Dubois D, and Prade H, Fuzzy logics and generalized modus ponens revisited,
Cybernetics and Systems, Vol.15, 293-331, 1984.
[32] Dubois D, and Prade H, Fuzzy cardinality and the modeling of imprecise
quantification, Fuzzy Sets and Systems, Vol.16, 199-230, 1985.
[33] Dubois D, and Prade H, A note on measures of specificity for fuzzy sets,
International Journal of General Systems, Vol.10, 279-283, 1985.
[34] Dubois D, and Prade H, The mean value of a fuzzy number, Fuzzy Sets and
Systems, Vol.24, 279-300, 1987.
Bibliography 253
[35] Dunyak J, Saad IW, and Wunsch D, A theory of independent fuzzy probabil-
ity for system reliability, IEEE Transactions on Fuzzy Systems, Vol.7, No.3,
286-294, 1999.
[36] Elkan C, The paradoxical success of fuzzy logic, IEEE Expert, Vol.9, No.4,
3-8, 1994.
[37] Elkan C, The paradoxical controversy over fuzzy logic, IEEE Expert, Vol.9,
No.4, 47-49, 1994.
[38] Esogbue AO, and Liu B, Reservoir operations optimization via fuzzy criterion
decision processes, Fuzzy Optimization and Decision Making, Vol.5, No.3,
289-305, 2006.
[39] Feng X, and Liu YK, Measurability criteria for fuzzy random vectors, Fuzzy
Optimization and Decision Making, Vol.5, No.3, 245-253, 2006.
[40] Feng Y, and Yang LX, A two-objective fuzzy k-cardinality assignment prob-
lem, Journal of Computational and Applied Mathematics, Vol.197, No.1, 233-
244, 2006.
[41] Fung RYK, Chen YZ, and Chen L, A fuzzy expected value-based goal pro-
graming model for product planning using quality function deployment, En-
gineering Optimization, Vol.37, No.6, 633-647, 2005.
[42] Gao J, and Liu B, New primitive chance measures of fuzzy random event,
International Journal of Fuzzy Systems, Vol.3, No.4, 527-531, 2001.
[43] Gao J, Liu B, and Gen M, A hybrid intelligent algorithm for stochastic multi-
level programming, IEEJ Transactions on Electronics, Information and Sys-
tems, Vol.124-C, No.10, 1991-1998, 2004.
[44] Gao J, and Liu B, Fuzzy multilevel programming with a hybrid intelligent
algorithm, Computers & Mathematics with Applications, Vol.49, 1539-1548,
2005.
[45] Gao J, and Lu M, Fuzzy quadratic minimum spanning tree problem, Applied
Mathematics and Computation, Vol.164, No.3, 773-788, 2005.
[46] Gao J, and Feng X, A hybrid intelligent algorithm for fuzzy dynamic inventory
problem, Journal of Information and Computing Science, Vol.1, No.4, 235-
244, 2006.
[47] Gao J, Credibilistic game with fuzzy information, Journal of Uncertain Sys-
tems, Vol.1, No.1, 74-80, 2007.
[48] Gao J, and Gao X, A new stock model for credibilistic option pricing, Journal
of Uncertain Systems, Vol.2, No.4, 243-247, 2008.
[49] Gao X, Some properties of continuous uncertain measure, International Jour-
nal of Uncertainty, Fuzziness & Knowledge-Based Systems, to be published.
(Also available at http://orsc.edu.cn/online/070512.pdf)
[50] Gil MA, Lopez-Diaz M, and Ralescu DA, Overview on the development of
fuzzy random variables, Fuzzy Sets and Systems, Vol.157, No.19, 2546-2557,
2006.
[51] González, A, A study of the ranking function approach through mean values,
Fuzzy Sets and Systems, Vol.35, 29-41, 1990.
254 Bibliography
[52] Guan J, and Bell DA, Evidence Theory and its Applications, North-Holland,
Amsterdam, 1991.
[53] Guo R, Zhao R, Guo D, and Dunne T, Random fuzzy variable modeling on
repairable system, Journal of Uncertain Systems, Vol.1, No.3, 222-234, 2007.
[54] Ha MH, Wang XZ, and Yang LZ, Sequences of (S) fuzzy integrable functions,
Fuzzy Sets and Systems, Vol.138, No.3, 507-522, 2003.
[55] Ha MH, Li Y, and Wang XF, Fuzzy knowledge representation and reasoning
using a generalized fuzzy petri net and a similarity measure, Soft Computing,
Vol.11, No.4, 323-327, 2007.
[56] Hansen E, Global Optimization Using Interval Analysis, Marcel Dekker, New
York, 1992.
[57] He Y, and Xu J, A class of random fuzzy programming model and its applica-
tion to vehicle routing problem, World Journal of Modelling and Simulation,
Vol.1, No.1, 3-11, 2005.
[58] Heilpern S, The expected value of a fuzzy number, Fuzzy Sets and Systems,
Vol.47, 81-86, 1992.
[59] Higashi M, and Klir GJ, On measures of fuzziness and fuzzy complements,
[60] Higashi M, and Klir GJ, Measures of uncertainty and information based on
possibility distributions, International Journal of General Systems, Vol.9, 43-
58, 1983.
[61] Hisdal E, Conditional possibilities independence and noninteraction, Fuzzy
Sets and Systems, Vol.1, 283-297, 1978.
[62] Hisdal E, Logical Structures for Representation of Knowledge and Uncer-
tainty, Physica-Verlag, Heidelberg, 1998.
[63] Hong DH, Renewal process with T-related fuzzy inter-arrival times and fuzzy
rewards, Information Sciences, Vol.176, No.16, 2386-2395, 2006.
[64] Hu BG, Mann GKI, and Gosine RG, New methodology for analytical and
optimal design of fuzzy PID controllers, IEEE Transactions on Fuzzy Systems,
Vol.7, No.5, 521-539, 1999.
[65] Hu LJ, Wu R, and Shao S, Analysis of dynamical systems whose inputs are
fuzzy stochastic processes, Fuzzy Sets and Systems, Vol.129, No.1, 111-118,
2002.
[66] Inuiguchi M, and Ramı́k J, Possibilistic linear programming: A brief review
of fuzzy mathematical programming and a comparison with stochastic pro-
gramming in portfolio selection problem, Fuzzy Sets and Systems, Vol.111,
No.1, 3-28, 2000.
[67] Ishibuchi H, and Tanaka H, Multiobjective programming in optimization of
the interval objective function, European Journal of Operational Research,
Vol.48, 219-225, 1990.
[68] Jaynes ET, Information theory and statistical mechanics, Physical Reviews,
Vol.106, No.4, 620-630, 1957.
Bibliography 255
[69] Ji XY, and Shao Z, Model and algorithm for bilevel Newsboy problem
with fuzzy demands and discounts, Applied Mathematics and Computation,
Vol.172, No.1, 163-174, 2006.
[70] Ji XY, and Iwamura K, New models for shortest path problem with fuzzy arc
lengths, Applied Mathematical Modelling, Vol.31, 259-269, 2007.
[71] John RI, Type 2 fuzzy sets: An appraisal of theory and applications, Interna-
tional Journal of Uncertainty, Fuzziness & Knowledge-Based Systems, Vol.6,
No.6, 563-576, 1998.
[72] Kacprzyk J, and Esogbue AO, Fuzzy dynamic programming: Main develop-
ments and applications, Fuzzy Sets and Systems, Vol.81, 31-45, 1996.
[73] Kacprzyk J, Multistage Fuzzy Control, Wiley, Chichester, 1997.
[74] Karnik NN, Mendel JM, and Liang Q, Type-2 fuzzy logic systems, IEEE
Transactions on Fuzzy Systems, Vol.7, No.6, 643-658, 1999.
[75] Karnik NN, Mendel JM, and Liang Q, Centroid of a type-2 fuzzy set, Infor-
mation Sciences, Vol.132, 195-220, 2001.
[76] Kaufmann A, Introduction to the Theory of Fuzzy Subsets, Vol.I, Academic
Press, New York, 1975.
[77] Kaufmann A, and Gupta MM, Introduction to Fuzzy Arithmetic: Theory and
