Object-oriented approach to the reduction of matrix bandwidth, profile and wavefront

The majority of problems arising in science and engineering require the solution of a large set of linear algebraic equations such as, in matrix form. Ax = b. This type of equation is usually solved using some form of Gaussian elimination. It is necessary to the finite element users that the nodes and elements are numbered correctly since nearly all computer programs contain linear algebra solution routines. These are usually expressly written to operate efficiently on matrices possessing small bandwidths, profiles or wavefronts (frontwidths). The object-oriented implementation of bandwidth, profile and wavefront reduction is based on an algorithm published by Sloan, which seems to perform consistently better to that of the widely used reverse Cuthill-McKee method and the Gibbs-King method. This article presented for the first time during CST'96 Conference [Gajewski, R.R., Lompies, P., Object-oriented implementation of bandwidth, profile and wavefront reduction algorithms, In: Advances in Computational Structures Technology, ed. B.H.V. Topping, Civil Comp Press, 1996, pp. 115-120.] provides a full description of the implemented classes, their hierarchy and implementation. It is also illustrated by examples of practical calculations. (C) 1999 Elsevier Science Ltd and Civil-Comp Ltd. All rights reserved.