Some features of solving the problems of mapping for allocation of chemical elements on stellar surfaces as ill-posed problems with the use of multiprocessor systems

Abstract

The problem of mapping for allocation of chemical elements on stellar surfaces is considered as an ill-posed one. The minimization of Tikhonov’s functional with the choice of a regularization parameter according to the finite-dimensional generalized discrepancy principle is used to construct a regularizing algorithm for solving the problem under consideration. The method based on projections of conjugate gradients onto a set of non-negative vectors is chosen as a method of minimization. An approach to numerical solution of the minimization problem is considered for the above mapping problem. A multiprocessor computer is used to solve several model problems numerically. Some schemes of parallelization are offered and special features of their realization are discussed.