Applications, Van Nostrand Reinhold, New York, 1985.
[78] Kaufmann A, and Gupta MM, Fuzzy Mathematical Models in Engineering
and Management Science, 2nd ed, North-Holland, Amsterdam, 1991.
[79] Ke H, and Liu B, Project scheduling problem with stochastic activity duration
times, Applied Mathematics and Computation, Vol.168, No.1, 342-353, 2005.
[80] Ke H, and Liu B, Project scheduling problem with mixed uncertainty of ran-
domness and fuzziness, European Journal of Operational Research, Vol.183,
No.1, 135-147, 2007.
[81] Klement EP, Puri ML, and Ralescu DA, Limit theorems for fuzzy random
variables, Proceedings of the Royal Society of London Series A, Vol.407, 171-
182, 1986.
[82] Klir GJ, and Folger TA, Fuzzy Sets, Uncertainty, and Information, Prentice-
Hall, Englewood Cliffs, NJ, 1980.
[83] Klir GJ, and Yuan B, Fuzzy Sets and Fuzzy Logic: Theory and Applications,
Prentice-Hall, New Jersey, 1995.
[84] Knopfmacher J, On measures of fuzziness, Journal of Mathematical Analysis
and Applications, Vol.49, 529-534, 1975.
[85] Kosko B, Fuzzy entropy and conditioning, Information Sciences, Vol.40, 165-
174, 1986.
[86] Kruse R, and Meyer KD, Statistics with Vague Data, D. Reidel Publishing
Company, Dordrecht, 1987.
[87] Kwakernaak H, Fuzzy random variables–I: Definitions and theorems, Infor-
[88] Kwakernaak H, Fuzzy random variables–II: Algorithms and examples for the
discrete case, Information Sciences, Vol.17, 253-278, 1979.
256 Bibliography
[89] Lai YJ, and Hwang CL, Fuzzy Multiple Objective Decision Making: Methods
and Applications, Springer-Verlag, New York, 1994.
[90] Lee ES, Fuzzy multiple level programming, Applied Mathematics and Com-
putation, Vol.120, 79-90, 2001.
[91] Lee KH, First Course on Fuzzy Theory and Applications, Springer-Verlag,
Berlin, 2005.
[92] Lertworasirkul S, Fang SC, Joines JA, and Nuttle HLW, Fuzzy data en-
velopment analysis (DEA): a possibility approach, Fuzzy Sets and Systems,
Vol.139, No.2, 379-394, 2003.
[93] Li DY, Cheung DW, and Shi XM, Uncertainty reasoning based on cloud
models in controllers, Computers and Mathematics with Appications, Vol.35,
No.3, 99-123, 1998.
[94] Li DY, and Du Y, Artificial Intelligence with Uncertainty, National Defense
Industry Press, Beijing, 2005
[95] Li HL, and Yu CS, A fuzzy multiobjective program with quasiconcave mem-
bership functions and fuzzy coefficients, Fuzzy Sets and Systems, Vol.109,
No.1, 59-81, 2000.
[96] Li J, Xu J, and Gen M, A class of multiobjective linear programming
model with fuzzy random coefficients, Mathematical and Computer Modelling,
Vol.44, Nos.11-12, 1097-1113, 2006.
[97] Li P, and Liu B, Entropy of credibility distributions for fuzzy variables, IEEE
Transactions on Fuzzy Systems, Vol.16, No.1, 123-129, 2008.
[98] Li SM, Ogura Y, and Nguyen HT, Gaussian processes and martingales for
fuzzy valued random variables with continuous parameter, Information Sci-
ences, Vol.133, 7-21, 2001.
[99] Li SM, Ogura Y, and Kreinovich V, Limit Theorems and Applications of
Set-Valued and Fuzzy Set-Valued Random Variables, Kluwer, Boston, 2002.
[100] Li SQ, Zhao RQ, and Tang WS, Fuzzy random homogeneous Poisson pro-
cess and compound Poisson process, Journal of Information and Computing
Science, Vol.1, No.4, 207-224, 2006.
[101] Li X, and Liu B, The independence of fuzzy variables with applications,
International Journal of Natural Sciences & Technology, Vol.1, No.1, 95-100,
2006.
[102] Li X, and Liu B, A sufficient and necessary condition for credibility measures,
International Journal of Uncertainty, Fuzziness & Knowledge-Based Systems,
Vol.14, No.5, 527-535, 2006.
[103] Li X, and Liu B, New independence definition of fuzzy random variable and
random fuzzy variable, World Journal of Modelling and Simulation, Vol.2,
No.5, 338-342, 2006.
[104] Li X, and Liu B, Maximum entropy principle for fuzzy variables, Interna-
tional Journal of Uncertainty, Fuzziness & Knowledge-Based Systems, Vol.15,
Supp.2, 43-52, 2007.
Bibliography 257
[105] Li X, and Qin ZF, Expected value and variance of geometric Liu process, Far
East Journal of Experimental and Theoretical Artificial Intelligence, Vol.1,
No.2, 127-135, 2008.
[106] Li X, and Liu B, On distance between fuzzy variables, Journal of Intelligent
& Fuzzy Systems, Vol.19, No.3, 197-204, 2008.
[107] Li X, and Liu B, Chance measure for hybrid events with fuzziness and ran-
domness, Soft Computing, Vol.13, No.2, 105-115, 2009.
[108] Li X, and Liu B, Hybrid logic and uncertain logic, Journal of Uncertain
Systems, Vol.3, No.2, 83-94, 2009.
[109] Li X, and Liu B, Foundation of credibilistic logic, Fuzzy Optimization and
Decision Making, to be published.
[110] Li X, and Liu B, Conditional chance measure for hybrid events, Technical
Report, 2008.
[111] Li X, and Liu B, Moment estimation theorems for various types of uncertain
variable, Technical Report, 2008.
[112] Li X, and Liu B, Cross-entropy and generalized entropy for fuzzy variables,
Technical Report, 2008.
[113] Li YM, and Li SJ, A fuzzy sets theoretic approach to approximate spatial
reasoning, IEEE Transactions on Fuzzy Systems, Vol.12, No.6, 745-754, 2004.
[114] Li YM, and Pedrycz W, The equivalence between fuzzy Mealy and fuzzy
Moore machines, Soft Computing, Vol.10, No.10, 953-959, 2006.
[115] Liu B, Dependent-chance goal programming and its genetic algorithm based
approach, Mathematical and Computer Modelling, Vol.24, No.7, 43-52, 1996.
[116] Liu B, and Esogbue AO, Fuzzy criterion set and fuzzy criterion dynamic
programming, Journal of Mathematical Analysis and Applications, Vol.199,
No.1, 293-311, 1996.
[117] Liu B, Dependent-chance programming: A class of stochastic optimization,
Computers & Mathematics with Applications, Vol.34, No.12, 89-104, 1997.
[118] Liu B, and Iwamura K, Modelling stochastic decision systems using
dependent-chance programming, European Journal of Operational Research,
Vol.101, No.1, 193-203, 1997.
[119] Liu B, and Iwamura K, Chance constrained programming with fuzzy param-
eters, Fuzzy Sets and Systems, Vol.94, No.2, 227-237, 1998.
[120] Liu B, and Iwamura K, A note on chance constrained programming with
fuzzy coefficients, Fuzzy Sets and Systems, Vol.100, Nos.1-3, 229-233, 1998.
[121] Liu B, Minimax chance constrained programming models for fuzzy decision
systems, Information Sciences, Vol.112, Nos.1-4, 25-38, 1998.
[122] Liu B, Dependent-chance programming with fuzzy decisions, IEEE Transac-
tions on Fuzzy Systems, Vol.7, No.3, 354-360, 1999.
[123] Liu B, and Esogbue AO, Decision Criteria and Optimal Inventory Processes,
Kluwer, Boston, 1999.
[124] Liu B, Uncertain Programming, Wiley, New York, 1999.
258 Bibliography
[125] Liu B, Dependent-chance programming in fuzzy environments, Fuzzy Sets

