On the use of parallel processors for implicit Runge-Kutta methods

From MaRDI portal





Using an implicit Runge-Kutta method for a system of stiff differential equations one has to solve in each integration step a nonlinear system of equations for the internal approximation vectors. The authors discuss and propose modified Newton schemes that allow computation of the Newton correction in parallel. They prove that the scheme converges and show by numerical experiments a speedup of 1.8 to 4.1 on a 6 processors Sequent Symmetry.



Cites work









This page was built for publication: On the use of parallel processors for implicit Runge-Kutta methods

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1310614)