Directed Graphs
So far, we have been representing instances of the stable marriage problem
simply by describing them. Is there a more efficient way to represent the problem?
It would be particularly useful to find a representation that is amenable to
computer computations. Balinski and Ratier, [BM1]
among others, choose to represent stable marriage problem instances as directed
graphs. Consider an instance of the stable marriage problem with n men and l
women. As was discussed before, it is not necessary that a stable marriage instance
contain the same number of men and women, although in such a situation not every
person will be matched. Let the men and women rank each other, with each man
leaving off his list only women who are so objectionable to him that he would
rather be single than with one of them, and let the women do the same for the
men. Let G be a directed graph with n*l vertices, one corresponding to each
potential couple, with the vertices placed on a n by l grid. So if m, m’
are men and w, w’ are women, G will include vertices (m, w), (m, w’),
(m’, w) and (m’, w’) in the configuration shown in figure
1.
The vertices in figure 1 represent the possible couples. Vertex 1 represents
(m, w), vertex 2 represents (m, w’) and so on.
We can add arcs to this graph to represent the preferences of the stable marriage
participants. A vertical arc (in figure 1 an arc going in either direction between
1 and 3 or 2 and 4, in either direction) represents the preference of the woman
at the head of the column. If, for example, woman w prefers man m to man m’,
we will draw the arc arc (3, 1) in the picture in figure 1. Similarly, horizontal
arcs represent the preferences of the men. In a stable marriage instance with
more than two participants of each gender, there will be an arc between every
pair of ranked men in any woman’s column, and similarly for every man’s
row. The majority of these arcs are superfluous, however, since preference is
transitive. So we can delete any arc implied by transitivity, and determine
the preference order by examining the path from the least to most preferred
in each column or row. A vertex may have a degree as high as four. A vertex
with higher indegree than outdegree represents a couple that is popular with
both genders, while one with higher outdegree than indegree represents a couple
that is less popular. Figure 2, from Balinski and Ratier in 1998, shows the
directed graph for a stable marriage problem instance with 7 men and 7 women.
I’ve added a possible preference list for clarity. Note that not every
possible couple has a vertex in the graph. The vertex can be omitted if either
the man in that row or the woman in that column did not rate the other. No pair
in which one participant finds the other so undesirable that he or she would
rather be single could possibly be part of a stable marriage, so it does us
no good to consider it.
Preferences:
man | m1 | m2 | m3 | m4 | m5 | m6 | m7 |
1st choice | w7 | w1 | w1 | w6 | w5 | w4 | w6 |
2nd choice | w6 | w3 | w3 | w5 | w3 | w1 | w5 |
3rd choice | w4 | w4 | w4 | w4 | w2 | w2 | w2 |
4th choice | w3 | w5 | w2 | w2 | w1 | w3 | w1 |
5th choice | w5 | w6 | w5 | w1 | w6 | ||
6th choice | w2 | w6 | w5 | ||||
7th choice | w7 |
woman | w1 | w2 | w3 | w4 | w5 | w6 | w7 |
1st choice | m7 | m1 | m6 | m4 | m2 | m1 | m3 |
2nd choice | m6 | m4 | m5 | m2 | m3 | m2 | m1 |
3rd choice | m5 | m5 | m3 | m1 | m4 | m3 | |
4th choice | m4 | m6 | m1 | m5 | m4 | ||
5th choice | m3 | m7 | m6 | m7 | |||
6th choice | m2 | m7 | |||||
7th choice |
[BM1]
As demonstrated on the next page, the graph constructed
from a stable marriage instance allows us to use the tools of graph theory to
search for proofs and algorithms relating to the stable marriage problem.