and Systems, Vol.109, No.1, 97-106, 2000.
[126] Liu B, Uncertain programming: A unifying optimization theory in various un-
certain environments, Applied Mathematics and Computation, Vol.120, Nos.1-
3, 227-234, 2001.
[127] Liu B, and Iwamura K, Fuzzy programming with fuzzy decisions and fuzzy
simulation-based genetic algorithm, Fuzzy Sets and Systems, Vol.122, No.2,
253-262, 2001.
[128] Liu B, Fuzzy random chance-constrained programming, IEEE Transactions
on Fuzzy Systems, Vol.9, No.5, 713-720, 2001.
[129] Liu B, Fuzzy random dependent-chance programming, IEEE Transactions on
Fuzzy Systems, Vol.9, No.5, 721-726, 2001.
[130] Liu B, Theory and Practice of Uncertain Programming, Physica-Verlag, Hei-
delberg, 2002.
[131] Liu B, Toward fuzzy optimization without mathematical ambiguity, Fuzzy
Optimization and Decision Making, Vol.1, No.1, 43-63, 2002.
[132] Liu B, and Liu YK, Expected value of fuzzy variable and fuzzy expected value
models, IEEE Transactions on Fuzzy Systems, Vol.10, No.4, 445-450, 2002.
[133] Liu B, Random fuzzy dependent-chance programming and its hybrid intelli-
gent algorithm, Information Sciences, Vol.141, Nos.3-4, 259-271, 2002.
[134] Liu B, Inequalities and convergence concepts of fuzzy and rough variables,
Fuzzy Optimization and Decision Making, Vol.2, No.2, 87-100, 2003.
[135] Liu B, Uncertainty Theory, Springer-Verlag, Berlin, 2004.
[136] Liu B, A survey of credibility theory, Fuzzy Optimization and Decision Mak-
ing, Vol.5, No.4, 387-408, 2006.
[137] Liu B, A survey of entropy of fuzzy variables, Journal of Uncertain Systems,
Vol.1, No.1, 4-13, 2007.
[138] Liu B, Uncertainty Theory, 2nd ed., Springer-Verlag, Berlin, 2007.
[139] Liu B, Fuzzy process, hybrid process and uncertain process, Journal of Un-
certain Systems, Vol.2, No.1, 3-16, 2008.
[140] Liu B, Theory and Practice of Uncertain Programming, 2nd ed., Springer-
Verlag, Berlin, 2009.
[141] Liu B, Some research problems in uncertainty theory, Journal of Uncertain
Systems, Vol.3, No.1, 3-10, 2009.
[142] Liu B, Theory and Practice of Uncertain Programming, 3rd ed., http://orsc.
edu.cn/liu/up.pdf.
[143] Liu LZ, and Li YZ, The fuzzy quadratic assignment problem with penalty:
New models and genetic algorithm, Applied Mathematics and Computation,
Vol.174, No.2, 1229-1244, 2006.
[144] Liu SF, and Lin Y, Grey Information: Theory and Practical Applications,
Springer-Verlag, London, 2006.
Bibliography 259
[145] Liu XC, Entropy, distance measure and similarity measure of fuzzy sets and
their relations, Fuzzy Sets and Systems, Vol.52, 305-318, 1992.
[146] Liu XW, Measuring the satisfaction of constraints in fuzzy linear program-
ming, Fuzzy Sets and Systems, Vol.122, No.2, 263-275, 2001.
[147] Liu YH, How to generate uncertain measures, Proceedings of Tenth National
Youth Conference on Information and Management Sciences, August 3-7,
2008, Luoyang, pp.23-26.
[148] Liu YK, and Liu B, Random fuzzy programming with chance measures
defined by fuzzy integrals, Mathematical and Computer Modelling, Vol.36,
Nos.4-5, 509-524, 2002.
[149] Liu YK, and Liu B, Fuzzy random programming problems with multiple
criteria, Asian Information-Science-Life, Vol.1, No.3, 249-256, 2002.
[150] Liu YK, and Liu B, Fuzzy random variables: A scalar expected value opera-
tor, Fuzzy Optimization and Decision Making, Vol.2, No.2, 143-160, 2003.
[151] Liu YK, and Liu B, Expected value operator of random fuzzy variable and
random fuzzy expected value models, International Journal of Uncertainty,
Fuzziness & Knowledge-Based Systems, Vol.11, No.2, 195-215, 2003.
[152] Liu YK, and Liu B, A class of fuzzy random optimization: Expected value
models, Information Sciences, Vol.155, Nos.1-2, 89-102, 2003.
[153] Liu YK, and Liu B, On minimum-risk problems in fuzzy random decision
systems, Computers & Operations Research, Vol.32, No.2, 257-283, 2005.
[154] Liu YK, and Liu B, Fuzzy random programming with equilibrium chance
constraints, Information Sciences, Vol.170, 363-395, 2005.
[155] Liu YK, Fuzzy programming with recourse, International Journal of Uncer-
tainty, Fuzziness & Knowledge-Based Systems, Vol.13, No.4, 381-413, 2005.
[156] Liu YK, Convergent results about the use of fuzzy simulation in fuzzy opti-
mization problems, IEEE Transactions on Fuzzy Systems, Vol.14, No.2, 295-
304, 2006.
[157] Liu YK, and Wang SM, On the properties of credibility critical value func-
tions, Journal of Information and Computing Science, Vol.1, No.4, 195-206,
2006.
[158] Liu YK, and Gao J, The independence of fuzzy variables with applications to
fuzzy random optimization, International Journal of Uncertainty, Fuzziness
& Knowledge-Based Systems, Vol.15, Supp.2, 1-20, 2007.
[159] Loo SG, Measures of fuzziness, Cybernetica, Vol.20, 201-210, 1977.
[160] Lopez-Diaz M, and Ralescu DA, Tools for fuzzy random variables: Embed-
dings and measurabilities, Computational Statistics & Data Analysis, Vol.51,
No.1, 109-114, 2006.
[161] Lu M, On crisp equivalents and solutions of fuzzy programming with different
chance measures, Information: An International Journal, Vol.6, No.2, 125-
133, 2003.
[162] Lucas C, and Araabi BN, Generalization of the Dempster-Shafer Theory:
A fuzzy-valued measure, IEEE Transactions on Fuzzy Systems, Vol.7, No.3,
255-270, 1999.
260 Bibliography
[163] Luhandjula MK, Fuzziness and randomness in an optimization framework,

