Confusion regarding steps in bipartite matching proof

187 Views Asked by At

Can someone please explain how it follows that $|N(S)|x \geq |S|x$? What I'm asking is why is it necessary to use the value of x to derive the inequality?

Theorem 5.2.7. Let G be a bipartite graph with vertex partition L, R where $|L| \leq |R|$. If G is degree-constrained, then there is a matching that covers L.

Proof.

The proof is by contradiction. Suppose that G is degree constrained but that there is no matching that covers L. By Theorem 5.2.5, this means that there must be a bottleneck $S \subseteq L$. Let x be a value such that $deg(l) \geq x \geq deg(r)$ for every $l \epsilon L$ and $r \epsilon R$. Since every edge incident to a node in S is incident to a node in N(S), we know that \begin{aligned} |N(S)|x \geq |S|x \end{aligned} and thus that \begin{aligned} |N(S)| \geq |S| \end{aligned} This means that S is not a bottleneck, which is a contradiction. Hence G has a matching that covers L.

EDIT:

For clarification (hopefully), this theorem builds on the definition of a bottleneck where:

\begin{aligned} G \ is \ the \ bipartite \ graph \ (V,E) \end{aligned}

So the graph is partitioned in to L and R, where:

\begin{aligned} S \subseteq L \end{aligned}

\begin{aligned} N(S) = \{r \ | \ \{s,r\} \ \epsilon \ E \ for \ some \ s \ \epsilon \ S \} \end{aligned}

Meaning that N(S) is a neighbor of some vertex in s, and:

\begin{aligned} |S| \geq |N(S)| \end{aligned}

implies that a bottleneck exists.

1

There are 1 best solutions below

1
On BEST ANSWER

It's kind of subtle but everything is in 'Since every edge incident to a node in $S$ is incident to a node in $N(S)$'.

Here's how you can see this. Let $e(S)$ and $e(N(S))$ respectively denote the number of edges incident to some vertex of $S$ and $N(S)$.

Then, $e(N(S)) \geq e(S)$, since every edge counted in $e(S)$ has to be counted in $e(N(S))$ (as every edge with an endpoint in $S$ has an its other endpoint in $N(S)$.

We have $|S|x \leq e(S)$ since $x \leq deg(l)$ for all $l \in S$.

And we have $|N(S)|x \geq e(N(S))$ since $x \geq deg(r)$ for all $r \in N(S)$.

Putting all this together in the correct order, you get the inequality chain

$|S|x \leq e(S) \leq e(N(S)) \leq |N(S)|x$.