Parallel Numerical Routines
The NAG Library has been extended with additional algorithms engineered for efficiency on the current generation of computer systems. With new functions in important areas that include optimization, statistics, matrix operations, PDE and ODE solvers. To take advantage of the latest processor and memory configurations, many algorithms have been implemented so that they are more efficient when running on multiple cores. Routines, such as Particle Swarm Optimization, have been chosen because the algorithm is intrinsically parallel and can be applied to solving important classes of problems. The routines will be available in two packages appropriate to either shared or distributed memory computation. The NAG Library for SMP & multicore uses OpenMP (Open Multi-Processing) to effect parallelism for shared memory system. A separate package, the NAG Parallel Library, is designed for distributed memory systems and is based on MPI (Message Passing Interface) techniques.