On the numerical solutions of two-point singularly perturbed boundary value problems (Q759576)

We consider a class of singularly perturbed two-point boundary value problems with various types of boundary conditions and examine the layer behavior of the solution. By constructing a more regular modified problem with a correction term we discuss how to deal with the layer separately and prove error estimates which are uniform in \(\epsilon\). For standard finite difference or finite element methods we then show that the system of equations may be reduced to a positive matrix problem for which estimates in maximum norm are available. These estimates depend on the smoothness of the solution to the separated problems on the whole interval of definition. Finally, we perform a numerical test to illustrate the method of separating the calculation of the layer.