Boosting the computation of the matrix exponential

作者：

Highlights：

• Two new Taylor algorithms for the computation of the matrix exponential are proposed.

• They are based on matrix polynomial evaluation methods more efficient than Paterson–Stockmeyer method.

• In tests they showed higher efficiency and accuracy than state-of-the-art Padé based methods.

• They also showed a higher efficiency than state-of-the-art Taylor methods, and higher accuracy in some tests.

摘要

•Two new Taylor algorithms for the computation of the matrix exponential are proposed.•They are based on matrix polynomial evaluation methods more efficient than Paterson–Stockmeyer method.•In tests they showed higher efficiency and accuracy than state-of-the-art Padé based methods.•They also showed a higher efficiency than state-of-the-art Taylor methods, and higher accuracy in some tests.

论文关键词：Matrix exponential,Scaling and squaring,Taylor series,Efficient matrix polynomial evaluation

论文评审过程：Received 4 February 2018, Revised 16 July 2018, Accepted 13 August 2018, Available online 9 September 2018, Version of Record 9 September 2018.

论文官网地址：https://doi.org/10.1016/j.amc.2018.08.017