Function include/PS_mp.m
computes the Taylor approximants to the matrix exponential in arbitrary precision; function include/cosm_hsd.m
is the counterpart that works in double precision and uses only double,
single, and half (simulated by chop
) precisions (no mp
computation). Function include/PS_mp_gen.m
is for general polynomials of matrices and takes the scalar coefficients (of decaying modulus) as one of the inputs.
Details on the underlying algorithms can be found in the technical report:
X. Liu. Mixed-precision Paterson–Stockmeyer method for evaluating polynomials of matrices, preprint, https://arxiv.org/abs/2312.17396.
All codes used for generating the data in the above report are included in this repository.
The code in this repository may require the Advanpix Multiprecision Computing Toolbox for MATLAB (www.advanpix.com).
See license.txt
for licensing information.