by Maarten Behn
Group 5

Exercise 2.1

Compute a maximum matching of the graph given below.
Choose the shortest paths in DM such that you apply at least 3 times the ’shrink’ operation of the algorithm.




Exercise 2.2

Consider the following simple algorithm to compute a matching in a given graph .
We fix some .

  1. Let be the current matching.
  2. If there is a set of edges and a set of edges such that is a matching, update to . Repeat.
    Show the following:

a) The running time of the algorithm is .

Given:

Let
Let be a matching in
Let

The Algorithm written in steps:

Go over all possible
There are different .
Go over all possible
There are different .
Check if is a matching
This takes operations
because we need to check if every new edge has no neigbor in the matching.


because ,
because
because

b) Let M be the matching output by the algorithm and let be a maximum matching in G.

Show that

Lets take
Each connected component is either:

  • a cycle: the number of edges from and are equal.
  • a path: the number differs by 1 if it starts and ends with an edge from , it has one more edge from than from .

Since no local improvement of exchanging edges from with edges from is possible,
the number of edges cannot exceed the number of edges by more than a factor of .



Exercise 2.3

Consider the game Slither by W. Anderson.
Given an undirected, simple connected graph , two players choose in turns an edge e with the following rule:
was not yet chosen and the set of up to now chosen edges (including ) represents a simple path.
The player who cannot choose an edge according to the rules loses.
Show: If contains a perfect matching, then there is winning-strategy for the first player.

Given

Let
Let be a perfect matching in

Observation

If and both end edges are in
no futher edge can be chosen.

Proof

An futher edge can only be choosen if it leads to a vertex not in .
If the vertex would be in , would not be a simple path.
covers all vertecies in , because is perfekt,
therefore there is no edge that leads to a vertex not in .

Winning strategy

The first player always chooses an edge in M.
Therefore the second player must allways choose an edge not in .
The second player will loose after the first player chooses the last edge in
because and both end edges are in .

Assumption

  1. The first player allways choose an edge in .
  2. Both end edges are in when the second player chooses.
  3. P is an alternating path.

Proof

by induction over

Start

The first player chooses an edge in . (1. Assumption)
2. Assumption: both end edges are in
3. Assumption: the path contains only one edge in

Step n’ = n + 1

  • If (the second player chooses)
    The second player must choose an edge (2. Assumption)
    therefore ( is perfekt) and ( is simple).

3. Assumption: The second player extended the end edges whitch are in with an edge that is not in .

  • If (the first player chooses)

    because would need to contain two cosequtive edges that are not in . (3. Assumption)

1. Assumption the first player can choose
2. and 3. Assumption: The player first will extend the end edge that is not .

At every step in the induction all three assumptions are true.
I have also marked this with the arrows

Source

https://doi.org/10.1016/0095-8956(74)90029-X