Data

Tools

Indexes

Applications

Experiments

Open Source Science

randnla.jl

Julia algorithms for randomized numerical linear algebra.

https://github.com/rs-coop/randnla.jl

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
    Found .zenodo.json file
  • ○
    DOI references
  • ✓
    Academic publication links
    Links to: arxiv.org
  • ○
    Academic email domains
  • ○
    Institutional organization owner
  • ○
    JOSS paper metadata
  • ○
    Scientific vocabulary similarity
    Low similarity (13.2%) to scientific vocabulary

Keywords

julia linear-algebra randomized-algorithms
Last synced: 9 months ago · JSON representation ·

Repository

Julia algorithms for randomized numerical linear algebra.

Basic Info
  • Host: GitHub
  • Owner: RS-Coop
  • License: mit
  • Language: Julia
  • Default Branch: main
  • Homepage:
  • Size: 14.6 KB
Statistics
  • Stars: 0
  • Watchers: 1
  • Forks: 0
  • Open Issues: 0
  • Releases: 0
Topics
julia linear-algebra randomized-algorithms
Created almost 3 years ago · Last pushed over 2 years ago
Metadata Files
Readme License Citation

README.md

RandNLA.jl: Randomized algorithms for numerical linear algebra

Author: Cooper Simpson

RandNLA.jl is a pure julia package for fundamental operations and state-of-the-art research. The core is lightweight, but a variety of extensions allow for easy plug-and-play use in other packages. This is still very much a work in progress, and a full roadmap is given in Roadmap.

License & Citation

All source code is made available under an MIT license. You can freely use and modify the code, without warranty, so long as you provide attribution to the authors. See LICENSE for the full text.

This repository can be cited using the GitHub action in the sidebar, or using the metadata in CITATION.cff.

Features

  • AbstractFloat support?
  • GPU support?
  • Sparse support?
  • Complex support?

Installation

This package can be installed just like any other Julia package. From the terminal, after starting the Julia REPL, run the following: julia julia> ] pkg> add RandNLA

Testing

To test the package, run the following command in the REPL: julia using Pkg Pkg.test(test_args=[<specific tests>])

Usage

Load the package as usual: julia using RandNLA

Design Principles

  1. Provide a consolidated suite of algorithms, both well established and state of the art
  2. Interface nicely with other Julia linear algebra packages such as Krylov.jl
  3. Leverage Julia to provide efficient implementations while staying adaptable and updateable

References

Papers

Julia Packages

Python Packages

C++ Pacakges

Roadmap

  1. Basic Sketching
    • Agnostic to RNG, i.e. use AbstractRNG type
    • Left vs. right sketching
    • Explicit vs implicit, i.e. is the sketching opeartor realized
    • Dense sketching operators
      • Rademacher
      • Uniform
      • Gaussian
      • Haar
      • Can allow for abitrary distribution as well
      • Methods for constructing these distributions?
    • Sparse sketching opeartors
      • Short-axis
      • Long-axis
      • iid
  2. Property estimation
    • Trace
    • Norm
  3. Decompositions
    • SVD
    • Nystrom (positive definite and indefinite)
    • ID
    • CUR
  4. Advanced Sketching
    • Leverage-score sampling
  5. Advanced algorithms
    • Randomized finite differences

Owner

  • Name: Cooper Simpson
  • Login: RS-Coop
  • Kind: user

Citation (CITATION.cff) 

cff-version: 1.2.0
message: "If you use this software, please cite it as below."
authors:
- family-names: "Simpson"
  given-names: "Cooper"
  orcid: "https://orcid.org/0000-0003-3922-2797"
title: "RandNLA.jl"
version: 0.0.0
doi: 10.1/2
date-released: 2023-12-01
url: "https://github.com/rs-coop/RandNLA.jl"
note: "Julia algorithms for randomized numerical linear algebra"

GitHub Events

Total
Last Year