A modified piecewise linear Markov approximation of Markov operators (Q2490203): Difference between revisions

\textit{J. Ding} and \textit{T. Y. Li} [Nonlinear Anal., Theory Methods Appl. 17, No. 8, 759--772 (1991; Zbl 0758.28014)] proposed a piecewise linear numerical method based on a Markov finite approximation of an integrable function for the computation of absolutely continuous invariant measures associated with chaotic interval mappings. The present paper proposes a modified piecewise linear Markov approximation method for functions of one variable. The novelty consists in a new discretization operator \(Q_n\) for approximating a function in \(L^1(0, 1)\). Several numerical examples compare the original piecewise linear Markov method of Ding and Li (loc. cit.), and the newly proposed piecewise linear Markov method, when applied to computing fixed densities of Frobenius-Perron operators. The results show an improved approximation for the new (and simpler) method in comparison to the original one. Furthermore, stability and convergence properties of the novel piecewise linear method based on Markov finite approximations are studied in terms of the \(L^1\)-norm and the bounded variation norm. The new technique provides a simple and fast numerical scheme for computing fixed densities of Markov operators, and can be successfully employed in the statistical study of stochastic analysis of dynamical systems.