No label defined (Q4026082)

[For the entire collection see Zbl 0745.00036.] The paper is concerned with the following question: \((Q)\): To what extent is the (discrete) dynamics of the approximate solutions generated by a numerical method, a faithful description of the dynamics of the corresponding system of ordinary differential equations? The paper is mostly expository. It benefits from the rich experience of the author documented by several publications of him and coauthors. A rich list of further valuable references is included. The standard analysis of linear multistep and Runge-Kutta methods with respect to error bounds and absolute stability is reviewed. Question \((Q)\) is discussed by means of a simple example due to Brezzi et al. (1984) where the time-continuous problem exhibits Hopf bifurcation with globally attractive equilibrium or limit cycle --- a situation which is dramatically changed after discretization by Euler method. Results are quoted concerning hyperbolic equilibria and periodical orbits of autonomous systems and leading to an optimistic standpoint for \((Q)\) --- at least for sufficiently small steplengths. It is pointed out, that --- on the other hand --- spurious equilibria, spurious periodical orbits and even spurious chaos have been found in the literature which show pathological behavior of discretizations --- a pessimistic point of view in answering \((Q)\). Finally the author draws certain conclusions which result into an affirmative answer to \((Q)\), particularly by library software packages with variable steplength strategies are used.