Stabilized times schemes for high accurate finite differences solutions of nonlinear parabolic equations (Q2399226)

This paper of review character starts from considering parabolic equations and ways to stabilize the finite difference methods approximating those equations. The authors cite as first work in this direction an unpublished report from 1998 whereas the Crank-Nicolson method (1948) goes in the same direction as does a series of papers of Russian mathematicians like D'yakonov or Samarskij (mostly published in Numerical Mathematics and Mathematical Physics which has been published by the AMS in parallel in English, now the Springer journal Computational Mathematics and Mathematical Physics) beginning from the sixties (e.g. [\textit{A. A. Samarskii}, U.S.S.R. Comput. Math. Math. Phys. 2, 894--926 (1962; Zbl 0273.65078); translation from Zh. Vychisl. Mat. Mat. Fiz. 2, 787--811 (1962)]). The authors obtain stability results in the well-known way of \(L_2\) estimates, similar to those of the \(\theta\) scheme. They also cite and explain a parallel development in the Fourier space going back to Temam and coworkers. They further obtain stability results for the semilinear case (including the case of a convex-concave splitting of the nonlinear source function). Further they mention the Richardson extrapolation and preconditioning by lower-order operators (e.g. Laplace preconditioning for the biharmonic operator), and compact difference schemes, and obtain stability results for the case that the higher-order operator is not symmetric but positive definite (with small nonsymmetric part). In their numerical experiments mainly the behaviour of the considered schemes for Navier-Stokes equations in the stream function -- vorticity formulation on the driven cavity problem is reported. The final outcome is that a not properly defined scheme is the best one. It would have been a benefit if the paper was checked from the language point of view. This is not only a question of misprints (and there are misprints also in some of the formulas).