Parallel simulation of axon growth in the nervous system

In this paper we discuss a model from neurobiology, which describes the outgrowth of axons from neurons in the nervous system. The model combines ordinary differential equations, defining the movement of the axons, with parabolic partial differential equations. The parabolic equations model the concentrations of chemicals. The axons are guided by the gradients of these chemoattractant and chemorepellant concentrations. We briefly discuss the numerical techniques that we have used to solve this coupled parabolic-gradient system. Special attention is given to the parallel implementation on the SGI Origin 3000 (Teras), a multiprocessor machine. For that purpose we use the OpenMP standard. Several parallelization strategies are introduced and tested on the basis of a test example. Simulation results as well as performance results are reported. (C) 2003 Elsevier B.V. All rights reserved.