Boundedness of convolution operators and input-output maps between weighted spaces (Q1024744)

Let \(U\), \(Y\) be separable Hilbert spaces over \(\mathbb{R}\) or \(\mathbb{C}\), and let \(B(U, Y)\) be the space of bounded linear operators from \(U\) into \(Y\). The convolution of a \(U\)-valued function \(f: \mathbb{R}_+\to U\) and an operator-valued function \(H: \mathbb{R}_+\to{\mathcal B}(U, Y)\) is defined to be the \(Y\)-valued function \(H* f: \mathbb{R}_+\to Y\) given by \[ (H* f)(t)= \int^t_0 H(s) f(t-s)\,ds,\quad\text{where }t\in \mathbb{R}_+= \{t\in\mathbb{R}: t\geq 0\}. \] For functions \(m, w: \mathbb{R}_+\to\mathbb{R}_+\), define \({\mathcal S}={\mathcal S}(m,w,H)\in[0,+ \infty]\) by \[ {\mathcal S}= \sup\Biggl\{\int^\infty_0\|(H* f)(t)\|^2 m(t)\,dt: \int^\infty_0\| f(s)\|^2 w(s)\,ds\leq 1\Biggr\}. \] Problem 1.1. For \(U\), \(Y\) and \(H\) as above all fixed, find all weight functions \(m\), \(w\) such that \({\mathcal S}<\infty\) and estimate \({\mathcal S}\) in terms of \(m\), \(w\), \(H\). Problem 1.2. For \(U\), \(Y\) and the weights \(m\), \(w\) fixed, find all kernels \(H\) such that \({\mathcal S}<\infty\), and estimate \({\mathcal S}\). One of the main new results is the following. Result: Let \(w: (0,\infty)\to \mathbb{R}_+\) be a decreasing function with certain properties (including, for example, \(w(t)= t^{-\rho}\) or \((1+ t)^{-\rho}\) with \(\rho\geq 0\)). Let \(U\), \(Y\) be separable Hilbert spaces, and \(H\) operator-valued. Then, with this particular \(w\), Problem 1.2 is solved for \(m= w\): \(\| H* f\|_{L^2_Y(w)}\leq M\| f\|_{L^2_U(w)}\) for all \(f\in L^2_U(w)\) holds if and only if \[ {\mathcal L}H(z)= \int^\infty_{t= 0} e^{-zt}H(t)\,dt \] has a bounded analytic continuation to \(\{z\in\mathbb{C}:\text{Re}(z)> 0\}\), bounded by \(M\). In the case that \(U= Y=\mathbb{C}\) and \(H\geq 0\), this condition obviously holds if and only if \(H\in L^1(\mathbb{R}_+)\). The above result uses Hardy spaces and Bergman-type spaces, Laplace transforms, the Paley-Wiener theorem, and reproducing kernels. When the kernel function is a strongly continuous semigroup of bounded linear Hilbert space operators, which arises from input-output maps of certain linear systems, then the most obvious sufficient condition for boundedness, obtained by taking norm signs inside the integrals, is also necessary in many cases, but not in general.