A parallel computational algorithm for an inverse problem of low frequency

Abstract We propose a parallel computational algorithm for an inverse problem of low frequency, which is shown to have the property of preserving symmetry. The corresponding discrete operator is shown to have the property also. We construct a simple orthogonal matrix to decompose the linear system resulting from discretization of the problem into four subsystems. This parallel algorithm reduces the computing time by a factor at least 16, and hence makes it possible to treat large scale inverse in practice.