Skip to content

kose-y/DistStat.jl

Repository files navigation

DistStat.jl

codecov build Actions Status

DistStat.jl: Towards unified programming for high-performance statistical computing environments in Julia

Installation Information

Dependencies:

  • Julia >= 1.4
  • An MPI installation (tested on OpenMPI, MPICH, and Intel MPI)
  • MPI.jl >= 0.15.0 (and its dependencies)
    • Select proper MPI backend when building MPI.jl, as described in this page)
  • CustomUnitRanges.jl
  • See Project.toml

For CUDA support:

  • CUDA >= 9.0
  • CUDA.jl, GPUArrays.jl (and their dependencies)
  • CUDA-aware MPI installation (of OpenMPI, MPICH, and Intel MPI, only OpenMPI supports CUDA)
  • MPI.jl should be built with the environment variable JULIA_MPI_BINARY=system; see this page).

To install the package, run the following code in Julia.

using Pkg
pkg"add https://github.com/kose-y/DistStat.jl"

Examples

Examples of nonnegative matrix factorization, multidimensional scaling, and l1-regularized Cox regression is provided in the directory examples/. Settings for multi-gpu experiments and multi-instance cloud experiments are also provided.

Acknowledgement

This work was supported by AWS Cloud Credits for Research. This research has been conducted using the UK Biobank Resource under application number 48152.

Citation

Ko S, Zhou H, Zhou J, and Won J-H (2020+). DistStat.jl: Towards Unified Programming for High-Performance Statistical Computing Environments in Julia. arXiv:2010.16114.

About

Distributed computation for statistics

Resources

License

Stars

Watchers

Forks

Packages

No packages published