next up previous contents
Next: Appendix E. Analytic continuation: Up: Appendix D. Commuting linear Previous: Appendix D.1. Commuting two

 

Appendix D.2. Commuting integrals and derivatives

Consider differentiating the integral tex2html_wrap_inline17453 with respect to t Theorem D.2 generalizes theorem 9.42 of [73] and establishes conditions general enough to allow the commutation of the derivative and integral for the functions appearing in this paper. Define tex2html_wrap_inline17457 to be the partial derivative of F with respect to its second argument, evaluated at (x,t).

: If

(1) F(x,t) and tex2html_wrap_inline17465 are defined for tex2html_wrap_inline17467, where tex2html_wrap_inline17469, and where tex2html_wrap_inline17471 is convex,

(2) tex2html_wrap_inline17473 exists tex2html_wrap_inline17475,

(3) tex2html_wrap_inline17477 and b>0, tex2html_wrap_inline17481 with f(x)>0 for tex2html_wrap_inline17485, and tex2html_wrap_inline17487 such that tex2html_wrap_inline17489 and tex2html_wrap_inline17491, tex2html_wrap_inline17493, tex2html_wrap_inline17495,

then tex2html_wrap_inline17497 on tex2html_wrap_inline17499.

Proof: Let tex2html_wrap_inline17501 for tex2html_wrap_inline17503. By (1) and the mean value theorem, tex2html_wrap_inline17505 with tex2html_wrap_inline17507, tex2html_wrap_inline17509 such that tex2html_wrap_inline17511. Using this and (3) we have that for any tex2html_wrap_inline17513, tex2html_wrap_inline17515, tex2html_wrap_inline17517 and a nowhere-negative (in tex2html_wrap_inline17519) f(x) obeying tex2html_wrap_inline17489 such that if tex2html_wrap_inline17525 and tex2html_wrap_inline17527, then tex2html_wrap_inline17529. From this and (2) it follows that for all b>0, tex2html_wrap_inline17517, and a nowhere-negative (in tex2html_wrap_inline17519) f(x) obeying tex2html_wrap_inline17539 such that if tex2html_wrap_inline17525, then tex2html_wrap_inline17543. Taking the limit tex2html_wrap_inline17545, noting that tex2html_wrap_inline17547, and finally taking tex2html_wrap_inline17549 with tex2html_wrap_inline17551, we arrive at the desired result. QED.

The functions F(x,t) of interest have the form tex2html_wrap_inline17555 with Re(t)>-1 and c>0. For these functions it may be shown that the conditions of theorem D.2 hold.



David Wolf
Tue Mar 25 08:11:49 CST 1997