Page 626 - Discrete Mathematics and Its Applications
P. 626

9.4 Closures of Relations  605


                                                        W 3 has 1 as its (i, j)th entry if there is a path from v i to v j that has only v 1 = a, v 2 = b,
                                                     and/or v 3 = c as its interior vertices, if any. We now have paths from d to a, namely, d, c, a,
                                                     and from d to d, namely, d, c, d. Hence,

                                                              ⎡           ⎤
                                                               0   0  0  1
                                                               1   0  1  1
                                                              ⎢           ⎥
                                                        W 3 =  ⎢          ⎥ .
                                                              ⎣ 1  0  0  1 ⎦
                                                               1   0  1  1

                                                        Finally, W 4 has 1 as its (i, j)th entry if there is a path from v i to v j that has v 1 = a, v 2 = b,
                                                     v 3 = c, and/or v 4 = d as interior vertices, if any. Because these are all the vertices of the graph,
                                                     this entry is 1 if and only if there is a path from v i to v j . Hence,


                                                              ⎡           ⎤
                                                               10     11
                                                              ⎢ 10    11  ⎥
                                                        W 4 =  ⎢          ⎥  .
                                                               10     11
                                                              ⎣           ⎦
                                                               10     11

                                                     This last matrix, W 4 , is the matrix of the transitive closure.               ▲

                                                        Warshall’s algorithm computes M R by efficiently computing W 0 = M R , W 1 , W 2 ,...,
                                                                                      ∗
                                                     W n = M R . This observation shows that we can compute W k directly from W k−1 : There is a
                                                              ∗
                                                     path from v i to v j with no vertices other than v 1 , v 2 ,..., v k as interior vertices if and only if
                                                     either there is a path from v i to v j with its interior vertices among the first k − 1 vertices in the
                                                     list, or there are paths from v i to v k and from v k to v j that have interior vertices only among the
                                                     first k − 1 vertices in the list. That is, either a path from v i to v j already existed before v k was
                                                     permitted as an interior vertex, or allowing v k as an interior vertex produces a path that goes
                                                     from v i to v k and then from v k to v j . These two cases are shown in Figure 4.
                                                        The first type of path exists if and only if w [k−1]  = 1, and the second type of path exists if
                                                                                            ij
                                                     and only if both w [k−1]  and w [k−1]  are 1. Hence, w [k]  is 1 if and only if either w [k−1]  is 1 or both
                                                                    ik       kj               ij                      ij
                                                     w [k−1]  and w [k−1]  are 1. This gives us Lemma 2.
                                                      ik       kj


                                                     Case 1
                                                      v i                            v j

                                                                All interior vertices
                                                                in {v ,v , . . . , v k–1 }
                                                                   1 2
                                                                   v k


                                                     Case 2



                                                      v i    All interior vertices  v j
                                                                         }
                                                                       k
                                                                1 2
                                                             in {v ,v , . . . , v –1
                                                     FIGURE 4 Adding v k to the Set of
                                                    Allowable Interior Vertices.
   621   622   623   624   625   626   627   628   629   630   631