Eigenvalues of the Laplacian Through Boundary Integral Equations (Q3977067)

A numerical method for the eigenvalue problem of the two-dimensional Laplacian is considered. The authors look for a method of rapid calculations with a possibly small number of operations. The presented method is based on a boundary integral formulation. The integral equation is reduced to a matrix problem for an \(N\times N\) matrix \(A(\lambda)\). Instead of computing the determinant of \(A(\lambda)\) (it requires \(O(N^ 3)\) operations), the authors propose to compute the smallest eigenvalues \(\mu(\lambda)\) of \(A^ TA\) (which is possible to do by a few iterations). In order to reduce the number of the matrix-vector multiplications, the idea of multipole expansions is applied. Finally, methods for finding zeros of \(\mu(\lambda)\) are discussed. The presented method requires \(O(N)\) operations on each iteration step for calculating one eigenvalue.