Fuzzy Sets and Systems, Vol.77, 291-297, 1996.
[164] Luhandjula MK, and Gupta MM, On fuzzy stochastic optimization, Fuzzy
[165] Luhandjula MK, Optimisation under hybrid uncertainty, Fuzzy Sets and Sys-
tems, Vol.146, No.2, 187-203, 2004.
[166] Luhandjula MK, Fuzzy stochastic linear programming: Survey and future
research directions, European Journal of Operational Research, Vol.174, No.3,
1353-1367, 2006.
[167] Maiti MK, and Maiti MA, Fuzzy inventory model with two warehouses under
possibility constraints, Fuzzy Sets and Systems, Vol.157, No.1, 52-73, 2006.
[168] Maleki HR, Tata M, and Mashinchi M, Linear programming with fuzzy vari-
ables, Fuzzy Sets and Systems, Vol.109, No.1, 21-33, 2000.
[169] Merton RC, Theory of rational option pricing, Bell Journal of Economics and
Management Science, Vol.4, 141-183, 1973.
[170] Mizumoto M, and Tanaka K, Some properties of fuzzy sets of type 2, Infor-
mation and Control, Vol.31, 312-340, 1976.
[171] Mohammed W, Chance constrained fuzzy goal programming with right-hand
side uniform random variable coefficients, Fuzzy Sets and Systems, Vol.109,
No.1, 107-110, 2000.
[172] Molchanov IS, Limit Theorems for Unions of Random Closed Sets, Springer-
Verlag, Berlin, 1993.
[173] Nahmias S, Fuzzy variables, Fuzzy Sets and Systems, Vol.1, 97-110, 1978.
[174] Negoita CV, and Ralescu D, On fuzzy optimization, Kybernetes, Vol.6, 193-
195, 1977.
[175] Negoita CV, and Ralescu D, Simulation, Knowledge-based Computing, and
Fuzzy Statistics, Van Nostrand Reinhold, New York, 1987.
[176] Neumaier A, Interval Methods for Systems of Equations, Cambridge Univer-
sity Press, New York, 1990.
[177] Nguyen HT, On conditional possibility distributions, Fuzzy Sets and Systems,
Vol.1, 299-309, 1978.
[178] Nguyen HT, Fuzzy sets and probability, Fuzzy sets and Systems, Vol.90, 129-
132, 1997.
[179] Nguyen HT, Kreinovich V, and Zuo Q, Interval-valued degrees of belief: Ap-
plications of interval computations to expert systems and intelligent control,
Vol.5, 317-358, 1997.
[180] Nguyen HT, Nguyen NT, and Wang TH, On capacity functionals in interval
probabilities, International Journal of Uncertainty, Fuzziness & Knowledge-
Based Systems, Vol.5, 359-377, 1997.
[181] Nguyen HT, Kreinovich V, and Shekhter V, On the possibility of using com-
plex values in fuzzy logic for representing inconsistencies, International Jour-
nal of Intelligent Systems, Vol.13, 683-714, 1998.
Bibliography 261
[182] Nguyen HT, Kreinovich V, and Wu BL, Fuzzy/probability similar to frac-

