I have defined a continuous functional calculus on the bounded self-adjoint linear operators for functions continuous on the spectrum, and are defined on an interval. How do I deal with wanting to define $f(T)$ for functions defined on the whole real line? Can I simply extend it without any technical arguments?

  • 2
    $\begingroup$ The point is of course that you can uniformly approximate $f$ on the spectrum of $T$ by polynomials $p_n$ and put $f(T) = \lim p_n (T)$. This expression doesn't bother whether $f$ is defined on a larger subset of $f$ than the spectrum of $T$ as mac points out below. I strongly recommend having a look at the proof of the spectral theorem, for instance in Pedersen, Analysis now, section 4.4. $\endgroup$ – t.b. Jun 14 '11 at 10:45

If $f:\mathbb{R}\to\mathbb{C}$ is continuous and $T$ is bounded and self-adjoint, then $\sigma(T)$, the spectrum of $T$, is a compact subset of $\mathbb{R}$. So $g=f|_{\sigma(T)}$, the restriction of $f$ to $\sigma(T)$, is continuous, and you've already defined $g(T)$.

It's natural enough to simply define $f(T)$ to be $g(T)$. The map of "evaluation at $T$" will then be a nice homomorphism from the continuous functions $\mathbb{R}\to\mathbb{C}$ into the bounded linear operators, which is essentially what you want from a functional calculus.


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.