Computational improvement of gravitational mode calculation of basins discretised by orthogonal curvilinear grids

Abstract

The present note illustrates a criterion to improve the computational capability of the approaches proposed by Beltrami et al. [Beltrami, G.M., Bargagli, A., Briganti, R., 2003. Gravitational mode calculation of basins discretised by orthogonal curvilinear grids. Ocean Engineering 30, 833–853] for the direct numerical solution of the eigenvalue problem associated to the linear shallow-water equations when adiabatic boundary conditions apply. It is shown that—given the nature of its spatial differential operator—the problem can be solved by the singular value decomposition (SVD) of the real bidiagonal matrix resulting from a previous ad hoc Householder reduction of the operator matrix image. This procedure actually requires 1/8 of the random-access memory (RAM) needed by a standard library routine to compute all the eigenvalues and eigenvectors of the matrix image of the above-mentioned differential operator. Given the intrinsic limitation of a computing-machine RAM, this procedure dramatically improves the computational capability of both the proposed approaches.