MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Is $e^{(rA)} = (e^{A})^r$ when $r \in \mathbb{R}$ and $A$ is an element of a Banach algebra?

Clearly if $n$ is an integer, then

$e^{(nA)} = e^{A+A \cdots +A} = e^{A}e^{A}\cdots e^{A} = (e^{A})^n$,

where the second equality follows from the Baker-Hausdorff lemma and the fact that [A,A]=0. On the other hand, I think the equality is not generally true when $r \in \mathbb{C}$. But what about the reals?

Many thanks for your thoughts!

share|cite|improve this question
How do you define the non-integer powers for arbitrary Banach algebra elements? – fedja Dec 24 '10 at 15:35
Using the binomial series, I guess? – Mariano Suárez-Alvarez Dec 24 '10 at 15:36
The conventional definition is to use the limit of (1 + x/n)^n as n goes to infinity. (There's a slightly different variants that's more likely to converge, but I can't remember what it is.) – arsmath Dec 24 '10 at 16:09
The Taylor expansion of the logarithm diverges more often than not. There is no reason to expect it to converge for $e^A$. Of course, if everything is defined as a series and you can meaningfully plug series into series justifying all passages to the limits, the identity holds. On the other hand, if the right hand side just fails to exist or has multiple values like non-integer powers of complex numbers (the first example to look at), the question hardly makes sense. – fedja Dec 24 '10 at 16:59
You are asking why $(f\circ g)(A)= f(g(A))$. The answer is, it is just a matter of functional calculus; check any textbook on the topic. – Pietro Majer Dec 24 '10 at 17:38
up vote 4 down vote accepted

A particular case of a Banach algebra is the one-dimensional case $\mathbb C$, right?
And according to Pietro, if you can prove this in that case, you get the general case by saying "functional calculus". (At least for $A$ where it applies, such as normal $A$.)
But is it true for $\mathbb C$?

$A = 2\pi i$, $r=1/2$, $1 = \exp(2\pi i)$ and $1^{1/2} =^{?} \exp(\pi i) = -1$ .

share|cite|improve this answer
Oh dear -- you're of course correct, the claim is not generally true. Thanks! – soulphysics Dec 26 '10 at 17:21

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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