for large values of a, we can obtain an explicit expresion for v(z): 



v(z) = vA (Y) = [gA"^(z) - gA"\(3;ZZ)]]-l, (2.23) 



1 - 2x 



where 



gA(z) = (e^^ - 1)/A, (2.24) 



and, since Qq^z) = "z , then; 



vn(z) = [(z)-^ - [(IIZZ))-1]-1. (2.25) 



1 - 2x 



It is convenient for the remainder of our presentation to introduce 

 the following function: 



v(z,) - v(Zp) 

 6 = 6(Zp z^) = — . (2.26) 



•^v' (z^) v' (Z2) {z^ - f^) 



It can be shown that in the linear case, when v(z) = v (z) the 

 corresponding expressions >^ = A^ and "5 = 6^ become: 



(1 + A/C) - e^^l In (1 + A/C) - AZp ,,- -- .,, 

 YA = L_ . e"^^^r^2' '^ (2.27) 



(1 + A/C) - e^^2 In (1 + A/C) - Az^ 



and 



, = [eA(-Zi-Z2)'2 - e-A(zi-Z2)'2j/,( ) . Shl^U,-Z2)/2l ^ 



^ ^ A(zi-r2)/2 



In general, the extreme values of the conditional probability 



^yi.A = (ya^» yyi,Xy2= ^^a^' ^^^ = ^^S^ ^"^ ^'^^ absolute probability 

 p = (p°) are: 



350 



