Page 194 - Determinants and Their Applications in Mathematical Physics
P. 194

5.2 The Generalized Cusick Identities  179

          In particular,

                              2n−i

                       a i,2n =   φ s+i−1 ψ 2n−s ,  1 ≤ i ≤ 2n − 1.  (5.2.4)
                              s=1
          Let A 2n denote the skew-symmetric determinant of order 2n defined as
                                    A 2n = |a ij | 2n ,              (5.2.5)
          where a ij is defined by (5.2.3) for 1 ≤ i ≤ j ≤ 2n and a ji = −a ij , which
          implies a ii =0.
            Let H n and K n denote Hankelians of order n defined as

                                  |h ij | n ,  h ij = φ i+j−1
                           H n =                                     (5.2.6)
                                  |φ m | n , 1 ≤ m ≤ 2n − 1;

                                  |k ij | n ,  k ij = ψ i+j−1
                           K n =                                     (5.2.7)
                                  |ψ m | n , 1 ≤ m ≤ 2n − 1.
          All the elements φ r and ψ r which appear in H n and K n , respectively, also
          appear in a 1,2n and therefore also in A 2n . The principal identity is given
          by the following theorem.
          Theorem 5.1.
                                            2
                                              2
                                    A 2n = H K .
                                            n  n
          However, since
                                             2
                                     A 2n =Pf ,
                                             n
          where Pf n is a Pfaffian (Section 4.3.3), the theorem can be expressed in the
          form
                                    Pf n = H n K n .                 (5.2.8)
          Since Pfaffians are uniquely defined, there is no ambiguity in sign in this
          relation.
            The proof uses the method of induction. It may be verified from (4.3.25)
          and (5.2.3) that

                              Pf 1 = a 12 = φ 1 ψ 1 = H 1 K 1 ,


                    Pf 2 = φ 1 ψ 1  φ 1 ψ 2 + φ 2 ψ 1  φ 1 ψ 3 + φ 2 ψ 2 + φ 3 ψ 1

                                    φ 2 ψ 2      φ 2 ψ 3 + φ 3 ψ 2

                                                    φ 3 ψ 3

                             φ 1       ψ 1
                        =      φ 2      ψ 2
                           φ 2  φ 3  ψ 2  ψ 3

                                                                     (5.2.9)
                        = H 2 K 2
          so that the theorem is known to be true when n = 1 and 2.
   189   190   191   192   193   194   195   196   197   198   199