Data

Tools

Indexes

Applications

Experiments

Open Source Science

mphodlr_exp

Experimental code for mixed precision HODLR matrices

https://github.com/inexascale/mphodlr_exp

Science Score: 54.0%

This score indicates how likely this project is to be science-related based on various indicators:

  • ✓
    CITATION.cff file
    Found CITATION.cff file
  • ✓
    codemeta.json file
    Found codemeta.json file
  • ○
    .zenodo.json file
  • ✓
    DOI references
    Found 8 DOI reference(s) in README
  • ✓
    Academic publication links
    Links to: zenodo.org
  • ○
    Academic email domains
  • ○
    Institutional organization owner
  • ○
    JOSS paper metadata
  • ○
    Scientific vocabulary similarity
    Low similarity (10.3%) to scientific vocabulary
Last synced: 10 months ago · JSON representation ·

Repository

Experimental code for mixed precision HODLR matrices

Basic Info
  • Host: GitHub
  • Owner: inEXASCALE
  • Language: MATLAB
  • Default Branch: main
  • Homepage:
  • Size: 511 KB
Statistics
  • Stars: 2
  • Watchers: 3
  • Forks: 2
  • Open Issues: 0
  • Releases: 0
Created about 2 years ago · Last pushed about 1 year ago
Metadata Files
Readme Citation

README.md

mphodlr_exp

This repository contains the fully reproducible experimental code for the paper Mixed precision HODLR matrices [1].

Download

mphodlr_exp contains large files storage. To download the full repository, please ensure git lfs is properly set up (see here for details) and use the following commands:

GIT_LFS_SKIP_SMUDGE=1 git clone https://github.com/inEXASCALE/mphodlr_exp.git cd mphodlr_exp git lfs pull

Full repository containning all code and data can also be obtained in here.

Requirements

Due to large files storage, the software @precision, @hodlr, and @ampholdr, which can be downloaded from https://github.com/chenxinye/mhodlr. MATLAB 2024a or newer (with Statistics and Machine Learning Toolbox) is required. The experimental code was simulated in terms of the version Commit 706333a.

Instruction

Detailed guidance is referred to index:

  • The scripts plot_saylr3.m and plot_LeGresley.m are used to generate [Fig. 4.1, 1].

  • The scripts exp_rcerr.m and plot_exp_rcerr.m are used to generate the results for Fig. 5.1, 1.

  • The scripts exp_mvprod.m and plot_exp_mvprod.m are used to generate the results for Fig. 5.2, 1.

  • The scripts exp_lu.m and plot_exp_lu.m are used to generate the results for Fig. 5.3, 1.

  • The scripts exp_storage.m and plot_exp_storage.m are used to generate the results for Fig. 5.4, 1.

All test matrices stored in the folder data are from Amestoy et al. [2] and SuiteSparse collection [4]. The low precision arithmetics are simulated by chop [3]. One can perform all experiments at one go by running the command run_all. The generated results and figures are separately stored in results and figures, respectively.

References

[1] C. Erin, X. Chen and X. Liu, Mixed precision HODLR matrices, arXiv:2407.21637, (2024), https://doi.org/10.48550/arXiv.2407.21637.

[2] P. Amestoy, O. Boiteau, A. Buttari, M. Gerest, F. Jezequel, J.-Y. LExcellent, and T. Mary, Mixed precision low-rank approximations and their application to block lowrank LU factorization, IMA J. Numer. Anal., 43 (2022), pp. 21982227, https://doi.org/10.1093/imanum/drac037.

[3] N. J. Higham and S. Pranesh, Simulating low precision floating-point arithmetic, SIAM J. Sci. Comput., 41 (2019), pp. C585C602, https://doi.org/10.1137/19M1251308.

[4] T. A. Davis and Y. Hu, The University of Florida Sparse Matrix Collection, ACM Trans. Math. Software, 38 (2011), https://doi.org/10.1145/2049662.2049663.

Owner

  • Name: inEXASCALE
  • Login: inEXASCALE
  • Kind: organization

Citation (CITATION.bib) 

@misc{carsonMPM2024,
      title={Mixed precision HODLR matrices}, 
      author={Erin Carson and Xinye Chen and Xiaobo Liu},
      year={2024},
      eprint={2407.21637},
      archivePrefix={arXiv},
      primaryClass={math.NA},
      url={https://arxiv.org/abs/2407.21637}, 
}

GitHub Events

Total
  • Watch event: 2
  • Push event: 74
  • Fork event: 1
Last Year
  • Watch event: 2
  • Push event: 74
  • Fork event: 1