ERROR CONTROL IN NUMERICAL SOLUTIONS OF BOUNDARY INTEGRAL EQUATIONS

Abstract

A method of estimating the error committed in numerical solutions of integral equations is presented. It is shown how the error can be computed in spaces other than L^2, why this is reasonable and sometimes necessary. The appropriate function spaces are also shown to lead to bounded condition numbers even for first kind equations. [Vol. 11, No. 1 (1996), pp 32-36, Special issue on Applied Mathematics: Meeting the challenges presented by computational electromagnetics]