No label defined (Q4366389)

This paper is devoted to the numerical integration of the so-called generalized Burgers equation: \[ u_t+(\beta +\lambda u)u_x=\mu u_{xx} \] by means of finite differences. The author suggests to apply the second-order approximation of all the derivatives with respect to both variables \(t\) and \(x\) by three-point finite differences: forward, central, and backward. The computing process consists of five steps: -- a starting step computing the first two time lines, which uses the initial values, central and backward time differences, and the linearized equation, -- an iterative correction step, -- an explicit type scheme -- a kind of predictor for consecutive time steps, which uses the backward finite time difference and the linearized equation, -- an iterative correction step, -- and an additional step (when the whole solution is already determined), which improves the result by iterative correction on the whole time space grid. In the linear case, the stability analysis by means of the Fourier method is added. Moreover, the paper contains examples which compare the numerical solutions obtained by means of the algorithm proposed by the author with known solutions of the Burgers equation. The author's algorithm seems to work well in the cases under discussion.