Least squares methods for 2𝑚th order elliptic boundary-value problems

James H. Bramble, A. H. Schatz

Mathematics of Computation · 1971 · 71 citations · 12 references

DOIFull text

Open access

Concepts

Abstract

In this paper we consider a general class of boundary-value problems for 2 <italic>m</italic> th order elliptic equations including nonhomogeneous essential boundary conditions and nonselfadjoint problems. Approximation methods involving least squares approximation of the data are presented and corresponding error estimates are proved. These methods can be considered in the category of Rayleigh-Ritz-Galerkin methods and have the special feature that the trial functions need not satisfy the boundary conditions. A special case of the trial functions which is studied are spline functions defined on a uniform mesh of width <italic>h</italic> (or more generally piecewise polynomial functions). For a given "well set" boundary-value problem for a 2 <italic>m</italic> th order operator the theory presented will provide a method with any prescribed order of accuracy <italic>r</italic> which is optimal in the sense that the best approximation in the underlying subspace is of order of accuracy <italic>r</italic> .

References

12