Parallel Solutions of Stochastic Differential Equations

Abstract

The main purpose of this work was to develop a more time efficient solution to the Lotka- Volterra model. To compare runtimes, in addition to the serial version, the algorithm was implemented using the Message Passing Interface as well as on a GPU using CUDA. By comparing the best runtime for the three versions we concluded that the KMC method is best suited for the message passing interface. We have studied the accuracy of the KMC method and it turned out that there was a limit for which increasing the number of simulations did not affect the accuracy significantly. In order to reach this limit in a reasonable amount of time the use of parallel programming is essential.