The inversion of the Laplacian transform Pazy's Book "semigroups of linear operators and applications to Partial differential equations"

51 Views Asked by At

In Pazy's Book page 26, the author gives a proof of Lemma 7.1, the lemma 7.1 says that: Let $B$ be a bounded linear operator. If $\gamma>\|B\|$ then $$e^{tB}=\frac{1}{2\pi i}\int_{\gamma-i\infty}^{\gamma+i\infty}e^{\gamma t}R(\lambda;B)d\lambda$$ The convergence in this formula is in the uniform operator topology and uniformly in t on bounded intervals.

In the proof of this lemma, the author get $$e^{tB}=\frac{1}{2\pi i}\int_{C_{r}}e^{\lambda t}R(\lambda;B)d\lambda$$ where $C_{r}$ be the circle of radius $r$ centered at the origin. Choose $r$ such that $\gamma>r>\|B\|$.

Then the author says that since outside $C_{r}$ the integral of the above formula is analytic and $\|R(\lambda;B)\|\leq C|\lambda|^{-1}$ we can shift the path of integral from $C_{r}$ to the line $Re z=\gamma$, using Cauchy's theorem.

I can just prove this is right when $t\in [\alpha,\beta]$ in which $0<\alpha<\infty,0<\beta<\infty$, that is to say, I can just check pazy's proof is right when $t$ in a compact set away from zero, but can not justify whether $t\in [0,T]$ is right. So, How can I check the "$t\in[0,T]$"? How to shift the path of integral from $C_{r}$ to the line $Re z=\gamma$ when $t\in [0,T],T<\infty$?

Any hints are welcome, Thank you very much!!