Search

In this paper, we present a fast boundary integral equation method for the numerical conformal mapping and its inverse of bounded multiply connected regions onto a disk and annulus with circular slits regions. The method is based on two uniquely solvable boundary integral equations with Neumann-type and generalized Neumann kernels. The integral equations related to the mappings are solved numerically using combination of Nyström method, GMRES method, and fast multipole method. The complexity of this new algorithm is $$O((M + 1)n)$$ O ( ( M + 1 ) n ) , where $$M+1$$ M + 1 stands for the multiplicity of the multiply connected region and n refers to the number of nodes on each boundary component. Previous algorithms require $$O((M+1)^3 n^3)$$ O ( ( M + 1 ) 3 n 3 ) operations. The numerical results of some test calculations demonstrate that our method is capable of handling regions with complex geometry and very high connectivity. An application of the method on medical human brain image processing is also presented.

We study a numerical approach for solving integral equation with adjoint generalized Neumann kernel related to conformal mapping. Previously, computation of conformal mapping of M + 1 connected regions require solving at least M + 1 integral equations with adjoint generalized Neumann kernel separately. We apply global simpler GMRES which solve nonsymmetric system with multiple right-hand sides to solve M + 1 integral equations simultaneously. We also apply fast multipole method for several matrix vector products in every iteration of global simpler GMRES. Numerical example is given to illustrate the effectiveness of the proposed method.