helloGreat / sketched_fAb

Implementation of algorithms presented in "Randomized sketching for Krylov approximations of large-scale matrix functions"

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Randomized sketching for Krylov approximations of large-scale matrix functions

Overview

The code in this repository implements the algorithms proposed in the manuscript [1] (Link: https://arxiv.org/abs/2208.11447). In particular, it allows to reproduce all tables and figures in the manuscript.

Most scripts are not optimized for fast execution, as their purpose is comparing and plotting the error of different methods, not comparing their runtime.

Reproducing figures and tables from [1]

For reproducing the figures and tables from [1], run the scripts drive_*.m at the top level of the folder structure.

  • drive_convdiff.m: This script runs the experiment reported in "Section 5.1: Convection-diffusion example" of [1]. It generates the two plots shown in Figure 5.1.
  • drive_network.m: This script runs the experiment reported in "Section 5.2: Network example" of [1]. It generates the four plots shown in Figure 5.2.
  • drive_qcd.m: This script runs the first experiment reported in "Section 5.3: Lattice QCD" of [1]. It generates the four plots shown in Figure 5.3.
  • drive_qcd_timings.m: This script runs the second experiment reported in "Section 5.3: Lattice QCD" of [1]. It generates the LaTeX code for Table 5.1.

Citation

If you use our code in a scientific publication, please cite:

@techreport{GuettelSchweitzer2022,
  title={Randomized sketching for {K}rylov approximations of large-scale matrix functions},
  author={G{\"u}ttel, Stefan and Schweitzer, Marcel},
  year={2022},
  url={https://arxiv.org/abs/2208.11447}
}

License

This project is licensed under the MIT License.

References

[1] S. Güttel, M. Schweitzer, Randomized sketching for Krylov approximations of large-scale matrix functions, arXiv preprint arXiv:2208.11447 (2022) https://arxiv.org/abs/2208.11447

About

Implementation of algorithms presented in "Randomized sketching for Krylov approximations of large-scale matrix functions"

License:MIT License


Languages

Language:MATLAB 98.4%Language:M 0.8%Language:TeX 0.8%