Sparse matrix factorizations of transfer matrices

From MaRDI portal





For easing the computational effort in obtaining the leading eigenvalues a factorization of the transfer matrix into a product of sparse matrices is used. This paper is reviewing first the factorizations of the interactions round a face model in which the first and second neighbour interactions are contained. The sparse factors of a more general Ising model containing first, second and third neighbour interactions are also presented. For each considered case the formulation of the transfer matrix as a product of sparse factors is achieved by viewing the auxiliary spin sets required as a hierarchy of interacting spins. The majority of these spins interact through delta functions, giving rise to sparse nature of the factors. Each factor of the transfer matrix can be thought of as arising from the fundamental units (faces or wedges) in a given row. Using the sparse matrix representation, a significant reduction of the work requirement for computing the leading eigenvalues of the transfer matrix is achieved.











This page was built for publication: Sparse matrix factorizations of transfer matrices

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3761570)