A sparse matrix scheme for finite elements (Q1205237)

The authors propose an implementation of compact storage of the finite element matrix \(A\) arising from the discretization of problems which are described by a system of partial differential equations, so that some fixed number \(d>1\) of degrees of freedom correspond to each node. They show how to reduce approximately by \(d\) the storage required for indices of nonzero entries as compared with more conventional storage schemes and describe the implementation of matrix vector multiplication which may be exploited in iterative solvers.