The sensitivity of a computational L1-approximation technique for integral equations

摘要

A method of solving integral equations using an L1-approximation technique has previously been discussed. However, some questions concerning the sensitivity of such a method were still unanswered. In this paper, we make investigations towards the important problems of (a) closeness of parameter and eigenvalue(s) to one another, (b) noise in the original data, and (c) bifurcations in the solution of the integral equation. Results are given for these cases, illustrating satisfactory behavior of the technique for these so-called “sensitive” problems.