Spectral methods for linear inverse problems with unbounded operators (Q1191736)

Quite many linear inverse problems of mathematical physics may be described in abstract setting as linear operator equations: \(Ax=f\). The desired solution \(x\) is often given in terms of the Moore-Penrose generalized inverse of \(A\). In the case of \(A\), a closed, densely defined linear operator, not very many special results are known. The author makes use of spectral theory for bounded linear operators in Hibert space to develop a general framework for stable approximate solutions of abstract inverse problems posed in either implicit or explicit form, involving closed, densely defined linear operators which may be unbounded. The problem of convergence and stability are addressed and the methods are modified to provide a stable class of methods for evaluation of unbounded linear operators.