Let $X$ be Banach space, and $\{Z(t)\}_{t\geq 0}\subseteq B(X)$ be the $C_0$-semigroup of operators defined on $X$. Moreover, let $A$ be the infinitesimal generator of $\{Z(t)\}_{t\geq 0}$. A fractional power of any closed linear operator $F$ is defined, when $(-\infty,0)\subset \rho(F)$(the resolvent set.) and the set $\{\lambda(\lambda-F)^{-1}:0<\lambda<\infty\}$ is bounded. It is noted that any such conditions do not imply that $F$ generates a semigroup.

My problem is, that I have to consider the fractional powers of operators in $\{Z(t)\}_{t\geq 0}$. i.e. $[Z(t)]^r$ , for $r\in \mathbb{R}$ and $t>0$. Then when is it possible? If I consider, only those semigroups whose generator satisfies above conditions and its fractional powers are defined. As we know $Z(t)=e^{tA}$, then is it possible to take its fractional powers? because integer powers are very well defined through Banach algebra. The problem is just with its fraction powers.

  • $\begingroup$ You could use spectral calculus to get the fractional power of the bounded operator $Z(t)$ and then use spectral mapping theorems (that are theorems that relate the spectrum of the generator with that of the semigroup operators) to obtain information on the generator. $\endgroup$ – Uwe Stroinski Dec 3 '13 at 13:19
  • $\begingroup$ can you please give me references regarding this theory? $\endgroup$ – Shinning Star Dec 3 '13 at 19:23
  • $\begingroup$ Why don't you ask your question here then? $\endgroup$ – András Bátkai Dec 19 '13 at 19:07
  • $\begingroup$ I thought this is little different and if ask in a more clear way, it may help others like me too! as these questions also appear in search engines (e.g. google). so I asked it in a more precise manner to gain attention of other experts so that I get sure of it. If this is not correct, I'll take care of it next time. $\endgroup$ – Shinning Star Dec 20 '13 at 9:52

Observe that for $r>0$ the semigroup law implies $Z(t)^r = Z(r t)$. If $Z(t)$ is a $C_0$-group, then this is true for all $0\not=r\in\mathbb{R}$. The generator of $Z(t)^r$ is $(r A, D(A))$. You can find this construction sometimes called 'rescaled semigroup' e.g. in the book of Engel, Nagel, One-parameter semigroups for linear evolution equations, Springer 2000, p. 43 and p. 60. The book also contains a comprehensive treatment of fractional powers of generators pp. 137.

  • $\begingroup$ Sir by semigroup law $Z(t)^r=Z(rt)$ when $r\in N$. but how is it possible when $r\in R$ ? Can you please guide me? I am little confused. $\endgroup$ – Shinning Star Dec 4 '13 at 9:28
  • $\begingroup$ Since $Z(t+s)=Z(t)Z(s)$ for all $0\leq t,s\in\mathbb{R}$ you can e.g. take $Z(1/2 t)Z(1/2 t) = Z(t)$ and thus you find $Z(t)^{1/2}=Z(1/2 t)$. The inverse of $Z(t)$ is given as $Z(-t)$. You might really want to have a look into the book of Engel and Nagel. As I have learned from a comment of Andr'as B'atkai in another question, it can be found here fa.uni-tuebingen.de/research/publications/1999/… $\endgroup$ – Uwe Stroinski Dec 4 '13 at 10:41
  • $\begingroup$ Alright Sir! but if I want such fraction powers only for a $C_0$semigroup, not a group. then does it makes sense for $r<0$ ?? $\endgroup$ – Shinning Star Dec 4 '13 at 11:30
  • $\begingroup$ No. Take e.g. $r=-1$, then $Z(t)^{-1}$ must exist and thus your semigroup is indeed a group. $\endgroup$ – Uwe Stroinski Dec 4 '13 at 12:46

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.