PRECONDITIONING IN ECONOMIC STOCHASTIC GROWTH MODELS

摘要

Krylov subspace methods (KSM) have proven to be powerful methods for solving sparse linear systems arising in several engineering problems and, more recently, in computational economics. When used to solve stochastic growth models, KSM fail to converge for models describing very short time horizons. To improve their convergence properties in this application, one can use different preconditioners. The properties of different preconditioning techniques are investigated and reported.