The Fredholm equation of the first kind is widely encountered in the fields ofstructural mechanics, image processing, geological prospecting etc. Due to the ill-posedness, it is both theoretical and practical importance to develop effective methodsto solve the problem. To date, the Tikhonov regularization method is very effective insolving ill-posed problems. The basic idea is, based upon certain a priori informationon the exact solution, to transform the original problem into some feasible well-posedoptimization problems to get the regularization solutions. Then one can obtain a stablesolution method for the original problem.In this thesis, such a method is applied to solve the Fredholm equation of thefirst kind, where many kinds of existing strategies for choosing regularization param-eters and different scales of partitions for discretizing the integral equation are consid-ered, to show extensively and thoroughly the numerical performance of the method.Furthermore, to illustrate the computational performance for functions with differentsmoothness, the method is used to solve problems which have differentiable, contin-uous but not differentiable, and discontinuous solutions, respectively. The method issuccessfully applied to solve two Fredholm equations of the first kind arising fromgeological prospecting.
|