Parallelizing stochastic L-systems simulation of plant stem and branch growth

Abstract

Lindenmayer system or L-system has been recognized for a long time as a mathematical theory of plant development. The principle of system is based on a rewriting language in which each part of plant is produced by substituting along with production rules. Since substituting on each part of the plant can be performed simultaneously and independenty without interfering to each other parts it is clearly seen that the whole job can be separated and performed by group of computers. In this research a parallel algorithm for stochastic L-system has been developed and examined. The program was implemented in C programming and Message-Passing Interface package to enable the system running on virtual parallel machines. The result was investigated for feasibility to improve operating time of stochastic L-system. The speedup factor is poor for lower derivation length, but it become pretty good at derivation length greater than 7. Maximum value of speedup factor for this research is approximately 19.6 at derivation length of 9 with 9 processes operated simultaneously.