A new iterative Monte Carlo approach for inverse matrix problem (Q1298607)

The authors present new iterative Monte Carlo algorithms for the inverse matrix problem. The new algorithms are based on special choice of the iteration parameters which allow to control the convergence. They are applicable in the cases when not very accurate solution is needed and in finding special preconditioning matrices in the preconditioned iterative methods. The systematic (truncation) error which depends on the number of iterations and the stochastic (probable) error which depends on the probabilistic nature of the Monte Carlo methods are studied. The algorithms are well parallelized.