tal/smooth, International Journal of Uncertainty, Fuzziness & Knowledge-
Based Systems, Vol.7, 363-370, 1999.
[183] Nguyen HT, and Nguyen NT, On Chu spaces in uncertainty analysis, Inter-
national Journal of Intelligent Systems, Vol.15, 425-440, 2000.
[184] Nguyen HT, Some mathematical structures for computational information,
[185] Nguyen VH, Fuzzy stochastic goal programming problems, European Journal
of Operational Research, Vol.176, No.1, 77-86, 2007.
[186] Nilsson N, Probabilistic logic, Artificial Intelligence, Vol.28, 71-87, 1986.
[187] Øksendal B, Stochastic Differential Equations, 6th ed., Springer-Verlag,
Berlin, 2005.
[188] Pal NR, and Pal SK, Object background segmentation using a new definition
of entropy, IEE Proc. E, Vol.136, 284-295, 1989.
[189] Pal NR, and Pal SK, Higher order fuzzy entropy and hybrid entropy of a set,
[190] Pal NR, Bezdek JC, and Hemasinha R, Uncertainty measures for eviden-
tial reasonning I: a review, International Journal of Approximate Reasoning,
Vol.7, 165-183, 1992.
[191] Pal NR, Bezdek JC, and Hemasinha R, Uncertainty measures for evidential
reasonning II: a new measure, International Journal of Approximate Reason-
ing, Vol.8, 1-16, 1993.
[192] Pal NR, and Bezdek JC, Measuring fuzzy uncertainty, IEEE Transactions on
Fuzzy Systems, Vol.2, 107-118, 1994.
[193] Pawlak Z, Rough sets, International Journal of Information and Computer
Sciences, Vol.11, No.5, 341-356, 1982.
[194] Pawlak Z, Rough sets and fuzzy sets, Fuzzy Sets and Systems, Vol.17, 99-102,
1985.
[195] Pawlak Z, Rough Sets: Theoretical Aspects of Reasoning about Data, Kluwer,
Dordrecht, 1991.
[196] Pedrycz W, Optimization schemes for decomposition of fuzzy relations, Fuzzy
[197] Peng J, and Liu B, Stochastic goal programming models for parallel machine
scheduling problems, Asian Information-Science-Life, Vol.1, No.3, 257-266,
2002.
[198] Peng J, and Liu B, Parallel machine scheduling models with fuzzy processing
times, Information Sciences, Vol.166, Nos.1-4, 49-66, 2004.
[199] Peng J, and Liu B, A framework of birandom theory and optimization meth-
ods, Information: An International Journal, Vol.9, No.4, 629-640, 2006.
[200] Peng J, and Zhao XD, Some theoretical aspects of the critical values of biran-
dom variable, Journal of Information and Computing Science, Vol.1, No.4,
225-234, 2006.
262 Bibliography
[201] Peng J, and Liu B, Birandom variables and birandom programming, Com-
puters & Industrial Engineering, Vol.53, No.3, 433-453, 2007.
[202] Peng J, A general stock model for fuzzy markets, Journal of Uncertain Sys-
tems, Vol.2, No.4, 248-254, 2008.
[203] Peng ZX, Some properties of product uncertain measure, http://orsc.edu.cn/
online/081228.pdf.
[204] Puri ML, and Ralescu D, Differentials of fuzzy functions, Journal of Mathe-
matical Analysis and Applications, Vol.91, 552-558, 1983.
[205] Puri ML, and Ralescu D, Fuzzy random variables, Journal of Mathematical
Analysis and Applications, Vol.114, 409-422, 1986.
[206] Qin ZF, and Li X, Option pricing formula for fuzzy financial market, Journal
of Uncertain Systems, Vol.2, No.1, 17-21, 2008.
[207] Qin ZF, A fuzzy control system with application to production planning
problem, Proceedings of the Sixth Annual Conference on Uncertainty, Lu-
oyang, August 3-7, 2008, pp.71-77. (Also available at http://orsc.edu.cn/
process/080412.pdf)
[208] Qin ZF, and Gao X, Fractional Liu process and applications to finance, Pro-
ceedings of the Seventh International Conference on Information and Man-
agement Sciences, Urumchi, August 12-19, 2008, pp.277-280. (Also available
at http://orsc.edu.cn/process/080430.pdf)
[209] Raj PA, and Kumer DN, Ranking alternatives with fuzzy weights using max-
imizing set and minimizing set, Fuzzy Sets and Systems, Vol.105, 365-375,
1999.
[210] Ralescu DA, and Sugeno M, Fuzzy integral representation, Fuzzy Sets and
Systems, Vol.84, No.2, 127-133, 1996.
[211] Ralescu AL, and Ralescu DA, Extensions of fuzzy aggregation, Fuzzy Sets
and systems, Vol.86, No.3, 321-330, 1997.
[212] Ramer A, Conditional possibility measures, International Journal of Cyber-
netics and Systems, Vol.20, 233-247, 1989.
[213] Ramı́k J, Extension principle in fuzzy optimization, Fuzzy Sets and Systems,
Vol.19, 29-35, 1986.
[214] Ramı́k J, and Rommelfanger H, Fuzzy mathematical programming based on
some inequality relations, Fuzzy Sets and Systems, Vol.81, 77-88, 1996.
[215] Saade JJ, Maximization of a function over a fuzzy domain, Fuzzy Sets and
Systems, Vol.62, 55-70, 1994.
[216] Sakawa M, Nishizaki I, and Uemura Y, Interactive fuzzy programming for
multi-level linear programming problems with fuzzy parameters, Fuzzy Sets
and Systems, Vol.109, No.1, 3-19, 2000.
[217] Sakawa M, Nishizaki I, Uemura Y, Interactive fuzzy programming for two-
level linear fractional programming problems with fuzzy parameters, Fuzzy
[218] Shafer G, A Mathematical Theory of Evidence, Princeton University Press,
Princeton, NJ, 1976.
Bibliography 263
[219] Shannon CE, The Mathematical Theory of Communication, The University

