Page 25 - Vector Analysis
P. 25

§1.5 Determinants                                                                                  21

1. ki = j;

2. for each τ P t1, ¨ ¨ ¨ , i ´ 1u and ι P ti, i + 1, ¨ ¨ ¨ , n ´ 1u,

                      "    στ    if στ ă j ,                        "      σι         if σι ă j ,
                         στ + 1  if στ ě j ,                             σι + 1       if σι ě j .
            kτ     =                          and          kι+1  =

We now determine the sign of εσ1σ2¨¨¨σn´1 and εk1k2¨¨¨kn. Note that if a process of pair inter-
changes of the permutation (σ1, σ2, ¨ ¨ ¨ , σn´1) leads to (1, 2, ¨ ¨ ¨ , n ´ 1), then similar process
of pair interchanges of the permutation (k1, k2, ¨ ¨ ¨ , ki´1, j, ki+1, ¨ ¨ ¨ , kn), by leaving the i-th
slot fixed, leads to the permutation of degree n

            $ (1, 2, ¨ ¨ ¨ , j ´ 1, j + 1, ¨ ¨ ¨ , i ´ 1, j, i, ¨ ¨ ¨ , n) if i ą j,
            ’
            &

               (1, 2, ¨ ¨ ¨ , i ´ 1, j, i, ¨ ¨ ¨ , j ´ 1, j + 1, ¨ ¨ ¨ , n) if i ă j,

            ’                    (1, 2, ¨ ¨ ¨ , n)                             if j = i.
            %

For the case that i ‰ j, another |i ´ j|-times of pair interchanges leads to (1, 2, ¨ ¨ ¨ , n). To
be more precise, suppose that i ą j. We first interchange the (i ´ 2)-th and the (i ´ 1)-th
components, and then interchange that (i ´ 3)-th and the (i ´ 2)-th components, and so on.

After (i ´ j)-times of pair interchanges, we reach (1, 2, ¨ ¨ ¨ , n). Symbolically,

                                                                                                        ò

                                    (1, 2, ¨ ¨ ¨ , j ´ 1, j + 1, ¨ ¨ ¨ , i ´ 1, j , i, ¨ ¨ ¨ , n)

                                                          Ó τ(i ´ 2, i ´ 1)

                                                                                                     ò

                                  (1, 2, ¨ ¨ ¨ , j ´ 1, j + 1, ¨ ¨ ¨ , i ´ 2, j , i ´ 1, ¨ ¨ ¨ , n)

                                                          Ó τ(i ´ 3, i ´ 2)

                                                                                                      ò

                                  (1, 2, ¨ ¨ ¨ , j ´ 1, j + 1, ¨ ¨ ¨ , i ´ 3, j , i ´ 2 ¨ ¨ ¨ , n)

                                                          Ó...
                                                          Ó
                                                  (1, 2, ¨ ¨ ¨ , n).

Similar argument applies to the case i ă j; thus

                      εσ1σ2¨¨¨σn´1 = (´1)|i´j|εk1k2¨¨¨kn = (´1)i+j εk1k2¨¨¨kn .

As a consequence,

            det (A(ˆi, ˆj)) =                                            n´1

                                    ÿ                                    ź

                                                           εσ1σ2¨¨¨σn´1        bτ στ

                                 (σ1,σ2,¨¨¨ ,σn´1)PP(n´1)                τ =1

                         = (´1)i+j            ÿ                                ź

                                                             εk1k2¨¨¨kn               aℓkℓ .       ˝

                                    (k1,¨¨¨ ,kn)PP(n), ki=j                    1ďℓďn
                                                                                 ℓ‰i
   20   21   22   23   24   25   26   27   28   29   30