Professional Documents
Culture Documents
Ax
In special situations, when A operates on x, we get a scaled
version of x back again:
x
So we have Ax = x
When Ax = x
"Eigenvalue"
x "Eigenvector"
For an n x n matrix A, there are:
ll nneigenvalues,
eigenvalues,some
someofofwhich
whichmay
maybe
be
complex
complexand/or
and/orrepeated.
repeated.
ll At
Atleast
leastone
oneeigenvector
eigenvectorcorresponding
corresponding
to
toeach
eachdistinct
distincteigenvalue.
eigenvalue. These
Thesewill
will
be
becomplex
complexififthe
theeigenvalues
eigenvaluesare
are
complex.
complex.
ll Sometimes
Sometimesrepeated
repeatedeigenvalues
eigenvalueshave
have
associated
associatedwith
withthem
themgeneralized
generalized
eigenvectors.
eigenvectors.
How to find eigenvalues and eigenvectors:
Ax = x
( A I ) x = 0
2 0 2 How would
( A I ) = = 5 x1 = ( A 5 I ) x1 = 0 2 2 x1 = 0 you solve
1 this?
2 2 4
F
Using your favorite method; e.g., Gaussian elimination,
echelon forms, etc., we can get:
1
x1 = 1
1
c
x1 = c
c
Better yet, we normalize all eigenvectors so that they all
have length 1:
1 3
(corres. to 1 = 5 )
x1 = 1 3
1 3
1 6
x3 = 1 6 (corres. to 3 = 1)
2 6
Pause here to explore an application for the
eigenvalues/vectors: Consider the previous example.
Note that the eigenvectors are linearly independent.
They therefore form a basis for a 3-D vector space.
no input
Suppose we had a (homogeneous) system:
x& = Ax
x = Mx
where M is a matrix whose columns are the new basis
vectors.
When M is formed by columns that are eigenvectors, it is
called a modal matrix.
F
x& = M 1 AM x
1 6
1
6
2
6
3 0 2 1
6
1
2
1
3
M 1 AM = 1 1 0 0 3 2 1 6 1 1 3
2 2 2
1 3
1
3
1
3
2 2 1
2
6
0 1
3
(happens to be orthonormal)
1 0 0
= 0 3 0
0 0 5
= A These are the eigenvalues of A!
x&1 x1 1 0 0 x1 x1
x& = A x = 0 3 0 x = 3 x
2 2 2 2
x& 3 x3 0 0 5 x3 5 x3
A B
x& = Ax + Bu x = Mx x& = M 1 AM x + M 1 Bu
y = Cx + Du x& = Mx& y = CM x + Du
C D
Something different happens when we have one or more
eigenvalues that are repeated (multiple roots) of the
characteristic equation.
1 2 0
A = 0 1 0
3 3 5
0 = A I = (1 ) 2 (5 ) So 1 = 5, 2 = 3 = 1
0 2 0 a
( A I ) x = 0 0 0 x = 0 x2 = 0
3 3 4 34 a
If you can find xi+1 's that are linearly independent of all
previous vectors xi , then the new vectors are
generalized eigenvectors.
If the xi+1's are not linearly independent of previously
found vectors, continue on by solving:
( A i I ) xi + 2 = xi +1
0 2 0 4 5
Solve: 0 0 0 x = 0 x2 = 2
2
3 3 4 3 3
This vector is linearly independent of the previous
eigenvectors, so it is a generalized eigenvector.
Now form the modal matrix using the two regular
eigenvectors and the generalized eigenvector:
Put this in the
0 4 5 order you found
M = 0 0 2 them
1 3 3
generalized
regular
Compute similarity transformation:
3 4 3
8 1 1 2 0 0 4 5 5 0 0
A = M 1 AM = 1 4 5 0 0 1 0 0 0 2 = 0 1 1
8
0 1
2 0 3 3 5 1 3 3 0 0 1
5 0 0
A = 0 1 1
0 0 1
1 0 1
A I = 0 1 0 = (1 ) 2 ( 2 ) = 0
0 0 2
1 0 1 1
( A 2I )x = 0 1 0 x = 0 x1 = 0
0 0 0 1
For = 1:
0 0 1 1 0
( A I ) x = 0 0 0 x = 0 x 2 = 0, x3 = 1
0 0 1 0 0
( A i I ) i x = 0
( A i I ) i 1 x 0
1 1
denote these solutions v 1 , K , v m i . There will
be no more than m i of them (why?).
Rank ( A i I ) i = n mi
# solution = n (n mi )
Now compute a different chain of generalized
eigenvectors for each j = 1, K , mi :
( A i I )v1j = v 2j
( A i I ) v 2j = v 3j chain of generalized
M
eigenvectors
1
( A i I )v j i = vji
( A i I )v j i = 0 REGULAR eigenvector
ending each chain
( A i I )i 1 x = 0
( A i I )i 2 x 0 etc.