of Illinois Press, Urbana, 1949.
[220] Shao Z, and Ji XY, Fuzzy multi-product constraint newsboy problem, Applied
Mathematics and Computation, Vol.180, No.1, 7-15, 2006.
[221] Shih HS, Lai YJ, and Lee ES, Fuzzy approach for multilevel programming
problems, Computers and Operations Research, Vol.23, 73-91, 1996.
[222] Shreve SE, Stochastic Calculus for Finance II: Continuous-Time Models,
Springer, Berlin, 2004.
[223] Slowinski R, and Teghem J, Fuzzy versus stochastic approaches to multicrite-
ria linear programming under uncertainty, Naval Research Logistics, Vol.35,
673-695, 1988.
[224] Slowinski R, and Vanderpooten D, A generalized definition of rough approx-
imations based on similarity, IEEE Transactions on Knowledge and Data
Engineering, Vol.12, No.2, 331-336, 2000.
[225] Steuer RE, Algorithm for linear programming problems with interval objec-
tive function coefficients, Mathematics of Operational Research, Vol.6, 333-
348, 1981.
[226] Sugeno M, Theory of Fuzzy Integrals and its Applications, Ph.D. Dissertation,
Tokyo Institute of Technology, 1974.
[227] Szmidt E, Kacprzyk J, Distances between intuitionistic fuzzy sets, Fuzzy Sets
and Systems, Vol.114, 505-518, 2000.
[228] Szmidt E, Kacprzyk J, Entropy for intuitionistic fuzzy sets, Fuzzy Sets and
Systems, Vol.118, 467-477, 2001.
[229] Tanaka H, and Asai K, Fuzzy linear programming problems with fuzzy num-
bers, Fuzzy Sets and Systems, Vol.13, 1-10, 1984.
[230] Tanaka H, and Asai K, Fuzzy solutions in fuzzy linear programming problems,
IEEE Transactions on Systems, Man and Cybernetics, Vol.14, 325-328, 1984.
[231] Tanaka H, and Guo P, Possibilistic Data Analysis for Operations Research,
Physica-Verlag, Heidelberg, 1999.
[232] Torabi H, Davvaz B, Behboodian J, Fuzzy random events in incomplete prob-
ability models, Journal of Intelligent & Fuzzy Systems, Vol.17, No.2, 183-188,
2006.
[233] Wang G, and Liu B, New theorems for fuzzy sequence convergence, Proceed-
ings of the Second International Conference on Information and Management
Sciences, Chengdu, China, August 24-30, 2003, pp.100-105.
[234] Wang GJ, On the logic foundation of fuzzy reasoning, Information Sciences,
Vol.117, No.1-2, 47-88, 1999.
[235] Wang GJ, and Wang H, Non-fuzzy versions of fuzzy reasoning in classical
logics, Information Sciences, Vol.138, No.1-4, 211-236, 2001.
[236] Wang GJ, Formalized theory of general fuzzy reasoning, Information Sci-
ences, Vol.160, No.1-4, 251-266, 2004.
[237] Wang XZ, Wang YD, and Xu XF, A new approach to fuzzy rule generation:
fuzzy extension matrix, Fuzzy Sets and Systems, Vol.123, No.3, 291-306, 2001.
264 Bibliography
[238] Wang Z, and Klir GJ, Fuzzy Measure Theory, Plenum Press, New York, 1992.
[239] Yager RR, On measures of fuzziness and negation, Part I: Membership in the
unit interval, International Journal of General Systems, Vol.5, 221-229, 1979.
[240] Yager RR, On measures of fuzziness and negation, Part II: Lattices, Infor-
mation and Control, Vol.44, 236-260, 1980.
[241] Yager RR, A procedure for ordering fuzzy subsets of the unit interval, Infor-
[242] Yager RR, Generalized probabilities of fuzzy events from fuzzy belief struc-
tures, Information Sciences, Vol.28, 45-62, 1982.
[243] Yager RR, Entropy and specificity in a mathematical theory of evidence,
[244] Yager RR, On the specificity of a possibility distribution, Fuzzy Sets and
Systems, Vol.50, 279-292, 1992.
[245] Yager RR, Measures of entropy and fuzziness related to aggregation operators,
[246] Yager RR, Modeling uncertainty using partial information, Information Sci-
ences, Vol.121, 271-294, 1999.
[247] Yager RR, On the entropy of fuzzy measures, IEEE Transactions on Fuzzy
Systems, Vol.8, 453-461, 2000.
[248] Yager RR, On the evaluation of uncertain courses of action, Fuzzy Optimiza-
tion and Decision Making, Vol.1, 13-41, 2002.
[249] Yang L, and Liu B, On inequalities and critical values of fuzzy random vari-
able, International Journal of Uncertainty, Fuzziness & Knowledge-Based
Systems, Vol.13, No.2, 163-175, 2005.
[250] Yang L, and Liu B, A sufficient and necessary condition for chance distri-
bution of birandom variable, Information: An International Journal, Vol.9,
No.1, 33-36, 2006.
[251] Yang L, and Liu B, On continuity theorem for characteristic function of fuzzy
variable, Journal of Intelligent & Fuzzy Systems, Vol.17, No.3, 325-332, 2006.
[252] Yang L, and Li K, B-valued fuzzy variable, Soft Computing, Vol.12, No.11,
1081-1088, 2008.
[253] Yang N, and Wen FS, A chance constrained programming approach to trans-
mission system expansion planning, Electric Power Systems Research, Vol.75,
Nos.2-3, 171-177, 2005.
[254] Yazenin AV, On the problem of possibilistic optimization, Fuzzy Sets and
Systems, Vol.81, 133-140, 1996.
[255] You C, Multidimensional Liu process, differential and integral, Proceedings
of the First Intelligent Computing Conference, Lushan, October 10-13, 2007,
pp.153-158. (Also available at http://orsc.edu.cn/process/071015.pdf)
[256] You C, and Wen M, The entropy of fuzzy vectors, Computers & Mathematics
with Applications, Vol.56, No.6, 1626-1633, 2008.
Bibliography 265
[257] You C, Some extensions of Wiener-Liu process and Ito-Liu integral, Proceed-
ings of the Seventh International Conference on Information and Manage-
ment Sciences, Urumchi, August 12-19, 2008, pp.226-232. (Also available at
http://orsc.edu.cn/process/071019.pdf)
[258] You C, Some convergence theorems of uncertain sequences, Mathematical and
Computer Modelling, Vol.49, Nos.3-4, 482-487, 2009.
[259] Zadeh LA, Fuzzy sets, Information and Control, Vol.8, 338-353, 1965.
[260] Zadeh LA, Outline of a new approach to the analysis of complex systems and
decision processes, IEEE Transactions on Systems, Man and Cybernetics,
Vol.3, 28-44, 1973.
[261] Zadeh LA, The concept of a linguistic variable and its application to approx-
imate reasoning, Information Sciences, Vol.8, 199-251, 1975.
[262] Zadeh LA, Fuzzy sets as a basis for a theory of possibility, Fuzzy Sets and
Systems, Vol.1, 3-28, 1978.
[263] Zadeh LA, A theory of approximate reasoning, In: J Hayes, D Michie and
RM Thrall, eds, Mathematical Frontiers of the Social and Policy Sciences,
Westview Press, Boulder, Cororado, 69-129, 1979.
[264] Zhao R and Liu B, Stochastic programming models for general redundancy
optimization problems, IEEE Transactions on Reliability, Vol.52, No.2, 181-
191, 2003.
[265] Zhao R, and Liu B, Renewal process with fuzzy interarrival times and rewards,
Vol.11, No.5, 573-586, 2003.
[266] Zhao R, and Liu B, Redundancy optimization problems with uncertainty
of combining randomness and fuzziness, European Journal of Operational
Research, Vol.157, No.3, 716-735, 2004.
[267] Zhao R, and Liu B, Standby redundancy optimization problems with fuzzy
lifetimes, Computers & Industrial Engineering, Vol.49, No.2, 318-338, 2005.
[268] Zhao R, Tang WS, and Yun HL, Random fuzzy renewal process, European
Journal of Operational Research, Vol.169, No.1, 189-201, 2006.
[269] Zhao R, and Tang WS, Some properties of fuzzy random renewal process,
IEEE Transactions on Fuzzy Systems, Vol.14, No.2, 173-179, 2006.
[270] Zheng Y, and Liu B, Fuzzy vehicle routing model with credibility measure
and its hybrid intelligent algorithm, Applied Mathematics and Computation,
Vol.176, No.2, 673-683, 2006.
[271] Zhou J, and Liu B, New stochastic models for capacitated location-allocation
problem, Computers & Industrial Engineering, Vol.45, No.1, 111-125, 2003.
[272] Zhou J, and Liu B, Analysis and algorithms of bifuzzy systems, International
Journal of Uncertainty, Fuzziness & Knowledge-Based Systems, Vol.12, No.3,
357-376, 2004.
[273] Zhou J, and Liu B, Modeling capacitated location-allocation problem with
fuzzy demands, Computers & Industrial Engineering, Vol.53, No.3, 454-468,
2007.
266 Bibliography
[274] Zhu Y, and Liu B, Continuity theorems and chance distribution of random
fuzzy variable, Proceedings of the Royal Society of London Series A, Vol.460,
2505-2519, 2004.
[275] Zhu Y, and Liu B, Some inequalities of random fuzzy variables with applica-
tion to moment convergence, Computers & Mathematics with Applications,
Vol.50, Nos.5-6, 719-727, 2005.
[276] Zhu Y, and Ji XY, Expected values of functions of fuzzy variables, Journal
of Intelligent & Fuzzy Systems, Vol.17, No.5, 471-478, 2006.
[277] Zhu Y, and Liu B, Fourier spectrum of credibility distribution for fuzzy vari-
ables, International Journal of General Systems, Vol.36, No.1, 111-123, 2007.
[278] Zhu Y, and Liu B, A sufficient and necessary condition for chance distribution
of random fuzzy variables, International Journal of Uncertainty, Fuzziness &
Knowledge-Based Systems, Vol.15, Supp.2, 21-28, 2007.
[279] Zhu Y, Fuzzy optimal control with application to portfolio selection, http://
orsc.edu.cn/process/080117.pdf.
[280] Zimmermann HJ, Fuzzy Set Theory and its Applications, Kluwer Academic
Publishers, Boston, 1985.
List of Frequently Used Symbols
ξ, η, τ random, fuzzy, hybrid, or uncertain variables

