Skip to content
master
Switch branches/tags
Go to file
Code

Latest commit

 

Git stats

Files

Permalink
Failed to load latest commit information.
Type
Name
Latest commit message
Commit time
 
 
 
 
src
 
 
 
 
 
 
 
 
 
 
 
 
 
 

ADMM-Spark

This repository is an experimental implementation of the Alternating Direction Method of Multipliers (ADMM) algorithm on Apache Spark. Currently the implemented model is the L1-regularized logistic regression.

Dependencies

ADMM-Spark relies on the following libraries:

The sbt building tool will handle these dependencies. See the next section for details.

Installation

ADMM-Spark uses sbt (http://www.scala-sbt.org/index.html) to build the project and manage the dependencies. To create the jar file, first enter the project directory, and then type the sbt command.

cd ADMM-Spark
sbt

In sbt console, input make to build native C++ code that helps to speed up the computation, and then enter package to compile the Scala source code and create the jar file. If one also wants the jar file to include the dependent libraries so that it can be directly used in Spark applications, the assembly command can be used instead.

> make
Make dynamic library from C++ code
...
> package
[info] Packaging <PROJECT_DIR>/target/scala-2.11/admm_2.11-1.0.jar ...
[info] Done packaging.
> assembly
...
[info] Done packaging.

About

Implementation of ADMM algorithm on Apache Spark

Resources

Releases

No releases published

Packages

No packages published