We discuss some techniques that are likely to improve the efficiency of implementation of the code GAM [1,2,3] on a parallel computer. It is our experience that most of the time (about 85 %) spent in advancing the solution is devoted to solving the nonlinear systems underlying the integration procedure; the whole performance of the code will substantially benefit from an efficient parallelization of this part. The parallel version of the code GAM has been implemented using Fortran90 and MPI (Message Passing Interface).

References

[1] F. Iavernaro, F. Mazzia, Block-Boundary Value Methods for the solution of Ordinary Differential Equations, SIAM J.Sci. Comput. (to appear)

[2] F. Iavernaro, F. Mazzia, Solving ordinary differential equations by generalized Adams methods: properties and implementation techniques, Appl. Num. Math., 28 (2-4), 107-126, (1998).

[3] L. Brugnano, D. Trigiante, Solving Differential Problems by Multistep Initial and Boundary Value Methods, Gordon & Breach, Amsterdam, 1998.

MINISIMPOSIUM SESSION: 6. Novel methods for ODEs