ξ, η, τ random, fuzzy, hybrid, or uncertain vectors
µ, ν membership functions
φ, ψ probability, or credibility density functions
Φ, Ψ probability, or credibility distributions
Pr probability measure
Cr credibility measure
Ch chance measure
M uncertain measure
E expected value
V variance
H entropy
(Γ, L, M) uncertainty space
(Ω, A, Pr) probability space
(Θ, P, Cr) credibility space
(Θ, P, Cr) × (Ω, A, Pr) chance space
∅ empty set
< set of real numbers
<n set of n-dimensional real vectors
∨ maximum operator
∧ minimum operator
ai ↑ a a1 ≤ a2 ≤ · · · and ai → a
ai ↓ a a1 ≥ a2 ≥ · · · and ai → a
Ai ↑ A A1 ⊂ A2 ⊂ · · · and A = A1 ∪ A2 ∪ · · ·
Ai ↓ A A1 ⊃ A2 ⊃ · · · and A = A1 ∩ A2 ∩ · · ·
Index
algebra, 229 credibility distribution, 62

arithmetic canonical process, 206 credibility extension theorem, 51
bell-like uncertain variable, 167 credibility inversion theorem, 59
Borel algebra, 230 credibility measure, 45
Borel set, 230 credibility semicontinuity law, 49
Brownian motion, 205 credibility space, 52
canonical process, 205 credibility subadditivity theorem, 47
Cantor function, 234 critical value, 26, 90, 137, 187
Cantor set, 230 distance, 32, 100, 140, 190
chance asymptotic theorem, 124 entropy, 28, 94, 188
chance density function, 131 equipossible fuzzy variable, 61
chance distribution, 130 Euler-Lagrange equation, 238
chance measure, 117 event, 1, 45, 116, 153
chance semicontinuity law, 123 expected value, 14, 73, 132, 176
chance space, 115 exponential distribution, 10
chance subadditivity theorem, 121 exponential membership function, 89
Chebyshev inequality, 33, 102, 141 exponential uncertain variable, 167
conditional chance, 148 extension principle of Zadeh, 72
conditional credibility, 108 first identification function, 164
conditional fuzzy variable, 111 Fubini theorem, 236
conditional random variable, 151 fuzzy inference, 244
conditional membership function, 110 fuzzy random variable, 115
conditional probability, 40 fuzzy sequence, 104
conditional random variable, 41 fuzzy set, 244
conditional uncertain measure, 196 fuzzy variable, 57
conditional uncertain variable, 198 fuzzy vector, 57
connective symbols, 239 geometric canonical process, 207
control system, 216 hazard rate, 43, 112
convergence almost surely, 35, 104 Hölder’s inequality, 34, 102, 141, 192
convergence in chance, 143 hybrid inference, 245
convergence in credibility, 104 hybrid logic, 241
convergence in distribution, 36, 105 hybrid sequence, 143
convergence in mean, 36, 104 hybrid set, 245
convergence in probability, 35 hybrid variable, 124
countable additivity axiom, 1 hybrid vector, 128
countable subadditivity axiom, 153 hypothetical syllogism, 244
credibilistic logic, 240 identification function, 164
credibility asymptotic theorem, 50 independence, 11, 68, 173
credibility density function, 66 inference rule, 226
Index 269
Jensen’s inequality, 35, 103, 142, 193 product credibility theorem, 54

law of contradiction, 154, 221 product probability space, 4
law of excluded middle, 154, 221 product probability theorem, 4
law of truth conservation, 222 product uncertainty space, 161
Lebesgue integral, 234 random fuzzy variable, 115
Lebesgue measure, 231 random sequence, 35
Lebesgue-Stieltjes integral, 236 random variable, 5
Lebesgue-Stieltjes measure, 236 random vector, 5
Lebesgue unit interval, 4 rectangular uncertain variable, 165
Markov inequality, 33, 102, 141, 192 renewal process, 202
matching degree, 226, 245 second identification function, 166
maximality axiom, 46 self-duality axiom, 45, 153
maximum entropy principle, 30, 97 set function, 1
maximum uncertainty principle, 247 σ-algebra, 229
measurable function, 232 simple function, 233
measure inversion theorem, 169 singular function, 234
membership function, 58 step function, 233
Minkowski inequality, 34, 103, 142 stock model, 213
modus ponens, 244 trapezoidal fuzzy variable, 61
modus tollens, 244 trapezoidal uncertain variable, 165
moment, 25, 89, 137, 185 triangular fuzzy variable, 61
monotone convergence theorem, 235 triangular uncertain variable, 165
monotone class theorem, 230 third identification function, 169
monotonicity axiom, 45, 153 truth function, 240
nonnegativity axiom, 1 truth value, 220
normal membership function, 87 uncertain calculus, 205
normal probability distribution, 10 uncertain control, 216
normal uncertain variable, 183 uncertain differential equation, 211
normality axiom, 1, 45, 153 uncertain dynamical system, 217
operational law, 175 uncertain filtering, 216
optimistic value, see critical value uncertain finance, 213
option pricing, 213 uncertain inference, 225
pessimistic value, see critical value uncertain logic, 219
portfolio selection, 215 uncertain measure, 154
power set, 229 uncertain process, 201
probabilistic logic, 240 uncertain sequence, 194
probability continuity theorem, 2 uncertain set, 225
probability density function, 9 uncertain variable, 162
probability distribution, 7 uncertain vector, 162
probability inversion theorem, 9 uncertainty asymptotic theorem, 158
probability measure, 3 uncertainty density function, 172
probability space, 3 uncertainty distribution, 170
product measure axiom, 159 uncertainty space, 158
product credibility axiom, 54 uniform distribution, 10
product credibility space, 55 variance, 23, 86, 135, 183
Axiom 1. (Normality) M{Γ} = 1 for the universal set Γ.
Axiom 2. (Monotonicity) M{Λ1 } ≤ M{Λ2 } whenever Λ1 ⊂ Λ2 .
Axiom 3. (Self-Duality) M{Λ} + M{Λc } = 1 for any event Λ.
Axiom 4. (Countable Subadditivity) For every countable sequence of events
Λ1 , Λ2 , · · · , we have (∞ ) ∞
M M{Λi }.
[ X
Λi ≤
i=1 i=1
Axiom 5. (Product Measure) Let (Γk , Lk , Mk ) be uncertainty spaces for

