Efficient Leaping Methods for Stochastic Chemical Systems

Abstract. Well stirred chemical reaction systems which involve small numbers of mole-cules for some species have a stochastic behavior and can be modeled by a continuous time, discrete state Markov process. An exact method for simulating the time evolution of the system is the Stochastic Simulation Algorithm, but this method is extremely slow for realistic biological systems. We propose an adaptive leaping method based on local error formulas and we compare the adaptive method with the exact method and with the explicit tau leaping method with constant time steps. Key words. stochastic chemical kinetics, stochastic simulation algorithm, tau-leaping AMS subject classifications (2000). 60H35, 65C3