Integration 489
- [! J(()d( J f(()d( l
- 'Y ((-z-h)n((-z) - 'Y ((-z)n+l
- [! J(()d( J f(()d( l +
'Y (( - z - h)n-1(( - z)2 - 'Y (( - z)n+l ...
[/
!(() d( J f(() d( l
+ 'Y ((-z-h)((-z)n+1 - 'Y ((-z)n+l
- J f(()d(
n (( - z)n+l (7.20-5)
'Y
By the continuity property just proved, each of the expressions in brackets
in (7.20-5) tends to zero as h --t 0. For instance, the difference in the first
bracket can be written as
J
f(()d(/((-z) -1 f(()d(/((-z)
((-z-h)n 'Y ((-z)n
'Y
and since fi(() = f(()/(( -z) is continuous along 'Yi that difference tends
to zero ash --t 0. For the second bracket let h(() = !(()/((-z)2, and so
on. Hence, taking lill1:its in (7.20-5) as h --t O, we get
By letting F 1 (z) = F(z) we obtain, applying (7.20-6) repeatedly,
F'(z) = 1 · F2(z)
F"(z) = 1·F~(z)=1·2 · Fa(z)
F'"(z) = 1·2·F~(z)=1 · 2 · 3 · F 4 (z)
p(n)(z) = 1 · 2 · 3 · · · n · Fn+1(z)
'J f(()d(
=n. ((-z)n+l
'Y
The first formula above is the same as (7.20-1,).
(7.20-6)
(7.20-7)
Theorem 7.29 The function F(z), as well as all its derivatives, tend to
zero as z --t oo in Ro.