k = 1, 2, · · · , n. The product uncertain measure is M = M1 ∧ M2 ∧ · · · ∧ Mn ,
i.e.,
min Mk {Λk },

 sup


 Λ1 ×Λ2 ×···×Λn ⊂Λ 1≤k≤n
min Mk {Λk } > 0.5



if sup



Λ1 ×Λ2 ×···×Λn ⊂Λ 1≤k≤n




M{Λ} =  1 − min Mk {Λk },

sup
 Λ1 ×Λ2 ×···×Λn ⊂Λc 1≤k≤n
min Mk {Λk } > 0.5






 if sup
Λ1 ×Λ2 ×···×Λn ⊂Λc 1≤k≤n






0.5, otherwise.
Baoding Liu
Uncertainty Theory
Uncertainty theory is a branch of mathematics based on normality, mono-

tonicity, self-duality, countable subadditivity, and product measure axioms.
The goal of uncertainty theory is to study the behavior of subjective un-
certainty. For this new edition the entire text has been totally rewritten.
More importantly, uncertain process, uncertain calculus, uncertain differen-
tial equation, uncertain logic and uncertain inference are completely new.
This book provides a self-contained, comprehensive and up-to-date presen-
tation of uncertainty theory. The purpose is to equip the readers with an
axiomatic approach to deal with subjective uncertainty. Mathematicians,
researchers, engineers, designers, and students in the field of mathematics,
information science, operations research, industrial engineering, computer
science, artificial intelligence, and management science will find this work a
stimulating and useful reference.

Uncertainty Theory, 3rd Ed., 2009 Liu B.

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Uncertainty Theory, 3rd Ed., 2009 Liu B.

Uploaded by

Copyright:

Available Formats

Uncertainty Theory

Reference to this book should be made as follows:

2.15 Convergence Concepts . . . . . . . . . . . . . . . . . . . . . . 104

3 Chance Theory 115

4 Uncertainty Theory 153

5 Uncertain Process 201

6 Uncertain Calculus 205

7 Uncertain Differential Equation 211

8 Uncertain Logic 219

9 Uncertain Inference 225

A Measure and Integral 229

B Euler-Lagrange Equation 238

E Maximum Uncertainty Principle 247

F Why Uncertain Measure? 248

List of Frequently Used Symbols 267

There are various types of uncertainty. This fact provides a motivation to

The book is suitable for mathematicians, researchers, engineers, design-

A Guide for the Reader

Probability measure is essentially a set function (i.e., a function whose ar-

1.1 Probability Space

Definition 1.1 The set function Pr is called a probability measure if it sat-

Example 1.1: Let Ω = {ω1 , ω2 , · · · }, and let A be the power set of Ω.

Then Pr is a probability measure.

Then for any Borel set A, the set function

is a probability measure on <.

Probability Continuity Theorem

Proof: Step 1: Suppose {Ai } is an increasing sequence. Write Ai → A and

for k = 1, 2, · · · Thus we have

Step 2: If {Ai } is a decreasing sequence, then the sequence {A1 \Ai } is

= Pr{A1 } − lim Pr{Ai }

which implies that Pr{Ai } → Pr{A}.

Since Pr is increasing, we have

It follows from Steps 1 and 2 that Pr{Ai } → Pr{A}.

Example 1.4: Let Ω = {ω1 , ω2 , · · · }, A the power set of Ω, and Pr a

Product Probability Space

Pr{A1 × A2 × · · · × An } = Pr1 {A1 } × Pr2 {A2 } × · · · × Prn {An }

for any Ai ∈ Ai , i = 1, 2, · · · , n. This conclusion is called the product proba-

Pr{Ω} = Pr1 {Ω1 } × Pr2 {Ω2 } × · · · × Prn {Ωn } = 1.

Such a probability measure is called the product probability measure, denoted

Definition 1.3 Let (Ωi , Ai , Pri ), i = 1, 2, · · · , n be probability spaces, and

Infinite Product Probability Space

Pr {A1 × · · · × An × Ωn+1 × Ωn+2 × · · · } = Pr1 {A1 } × · · · × Prn {An }

for any measurable rectangle A1 × · · · × An × Ωn+1 × Ωn+2 × · · · and all

Pr = Pr1 × Pr2 × · · · (1.7)

Definition 1.4 Let (Ωi , Ai , Pri ), i = 1, 2, · · · be probability spaces, and Ω =

1.2 Random Variables

Example 1.8: A deterministic number c may be regarded as a special ran-

Definition 1.6 A random variable ξ is said to be

(e) discrete if there exists a countable sequence {x1 , x2 , · · · } such that

Definition 1.7 Let ξ1 and ξ2 be random variables defined on the probability

Theorem 1.3 The vector (ξ1 , ξ2 , · · · , ξn ) is a random vector if and only if

Proof: Write ξ = (ξ1 , ξ2 , · · · , ξn ). Suppose that ξ is a random vector on the

{ξ1 ∈ B} = {ξ1 ∈ B, ξ2 ∈ <, · · · , ξn ∈ <} = {ξ ∈ B × <n−1 } ∈ A

The vector ξ = (ξ1 , ξ2 , · · · , ξn ) is proved to be a random vector if we can

which implies that B c ∈ B; (iii) if Bi ∈ B for i = 1, 2, · · · , then {ξ ∈ Bi } ∈ A

which implies that ∪i Bi ∈ B. Since the smallest σ-algebra containing all

Definition 1.9 Let f : <n → < be a measurable function, and ξ1 , ξ2 , · · · , ξn

ξ(ω) = f (ξ1 (ω), ξ2 (ω), · · · , ξn (ω)), ∀ω ∈ Ω. (1.12)

Example 1.9: Let ξ1 and ξ2 be random variables on the probability space

(ξ1 + ξ2 )(ω) = ξ1 (ω) + ξ2 (ω), ∀ω ∈ Ω

and their product is

(ξ1 × ξ2 )(ω) = ξ1 (ω) × ξ2 (ω), ∀ω ∈ Ω.

The reader may wonder whether ξ(ω1 , ω2 , · · · , ωn ) defined by (1.9) is a