Eigenvalue estimation using linearly dependent eigenfunctions

A technique is developed for estimating the eigenvalues of one problem usin g an expansion of linearly dependent eigenfunctions of another problem. Suc h an expansion cannot be used with Galerkin's method because the linens dep endence of the eigenfunctions renders Galerkin's method singular Test resul ts indicate that the method converges and is as accurate as Galerkin's meth od with linearly independent trial functions. [S0739-3717(00)01804-3].