| Volume 7, Number 4, 2017, Pages 1402-1416 DOI：10.11948/2017085 |
| Approximation of the linear combination of $\varphi$-functions using the block shift-and-invert Krylov subspace method |
| Dongping Li,Yuhao Cong |
| Keywords:Matrix exponential, exponential integrators, block shift-and-invert Krylov subspace, a posteriori error estimates. |
| Abstract: |
| In this paper, we develop an algorithm in which the block shift-and-invert Krylov subspace method can be employed for approximating the linear combination of the matrix exponential and related exponential-type functions. Such evaluation plays a major role in a class of numerical methods known as exponential integrators. We derive a low-dimensional matrix exponential to approximate the objective function based on the block shift-and-invert Krylov subspace methods. We obtain the error expansion of the approximation, and show that the variants of its first term can be used as reliable a posteriori error estimates and correctors. Numerical experiments illustrate that the error estimates are efficient and the proposed algorithm is worthy of further study. |
PDF Download reader
|
|
|
|