Discrete simulation of large aggregates of neurons

Download

PDF - Publishers print Restricted to Registered users only Download (345Kb) | Request a copy

Description/Abstract

Realistic simulation of aggregates of neurons often utilises compartmental models which limit the scope of the simulations with single processor architectures to small or medium size networks (typically hundreds of neurons). An alternative approach, based on cell automata models, allows e1cient simulation of nervous tissue by modelling neurons as 2nite state automata. In this paper, data structures and algorithms appropriate for efficient simulation of message-based event-driven models of neurons in single processor architectures are presented. With these techniques, the simulation of large networks (of the order of 105 neurons with 102 synapses per neuron) becomes feasible.
All rights reserved.