On reproducing kernel and density problems (Q2358762): Difference between revisions

The paper is devoted to the analysis of reproducing kernel Hilbert spaces with particular emphasis on embedding and density properties in the spaces \(C(X)\) and \(L^{p}(X,\mu)\), where \(X\) is a locally compact Hausdorff space. The article begins with a gentle introduction to reproducing Hilbert spaces, \(\mathcal{H}_{K}\) being the (unique) completion of the space of all functions of the form \(K^{x}(y):=K(y,x)\) with the inner product \((K^{x},K^{y})_{K}:=K(y,x)\), \(x,y\in X\), where \(K:X\times X\rightarrow\mathbb{C}\) is a positive definite kernel. The main problem under study is to formulate conditions guaranteeing the density of \(\mathcal{H}_{K}\) in \(C(X)\). In this direction, the author proves the following theorem (Theorem 2.7). Let \(X\) be a locally compact Hausdorff space, \(K\) a bounded and continuous positive definite kernel with \(K^{x}\in C_{0}(X)\), for all \(x\in X\). Then the following are equivalent: (i) \(\mathcal{H}_{K}\) is dense in \(C_{0}(X)\); (ii) if \(\mu\) is a finite Radon measure on \(X\), then \(K\) is (strictly) \(L^{2}\)-positive definite; (iii) for any finite real non-null Radon measure \(\mu\), it holds that \[ g(x)=\int_{X}K(x,y) \, d\mu(y)\neq 0,\text{ a.e.} \] In the third section, the author concentrates on the case of \(X\) being a locally compact abelian group which allows the use of the Fourier transform and a more refined version of Theorem 2.7 (Theorem 3.7) is proved in this case. The paper is concluded with remarks which are, in fact, applications of the main results.