On the Scalable Computation of Large Sets of Integrals

We consider efficient strategies for the parallel and distributed computation of large sets of multivariate integrals. These arise in many applications such as computational chemistry, high energy physics and finite element problems. We use a hierarchical architecture which is highly scalable and allows parallelization not only at the domain level but also at the integral level. We present experimental results on a family of parametrized integrals and show that our approach leads to an efficient scalable load balanced algorithm. The methods are easily applicable to other areas where a large number of problems need to be solved efficiently on a parallel and distributed system. Keywords. High performance computing, cluster computing, load balancing, integration software library, task allocation.

• Type: Conference
• Year: 2003
• Where: ISCAPDCS
• Authors: Elise de Doncker, Ajay K. Gupta, Laurentiu Cucos
• Keywords: Efficient Scalable Load, ISCAPDCS 2003, ISCAPDCS 2007, Multivariate Integrals, Parametrized Integrals