Super-time-stepping for nonlinear singular perturbation problems

From MaRDI portal





We compare different methods for solving nonlinear systems resulting from the finite difference discretization of the nonlinear singularly perturbed boundary value problem \[ Su:= -\varepsilon u_{xx}+ b(x,u)= 0,\quad x\in X=[0,1],\quad u(0)= u(1)= 0.\tag{1} \] Here, \(\varepsilon\) is a perturbation parameter, \(0<\varepsilon\ll 1\), and \(b\) is a sufficiently smooth function. We are mainly interested in the time iterations, i.e. the methods based on convergence of solutions of the discrete analog of \[ u_t+ Su= 0,\quad (x,t)\in X\times[0,\infty), \] \[ u(0, t)= u(1,t)= 0\quad\text{for} \quad t\in[0,\infty),\quad u(x,0)= u_0(x)\quad\text{for} \quad x\in X, \] to the solution of the discrete analog of (1) as \(t\to\infty\). The only assumption on the initial function \(u_0\) is that it be bounded and measurable. In Section 2, we introduce a finite difference scheme to discretize (1). In addition to an equidistant mesh, a special mesh, dense in the layers, is considered. Then, we describe some standard iterative methods in Section 3. In Section 4, we turn our attention to the super-time-stepping method (STS). We also introduce a new, theoretically well-founded, quasi-optimal version of STS for nonlinear problems. Finally, in Section 5, we present numerical results and comparisons of the methods.











This page was built for publication: Super-time-stepping for nonlinear singular perturbation problems

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4378436)