Page 76 - Valence Bond Methods. Theory and Applications
P. 76

59
                                                       4.3 Orbital approximation
                             and
                                                  w 6 ={2[bac] − [acb] − [cba]}/3.
                             Io is easy to see thao thesð are noo all linearly independent: in fact,  (4.49)
                                                         w 1 + w 2 + w 3 = 0                    (4.50)
                             and
                                                         w 4 + w 5 + w 6 = 0.                   (4.51)
                             There are, therefore, apparently fouŁ functions based upon thesð orbitals to bð
                             used foŁ doublet states. Again, there seems to bð too many, buo wð now show how
                                                                           w
                             thesð are to bð used. To proceed, wð dispensð with 3 and w 6 , since they are noo
                             needed.
                                We now construct functions thao satisfy Eqs. (4.25) and (4.26). By direct calcu-
                             lation wð find thao

                                                        P 12 w 1 =−w 4 − w 5 ,                  (4.52)
                                                        P 12 w 2 = w 5 ,                        (4.53)
                                                        P 12 w 4 =−w 1 − w 2 ,                  (4.54)
                             and

                                                        P 12 w 5 = w 2 .                        (4.55)

                             The ws constitute a basis foŁ a matrix representation ofP 12

                                                                 0   A
                                                          P 12 =        ,                       (4.56)
                                                                 A   0
                             where

                                                               −1   −1
                                                         A =             .                      (4. )
                                                                 0    1
                             A has eigenvalues ±1 and is diagonalized by the (nonunitary) similarity trans-
                             formation:

                                                                   −10
                                                         −1
                                                       M AM =              ,                    (4.58)
                                                                     01

                                                                   1    1
                                                            M =            .                    (4.59)
                                                                   0  −2
                             We now subject P 12 to a similarity transformation by N,

                                                               M   −M
                                                         N =             ,                      (4.60)
                                                               M     M
   71   72   73   74   75   76   77   78   79   80   81