Numerical fixed grid methods for differential inclusions (Q2471823)

The following initial value problem for differential inclusions is considered \[ \dot x \in F(x(t)) ~\text{ a.e. in }~[0,T], ~~ x(0) = x_0. \] For the numerical approximation it is necessary to discretize both time and space. The authors consider that there are two meaningful types of spatial discretization: An approximation of the images of \(F\) by a finite set of points is generated and used to construct approximating trajectories or the phase space is decomposed into small cells and the cells visited by trajectories are recorded. Both methods are analysed and compared. Convergence results are proved which show the combined influence of time and (phase) space discretization. Details of computer implementation are discussed and two examples are presented (the normal form system for Hopf bifurcation and the Michaelis-Menten system for reaction kinetics).