A stopping criterion for the iterative solution of partial differential equations

摘要: Abstract A stopping criterion for iterative solution methods is presented that accurately estimates the error using low computational overhead. The proposed uses information from prior changes to estimate error. When are noisy or stagnating it reverts a less accurate but more robust, low-cost singular value approximate given residual. This estimator can also be applied linear matrix solvers such as Krylov subspace multigrid methods. Examples of criterion's ability non-linear and provided number different test cases in incompressible fluid dynamics.