# Cauculation of a conplex integrand. A question from the book PDE by A. Friedman

In the book Partial Diferential Equations by A. Friedman 1969. Part 2 on page 104, in the proof of theorem 2.1 (d).

A is a operator of type $(\psi,M)$ ($-A$generate a analytic semigroup), and $\Gamma$ is a smooth curve lying in the resolvent set of $-A$, $\lambda\in C$, the integrand $\frac{1}{2\pi i}\int_{\Gamma}(\lambda I+A)^{-1}Ax\frac{d\lambda}{\lambda}=0$, why? How to compute?

Thanks for any hint.

-

I am not sure what $(\psi,M)$ type is, and you cannot assume that all readers of this site have this 1969 book next to them but usually such things hold when the spectrum is bounded, and the curve $\Gamma$ can be continuously deformed to infinity on the resolvent set. The integrand is analytic in $\lambda$, so by Cauchy's theorem the integral does not depend on $\Gamma$, but on the other hand, a trivial estimate shows that the integral tends to $0$ when $\gamma=${ $z:|z|=R$} and $R\to\infty$. Therefore the integral is $0$.
Thank you. I get it. The $\Gamma$ here is not a closed curve, so I can combine it with a part of ball to get a closed curve and then prove the limit of integral on the "ball" is zero. By the wawy, who gave a -1 on my question, so sad. – caoxinru Apr 10 '13 at 9:30
If $\Gamma$ is not closed, it is unlikely that integral is $0$. The argument I wrote assumes that $\Gamma$ is a closed curve (or several closed curves) which surround the spectrum. I don't know who gave you -1, but I added 1 to cancel it, if it made you so upset:-) – Alexandre Eremenko Apr 10 '13 at 12:01