Parallel Computation of Multivariate Normal Probabilities

Keywords: Multivariate Normal Probabilities, Parallel computation, multivariate integration algorithms

Abstract: We present methods for the computation of Multivariate Normal Probabilities on Parallel/ Distributed systems. After a transformation of the initial integral, an approximation can be obtained using Monte-Carlo or quasi-random methods (based on quasi-random point sets such as Richtmeyer or Sobol points).
For low to moderate dimensions, parallel adaptive methods have been implemented and tested on parallel/ distributed systems including networks of workstations.
This article will compare the efficiency and accuracy obtained with various multivariate integration algorithms and parallelizations implemented on MPI.