A class of finite-element methods for singularly perturbed second-order differential equations

comparison-upwind elements global error bounds linear elements numerical results Petrov- Galerkin methods quadratic elements singular perturbation small parameter uniform convergence

Mathematics Subject Classification ID

Linear boundary value problems for ordinary differential equations (34B05) Numerical solution of boundary value problems involving ordinary differential equations (65L10) Finite element, Rayleigh-Ritz, Galerkin and collocation methods for ordinary differential equations (65L60) Error bounds for numerical methods for ordinary differential equations (65L70)

Review text

The paper deals with singular perturbed boundary value problems of the form \[ \varepsilon u(x)+a(x)u'(x)=f(x) \text{ for } x\in[0,1],\quad u(0)=\alpha,\quad u(1)=\beta \] where \(\varepsilon\) is a small parameter, \(a\) and \(f\) are sufficiently smooth functions with \(a(x)>0\). The problem is solved approximately by a family of Petrov-Galerkin methods: piecewise linear elements are used as test functions and special quadratic elements --- called comparison-upwind elements --- are used as trial functions; a nonstandard quadrature rule is used to evaluate the corresponding inner products. For two particular methods of this family it is shown that they are uniformly convergent (with respect to \(\varepsilon\)) to the orders \(O(h)\) and \(O(h^ 2)\) at the nodes. Global error bounds are derived in \(L^ 1\) and \(L^ 2\). Some numerical results demonstrate high accuracy at low computational costs.

Recommendations

Cites work

Cited in

(13)

This page was built for publication: A class of finite-element methods for singularly perturbed second-order differential equations

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1175204)