Giter Site home page Giter Site logo

kose-y / dist-primal-dual Goto Github PK

View Code? Open in Web Editor NEW
0.0 5.0 0.0 60 KB

Source code to accompany "Easily parallelizable and distributable class of algorithms for structured sparsity, with optimal acceleration".

License: MIT License

Python 100.00%

dist-primal-dual's Introduction

dist-primal-dual

This repository contains the Python package dist_pd for distributed computation of primal-dual problems in multiple GPUs with optimal acceleration, published in

  • S. Ko, D. Yu, J. Won. Easily Parallelizable and Distributable Class of Algorithms for Structured Sparsity, With Optimal Acceleration. J. Comput. Graph. Stat. 2019:28(4), 821- 2833.

Based on Tensorflow, we implemented distributed dense and sparse matrices, basic operations on them, and implementations of primal-dual algorithms based on these operations.

Installation

On Docker (recommended)

We have all the prerequisites readily installed in our docker containers. See docker/. The docker version contains an option to also download bigger datasets, which is shown to be scalable on multiple GPU devices.

On Linux

Prerequisites: Python 3, with tensorflow >=1.2 installed. The code is tested on Python 3.5. Tensorflow is not automatically installed by our setup script. To install tensorflow on your system, you can follow the instructions in this link. If you wish to run the code on Nvidia GPU(s), please read instructions on additional setup here.

One may run the following to install the code:

git clone https://github.com/kose-y/dist_pd.git
cd dist_pd
python setup.py install

Running the examples

You can run the expreriments in our paper using the code in examples/. Read examples/README.md for more information.

Obtaining the data

The "small" datasets can currently be obtained from Dropbox:

https://www.dropbox.com/s/d7tpa8insoq844g/ogrp_100_100_10_5000_X.mat
https://www.dropbox.com/s/t14b7p4dq66up9k/ogrp_100_100_10_5000.mat


https://www.dropbox.com/s/jc96pvzq4lo58ot/Zhu_1000_10_5000_20_0.7_100_X.mat
https://www.dropbox.com/s/tl8ode7ny8elole/Zhu_1000_10_5000_20_0.7_100.mat

If you are using Docker, these datasets come with the package. You may also choose to download a bigger dataset from Docker Hub: see docker/README.md.

If you are on linux, you can download the files using the script download_data.py.

Citation

If you use our package in your research, please cite the following item:

@article{doi:10.1080/10618600.2019.1592757,
    author = {Seyoon Ko and Donghyeon Yu and Joong-Ho Won},
    title = {Easily Parallelizable and Distributable Class of Algorithms for Structured Sparsity, with Optimal Acceleration},
    journal = {Journal of Computational and Graphical Statistics},
    volume = {28},
    number = {4},
    pages = {821-833},
    year  = {2019},
    publisher = {Taylor & Francis},
    doi = {10.1080/10618600.2019.1592757},
    URL = { 
        https://doi.org/10.1080/10618600.2019.1592757
    },
    eprint = { 
        https://doi.org/10.1080/10618600.2019.1592757
    }
}

dist-primal-dual's People

Contributors

kose-y avatar won-j avatar

Watchers

 avatar  avatar  avatar  avatar  avatar

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    ๐Ÿ–– Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. ๐Ÿ“Š๐Ÿ“ˆ๐ŸŽ‰

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google โค๏ธ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.