Phase Space Error Control for Dynamical Systems

From MaRDI portal





This very nice paper discusses the dynamics of variable stepsize methods. The authors note that adaptive schemes can perform badly near a stable fixed pointd and that nearly all adaptive explicit Runge-Kutta (ERK) methods admit stable spurious fixed points for arbitrarily small tolerances when using traditional error control. A new error criterion is introduced that bounds the truncation error by a fraction of the solution arclength. If the trapezoidal method is used to approximate the error then this gives the so-called phase space (PS) error control: \[ \|y_{n+ 1}- y_n-{h\over 2}(f(y_{n+ 1})+ f(y_n))\|\leq {h\over 2} \phi\|f(y_{n+ 1})+ f(y_n)\|,\quad \phi\in [0,1]. \] The admissability of PS-error control is shown and that it does not admit spurious fixed points or period-two solutions. In addition, the linear stability behaviour of adaptive ERKs under PS-error control is studied in terms of \(\phi\). Finally, the authors show how to choose \(h\) when the PS-error constraint is violated.











This page was built for publication: Phase Space Error Control for Dynamical Systems

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4509832)