ABSTRACT
This paper is concerned with the construction of efficient preconditioner for systems arising from discretization of a three-dimensional space-fractional diffusion equations. The matrix structure of the resulting linear systems is the summation of several 3-level Kronecker products. We propose to use one 3-level Kronecker product as a preconditioner. We obtain this Kronecker product through an alternating splitting iteration method, which is shown to be convergent. Optimal values of the iteration parameters are also obtained. The splitting iteration is then accelerated by a Krylov subspace method like GMRES. The components of the 3-level Kronecker product preconditioner have the same structure as the matrix derived from discretization of one-dimensional problem. Therefore, we use structure preserving approximation to the discrete one-dimensional problem as the building block for our preconditioner. Several numerical experiments are presented to show the effectiveness of our approach.
Acknowledgments
The authors are very grateful to the referees for their constructive comments and valuable suggestions, which greatly improved the original manuscript of the paper.
Disclosure statement
No potential conflict of interest was reported by the authors.