Intervals and OpenMP : towards an efficient parallel result-verifying nonlinear solver
Thomas Beelitz, Christian H Bischof, Bruno Lang · TUbilio (Technical University of Darmstadt) · 2003
Abstract. Nonlinear systems occur in diverse applications, i.e., in the steady state analysis of chemical processes. If safety concerns require the results to be provably correct then result-verifying algorithms relying on interval arithmetic should be used for solving these systems. Since such algorithms are very computationally intensive, parallelism must be ex-ploited to make them feasible in practice. We describe our framework for the verified solution of nonlinear systems and our approach to paralleliz-ing it with OpenMP. First numerical results show that the parallelization scheme is indeed successful but that the attainable speedup is limited for some unknown reason. 1