1
$\begingroup$

I apologize in advance for the rather vague question.

While reading the book White noise distribution theory by H.H. Kuo, in particular the section 13.3 I came across the following statement (I'll paraphrase it)

Since the Hida space $(S)^*$ is not Banach we cannot simply use the standard definition of the Bochner integral [...]. However note that $(S)^*=\bigcup_{p\geq 0} (S_{-p})^*$ and each $(S_{-p})^*$ is a separable Hilbert space. Then it's reasonable to say that a function $\Phi:M\to (S)^*$ is Bochner integrable if:

  1. $\Phi$ is weakly measurable.
  2. $\exists p\geq 0$ such that $\Phi(u)\in (S_{-p})^*$ for a.a. $u\in M$ and $\|\Phi(\cdot)\|_{-p}\in L^1(M)$.

Since the space of tempered distributions $S'(\mathbb R)$ can also be constructed in a similar fashion $S'(\mathbb R)=\bigcup_{p\geq 0} S_{-p}(\mathbb R)$ where $S_{-p}(\mathbb R)$ is the completition of $L^2(\mathbb R)$ with respect to the norm $|f|_{-p}:=\|A^{-p}f\|$ and $A$ is the Hamiltonian of the harmonic oscillator.

Can we extend the definition of Bochner integrals to functions taking values in the space of tempered distributions?

$\endgroup$
2
  • 6
    $\begingroup$ The Bochner intgeral can be defined on any locally convex space, as projecteuclid.org/journals/annals-of-functional-analysis/… shows. $\endgroup$
    – user130903
    Commented Nov 5, 2021 at 9:14
  • 1
    $\begingroup$ I might add to the above comment that the space of tempered distributions is not just any old l.c.s.--it is a nuclear Silva space, i.e., a union of a sequence of Hilbert spaces with nuclear embeddings. As such it is complete, separable and its bounded sets (which are automatically rel. compact) and convergent sequences arise in the natural way from those of the components. One can make a case for the claim that such spaces are better behaved (more like f.d. spaces) than Banach spaces. Should you ever need tensor products, they are a treat to work with (reference: Köthe). $\endgroup$ Commented Nov 5, 2021 at 14:58

0

You must log in to answer this question.