Fredholmness of convolution type operators

From MaRDI portal





Fredholm properties of convolution type operators on an unbounded domain \(D\) in \(\mathbb{R}^N\) is studied. Convolution type operators on \(\mathbb{R}^N\) are those which belong to the smallest algebra generated on \(L(L^p(\mathbb{R}^N))\) by \(\gamma I+aC(k)bI\), where \(\gamma=\)const, \(a(x)\) and \(b(x)\) are bounded and uniformly continuous functions while \(C(k)\) is a convolution operator with the integrable kernel: \[ C(k)\varphi(t):=\int_\mathbb{R}k(t-\tau)\varphi(\tau) \,d\tau,\qquad k\in L^1(\mathbb{R}^N) . \] Convolution type operators on a domain \(D\) are compressions of the corresponding operators from \(L(L^p(\mathbb{R}^N))\): \[ \chi_D\big(\gamma I+C(k)\big)\chi_DI\Big| _{L^p(D)}: L^p(D)\to L^p(D)\, . \] The domain \(D\subset\mathbb{R}^N\) under consideration might be a curved layer, a curved cylinder, cones with angular and even cuspidal edges. A Fredholm property criterion is found and formulated in terms of limit operators: to each operator \(B\) under investigation there is associated a family of operators on \(L^p(\mathbb{R}^N)\) (the operator spectrum of \(B\)) which is responsible for a behavior of operator \(B\) at infinity. The main results state: \(B\) is a Fredholm operator if and only if each operator from its spectrum is an invertible element in \(L(L^p(\mathbb{R}^N))\) and the norms of these inverses are uniformly bounded. In some cases (e.g. for slowly oscillating coefficients) rather effective inverses are found because the operators in the spectrum are simpler objects than the original operator. For the entire collection see [Zbl 1031.00013].











This page was built for publication: Fredholmness of convolution type operators

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4464542)