Deterministic and stochastic algorithms for mechanical systems under constraints.

We discuss the general philosophy underlying the statistical behaviour, the dynamical evolution and the integration of the equations of motion for systems subject to constraints. We also show how all this is related to the treatment of general non-Hamiltonian systems. Then we introduce a family of algorithms derivable from approximations of the evolution operator obtained via the Trotter formula. Generalizing the treatment to time-dependent force fields we also show how one can adapt those algorithms to ordinary stochastic differential equations. Copyright 2004 The Royal Society