Skip to content

Splits up NLOGO experiments among MPI tasks and executes them in parallel on a cluster.

License

Notifications You must be signed in to change notification settings

acfisher/mpirun_nlogo

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

12 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

mpirun_nlogo_experiment
======================

Divides up nlogo behaviorspaces among mpi processes and runs them in parallel.


Author : Aaron Fisher <[email protected]>

License : GNU GPL 3 ; See LICENSE.txt for details.


About
-----

This script make it possible to run large behaviorspace experiments on MPI clusters.  It relies on a significant amount of code from the project forked from ahrenberg/split_nlogo_experiment.  However this approach does not create an xml file per behaviorspace combination making it much more scalable as the file systems tend to get bogged down with 1000s of files in a directory.  

Requirements
------------

This software requires:
* An MPI enabled cluster with some version of mpirun.  It has been tested with openmpi.  
* Python 2.7 or higher with the mpi4py library installed.  
* NLOGO (easily installed in your home directory)
* A version of java that can run your version of NLOGO (easily installed in your home directory)

Talk to your system administrator if your system does not meet these requirements.

Basic usage
~~~~~~~~~~~
Start by downloading and unzipping your required version of NLOGO in your home directory:
wget http://ccl.northwestern.edu/netlogo/5.3.1/NetLogo-5.3.1-64.tgz
tar xvzf NetLogo-5.3.1-64.tgz

Start by cloning or unpacking your install of mpirun_nlogo:
git clone https://github.com/acfisher/mpirun_nlogo.git

Make a directory in your home directory to hold your parameter study:
mkdir param_study

If you are using the PBS batch system for your cluster (most clusters) copy the nlogo.pbs file to the param_study directory:
cp NetLogo-5.3.1/nlogo.pbs param_study/NLOGO jar file for your 

Copy the collect_data.py file to the param study directory:
cp NetLogo5.3.1/collect_data.py param_study/

Edit the nlogo.pbs file in the param study directory.  Add any module commands that may be necessisary for your cluster.  Edit the paths for your base/home directory, param study location, nlogo jar file for your netlogo install, command to launch java, mpirun command, and mpirun_nlogo.py location.

I'll add to this later.

About

Splits up NLOGO experiments among MPI tasks and executes them in parallel on a cluster.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 96.6%
  • Shell 3.4%