You are on page 1of 3

Math 206 Hamiltonian Cycles and Ores Theorem

Theorem 1 (Ore). (Sufficient Condition.) Given a graph G with n 3 vertices.


Suppose that for each pair of non-adjacent vertices
deg U + deg V n.
Then G has a Hamiltonian cycle.
Proof:
Among all simple paths in G, let P be a simple path of maximal length.
Renumber the vertices, if necessary, so that P = V1 , V2 , V3 , . . . , Vt .
The proof uses the following three ideas:
Idea 1: It is impossible for a simple path to have length t + 1 or greater, because
we stipulated that t was the largest length of all simple paths. So if we find that
a simple path has length t + 1, we have reached a contradiction in the proof.
Idea 2: Any vertex W which is adjacent to V1 must be one of {V2 , V3 , . . . , Vt1 }.
Otherwise W, V1 , V2 , . . . , Vt is a simple path of length t + 1 and we stipulated
that t was the largest length of all simple paths (Idea 1).
A similar argument proves that any vertex W which is adjacent to Vt must be
one of {V2 , V3 , . . . , Vt1 }.
An alternate way of expressing Idea 2 is: if W is not one of the vertices
{V2 , V3 , . . . , Vt1 }, then W cannot be adjacent to V1 (or Vt ).
Idea 3: If U and V are non-adjacent vertices, then there must be some vertex W
which is adjacent to both U and V .
Otherwise each of the remaining n 2 vertices can be adjacent to U or V , but
not both, which implies that deg U + deg V would be n 2, contrary to the
hypothesis.
The proof separates into two cases, depending on whether V1 and Vt are adjacent,
or not.

Case 1: V1 and Vt are adjacent.


Let U = V1 .
If t = n, then we are done; {V1 , V2 , . . . , Vt } is a cycle using all t = n vertices of
G.
Suppose t < n. We will show that this assumption leads to a contradiction.
Since there are n vertices, if t is less than n, then there exists a vertex V not in
{V1 , V2 , . . . , Vt }.
By Idea 2, V cannot be adjacent to U = V1 .
Since U and V are not adjacent, Idea 3 tells us that there must be some vertex
W which is adjacent to both U and V .
Since W is adjacent to U , it follows from Idea 2 that W is one of {V2 , V3 , . . . , Vt1 },
say W = Vj .
But now V, Vj , Vj+1 , Vj+2 , . . . , Vt , V1 , V2 , . . . , Vj1 is a simple path of length t + 1.
This contradicts Idea 1. Since the contradiction arose from the assumption that
t < n, it must be true that t = n, as desired.
Case 2: V1 and Vt are not adjacent.
By Idea 2, any vertex adjacent to V1 is one of {V2 , V3 , . . . , Vt1 }.
So the degree of V1 is at most t 2, since V1 is not incident to Vt or itself.
Now we can tweak Idea 3 to prove:
Idea 4: There is some vertex Vk in the set {V2 , V3 , V4 , . . . , Vt2 } such that Vk is
adjacent to Vt and Vk+1 is adjacent to V1 .
Suppose not. Then there are at least deg (Vt )1 vertices among {V2 , V3 , . . . , Vt2 }
which are adjacent to Vt . (The minus one is because we do not consider Vt1 .) If
for every Vk which is adjacent to Vt , we find that Vk+1 is not adjacent to V1 , then

deg V1 t 2 (deg Vt 1) = t 1 deg Vt .

Adding deg Vt to both sides gives


deg V1 + deg Vt t 1 n 1,
which violates the hypothesis that the sum of the degrees of the non-adjacent
vertices V1 and Vt is greater than or equal to n. This proves Idea 4.
Observe that
P 0 = V1 , V2 , V3 , . . . , Vk , Vt , Vt1 , Vt2 , . . . , Vk+1
is a simple path of maximal length t whose endpoints V1 and Vk+1 are adjacent.
Now follow the reasoning in Case 1 with this path P 0 . 
Theorem 2. (Necessary Condition) For each vertex V of a cycle C, precisely two
edges incident with V are in C. Hence, if H is a Hamiltonian cycle and deg V is
2, then both edges incident with V must be part of H.
Theorem 3. (Necessary Condition) If a graph G contains a bridge, that is, an edge
whose removal disconnects the graph, then G doe not have a Hamiltonian cycle.

You might also like