Let's define the operator $K$ as $$ \begin{eqnarray*} K :&H&\longrightarrow &\;\:Z&\\ &y\left( x;\sigma \right) &\mapsto &\!\!\!\!\int_{0}^{1}C\left( \sigma \right) y\left( x;\sigma \right) d\sigma& \end{eqnarray*}\: $$ where $H$, $Z$ are Hilbert spaces, $C\left( \sigma \right) $ is a linear bounded operator from $H$ to $Z$, $x$ and $\sigma$ are real parameters. My question is: how can I prove that $K$ is bounded, and what does its adjoint operator look like?

  • 2
    $\begingroup$ I don't think the question can be answered in the generality in which you've posed it, which relies at least on being able to interpret elements of $H$ as functions of a real parameter $\sigma$ (as well as possibly of some other unspecified parameter $x$). $\endgroup$ – LSpice May 6 at 16:59

I may assume that $H=Z=L^2(\mathbb R)$ and the mapping $K$ to be given by a distribution kernel $k(s,t)$ via a formula $$ Ku(s)=\int k(s,t) u(t) dt, $$ meaning that for $u,v\in C^\infty_c(\mathbb R)$, we have $\langle Ku, v\rangle_{\mathscr D'(\mathbb R),\mathscr D(\mathbb R)}=\langle k, v\otimes u\rangle_{\mathscr D'(\mathbb R^2),\mathscr D(\mathbb R^2)}$. Now your question could be reformulated as follows: are there general criteria for extending $K$ to a bounded operator on $L^2(\mathbb R)$? The answer is yes, some general criteria are available.

  1. Schur's criterion, for a locally integrable kernel, $$ \sup_s\int \vert{k(s,t)}\vert dt,\sup_t\int \vert{k(s,t)}\vert ds\quad \text{ both finite.} $$

  2. Singular integrals criteria, which can be applied for instance to $k=pv(1/(s-t))$, which is a Fourier multiplier by a bounded function.

  3. Operators which are products of type 1,2. In particular pseudo-differential operators of order 0, but it is less elementary.


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.