Scribd
Upload
46 views

Homework 6 Solutions

The document provides solutions to 7 problems related to graph theory and matching theory. 1. Exhibits a marriage system with more than one stable marriage by defining preferences on the vertices of K2,2. 2. Uses König's theorem to prove Hall's marriage theorem by showing that a vertex cover in a bipartite graph implies a matching. 3. Proves that any nonnegative matrix with equal row and column sums can be written as a sum of permutation matrices using induction on the number of nonzero entries.

Uploaded by

Ashish Gupta
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
46 views

Homework 6 Solutions

The document provides solutions to 7 problems related to graph theory and matching theory. 1. Exhibits a marriage system with more than one stable marriage by defining preferences on the vertices of K2,2. 2. Uses König's theorem to prove Hall's marriage theorem by showing that a vertex cover in a bipartite graph implies a matching. 3. Proves that any nonnegative matrix with equal row and column sums can be written as a sum of permutation matrices using induction on the number of nonzero entries.

Uploaded by

Ashish Gupta
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

Homework 6 Solutions

1. Exhibit a Marriage System which has more than one stable marriage.

Solution: Let ({a, b}, {x, y}) be a bipartition of K2,2 and define a system of preferences as
follows:

a : x>y
b : y>x
x : b>a
y : a>b

Both perfect matchings of this graph are stable marriages: {ax, by} is stable since a and b
are both adjacent to the vertex they most prefer. On the other hand, {ay, bx} is stable since
y and x are both adjacent to the vertex they most prefer.

2. Use Theorem 11.2 (König Egerváry) to prove Theorem 9.2 (Hall’s Marriage Theorem).

Solution: The ”only if” direction is obvious. For the ”if” direction, Let G be bipartite
with bipartition (A, B). If there is a matching of size |A|, then this matching covers A
and we are finished. Otherwise, β(G) = α0 (G) < |A| so we may choose a vertex cover
Y ⊆ V (G) with |Y | < |A|. Since Y is a vertex cover, N (A \ Y ) ⊆ B ∩ Y . But then we have
|A \ Y | = |A| − |Y ∩ A| > |Y ∩ B| ≥ |N (A \ Y )|, which completes the proof. ¤

Recall that a permutation matrix is a square matrix with all entries 0 or 1 in which every
row and every column has exactly one 1.
3. Let Q be an n × n nonnegative real matrix, let t ∈ R, and assume that every row and
every column sum to t. Prove that there exist permutation matrices P1 , . . . , Pk and real
P
numbers x1 , . . . , xk so that Q = ki=1 xi Pi . (Hint: proceed by induction on the number of
nonzero entries in Q, and show that there is a suitable permutation matrix P ).

Solution: We proceed by induction on the number of nonzero entries of Q. As a base,


observe that if Q is the zero matrix, then Q is an empty sum of permutation matrices. For
the inductive step, we may assume that Q has at least one nonzero entry. Let A be the set of
rows of Q and B be the set of columns, and define a simple bipartite graph G with bipartition
(A, B) by the rule that a ∈ A is adjacent to b ∈ B if the (a, b) entry in Q is nonzero. Let
2
X ⊆ A. By construction, every entry in the submatrix QX
B\N (X) is zero, so by summing up
the entries in each row, we see that the sum of the entries in the submatrix QX
N (X) is exactly
equal to t|X|. On the other hand, by summing up the entries in each column, we see that the
sum of the entries in this submatrix is at most t|N (X)| (here we use the assumption that all
entries are nonnegative). It then follows that |N (X)| ≥ |X| (here we use t > 0). Since this
is true for every X ⊆ A, it follows from Hall’s theorem that G has a perfect matching. This
corresponds to a set of n entries from Q with exactly one from each row and column. Let P
be the corresponding permutation matrix, and let x be the smallest value of Q appearing in
one of these entries. Now the matrix Q0 = Q − xP is a nonnegative matrix in which every
row and column sum to t − x which has more zero entries than Q. Thus, by induction, there
P
exist real numbers x1 , . . . xk and permutation matrices P1 , . . . , Pk with Q0 = ki=1 xi Pi , and
P
we then have Q = xP + ki=1 xi Pi as desired.

4. For every k ≥ 1 construct a simple graph in which every vertex has degree 2k which has
no perfect matching.

Solution: K2k+1 is such a graph.

5. For every k ≥ 1 construct a simple graph in which every vertex has degree 2k + 1 which
has no perfect matching. (Hint: start with k = 1 and then generalize).

Solution: For every k ≥ 1 let Hk be the graph obtained from a copy of K2k,2k+1 by adding a
k-edge matching to the partite set of size 2k + 1. Then Hk has 2k vertices of degree 2k + 1
and 1 vertex of degree 2k. Now, form a graph Gk by taking 2k + 1 copies of Hk , then adding
one new vertex, say v, and then adding an edge between v and each vertex of degree 2k in
each copy of Hk . Then Gk is (2k + 1)-regular, but cannot have a perfect matching, since
odd(G − {v}) = 2k + 1 > 1 = |{v}|.

6. Prove that a tree T has a perfect matching if and only if odd(T − v) = 1 for every
v ∈ V (T ).

Solution: If T has a perfect matching, then |V (T )| is even, so T − v must have an odd


number of components for every v. On the other hand, since T has a perfect matching,
odd(T − v) ≤ 1. Thus odd(T − v) = 1 for every v ∈ V (T ).
Next, assume that odd(T − v) = 1 for every v ∈ V (T ) and set M = {e ∈ E(T ) :
odd(T − e) = 2}. Let v ∈ V (T ) and let u be the (unique) vertex in the odd component of
3
T − v which is adjacent to v. It follows that uv ∈ M , and that this is the only edge incident
with v in M . Since this holds for every v, we find that M is a perfect matching.

7. Let G be a simple graph in which every vertex has degree 3. Prove that G has a perfect
matching if and only if there is a decomposition of G into 3-edge paths.

Solution: First suppose that G has a decomposition into three edge paths P1 , . . . , Pk . Let
M be the collection of edges which appear as the middle edge in one of P1 , . . . , Pk . Now
P
2k = 6|E(G)| = 3 v∈V (G) d eg(v) = |V (G)|, so |M | = 21 |V (G)|. Since P1 , . . . , Pk form a
decomposition of G, it follows that no two edges of M share an endpoint. Since |M | =
1
2
|V (G)|, it then follows that M is a perfect matching.
Next suppose that M is a perfect matching of G, and let H = G − M . Then H is
2-regular, so every component of H is a cycle. Assign a ”clockwise” orientation to each cycle
of H. Now for each edge uv ∈ M , complete M to a 3-edge path by adding the two edges
which follow u and v in the clockwise ordering of the cycles in H. This gives a decomposition
of G into 3-edge paths as required.

You might also like