EXAMPLE:
1 2 0
A = 0 1 0 1 = 2 = 1 3 = 5
3 3 5
n=3 m1 = 2 n m1 = 1
Consider =1
0 2 0
( A I ) = 0 0 0 r( A I ) = 2
3 3 4
0 0 0
( A I )2 = 0 0 0 r( A I ) 2 = 1
12 6 16
Index of = 1 is 2.
So solve ( A I )2 x1 = 0
( A I ) x1 0
1
to get x1 = 2 (generalized)
0
4
Now generating the chain: x 2 = ( A I ) x1 = 0
3
( A I ) x2 = 0
Note that if there are other linearly independent
solutions to
( A I )2 x = 0
T
Finally, for this example, we note that x = 0 0 1
is the regular eigenvector corresponding to 3 = 5 ,
so we get:
reverse order
0 4 1 5 0 0
M = 0 0 2 A = M 1 AM = 0 1 1
1 3 0 0 0 1
generalized
regular
ANOTHER EXAMPLE:
0 0 1 0
0 0 0 1
A= 4 = 0, m1 = 4, n m1 = 0
0 0 0 0
0 0 0 0
0 0 1 0
0 0 0 1
r( A 0 I ) = r =2
0 0 0 0
0 0 0 0
0
0 0 0
i= 2 =
0 0 0 0
r ( A 0 I )2 =r =0 longest chain
0 0 0 0
of eigenvectors
0 0 0 0
Find 2 linearly independent solutions to
( A I )2 x1 = ( A)2 x1 = 0
( A I ) x1 = ( A) x1 0
try x1 = [1 0 0 0 ] :
T
( A 0 I )x1 = 0 D No
same for x1 = 0 1 0 0
T
D No
try 0 0 1
0 0 0
x1 = : ( A 0 I ) =
1 1 0
0 0 0
generalized regular
0 0 0
and also: 0 0 1
x1 = : ( A 0 I ) =
0 0 0
1 1 0
generalized regular
so now,
1 0 0 0 0 1 0 0
0 0 1 0 0 0 0 0
M = , A = M 1 AM =
0 1 0 0 0 0 0 1
0 0 0 1 0 0 0 0
reg reg
gen gen
III. "Adjoint" method. This method requires computation
of the adjoint of A and must be done "by hand." It is
relatively tedious to do, but there is an example in
Brogan (page 257).
has A I = ( 2 )5
So it has eigenvalues: 1 = 2, algebraic multiplicity 5
2 = 0, algebraic multiplicity 1
NOTE:
NOTE: The The AAmatrix
matrix
itself
itselfhas
hasrank
rank
If we compute the rank of deficiency
deficiencyequal
equaltoto
A 2I 1 = 2 the
thegeometric
geometric
multiplicity
multiplicityof
ofany
any
we get 4, for a rank deficiency zero
zeroeigenvalues.
eigenvalues.
of 6-4=2. (Why?)
gm = 2 (Why?)
Because the
Therefore, in the Jordan canonical form for this
column
matrix, we will have 2 Jordan blocks for the of the modal
eigenvalue = 2 (and of course one trivial block matrix = 0
corresponding to = 0 ). We can calculate 2
regular eigenvectors and will need 3 generalized
eigenvectors. F
When we go through the exercise of finding the Jordan
form, we get:
2 1 0 0 0 0
0 2 1 0 0 0
0 0 2 0 0 0
0 0 0 2 1 0
0 0 0 0 2 0
0 0 0 0 0 0
How could you tell there is a 3x3 block and a 2x2 block, rather
than a 4x4 and a 1x1? Number of regular eigenvectors
Recall that the index of the eigenvalue is the smallest integer i
such that rank ( A i I ) i = n mi .
For this matrix and eigenvalue i , i = 3 , and this will be the
size of the largest Jordan block associated with eigenvalue i .
A Geometric Interpretation of All This Stuff:
qi qi qi qi
Ax = A ai ei = ai ( Aei ) = ai ( i ei ) = ( ai i ) ei N i
i =1 i =1 i =1 i =1
e
2
x
Ax
Plane
Plane(subspace)
(subspace)formed
formedby
by IfIfaavector
vectorxxstarts
startsout
outininthe
the
eigenvectors
eigenvectorsofofAA subspace,
subspace,ititstays
staysininthe
thesubspace
subspace
when
whenAAacts
actson
onit.
it.
Chapter 5: Functions of Vectors and Matrices
y , Ax (= y T Ax ): "Bilinear Form"
x, Ax (= x T Ax ) : "Quadratic Form"
x Ax = x Ax
T
( T
)
T
= x T AT x ,
(
) +
T
1 A A
x T Ax = x T Ax + x T AT x = x T x
2 2
any quadratic form can be written as a quadratic form with
a symmetric A-matrix. We therefore treat all quadratic
forms as is they contained symmetric matrices.
DEFINITIONS: Let Q = x T Ax
F
eigenvalues i :
If the real parts of eigenvalues i
Matrix A is . . . of A are: . . . .