Skip to content
This repository has been archived by the owner on Jan 28, 2020. It is now read-only.

Latest commit

 

History

History
81 lines (60 loc) · 3.98 KB

README.md

File metadata and controls

81 lines (60 loc) · 3.98 KB

Build Status Maintainability Test Coverage

NetLogo-HPC

Application to run Netlogo on HPC clusters which use SLURM for scheduling. Uses by default NetLogo 6.0.4.

Installing the application

  1. Assuming you are on a linux based HPC
  2. Change into the directory containing your model.
  3. Download the tar.gz file You can do that directly from the command line with wget https://github.com/daveol/NetLogo-HPC/releases/download/0.1.4/NetLogo-HPC-0.1.4.tar.gz
  4. Unpack the archive with tar -zxvf NetLogo-HPC-0.1.4.tar.gz

Using the application

  1. Change into the directory containing your model.
  2. Start screen or tmux program. If you are not familiar with screen and what it does, please see the wikipedia page on GNU screen
  3. execute netlogo-hpc --model model.nlogo --experiment experiment --table results.csv
  4. You can now detach from your session, and wait for the simulations to finish. If you would like to be notified by email when it finishes or crashes, you can use the following incantation : netlogo-hpc --model model.nlogo --experiment experiment --table results.csv && mail -s "Simulation finished" [email protected] < /dev/null || mail -s "Simulation crashed" [email protected] < /dev/null

Using extensions

Create a 'extensions' directory in the model experiment, and place any extensions you need there. Please note that at the moment saving experimental output in non-NetLogo native data types is not supported. For eample, you can not use a Table object as a output of a reporter to be saved in the output file.

Building the application

Please note, building has only been tested on Linux, but will probaly work on other unix type OSs, like OSX.

Building an application bundle

./sbt packArchive

Which will create the file target/NetLogo-HPC-*.tar.gz which contains the dependencies to run NetLogo-HPC. This be unpacked by runningtar xvf /path/to/NetLogo-HPC-*.tar.gz which then makes bin/netlogo-hpc which runs the application.

Using with NetLogo versions other than 6.0.4

By default, netlogo-hpc uses NetLogo 6.04 for its internal working. If you wish to use a older (or newer) version, change

libraryDependencies += "org.nlogo" % "netlogo" % "6.0.4"

in the build.sbt file to the desired version. Please note that the version you provide in the file must match the GitHub version tag.

Documentation for netlogo-hpc

 Arguments:
  --help
      shows this help information

  --model [file]
      Specify the NetLogo model file

  --experiment [name]
      Specify the experiment to run

  --table [file]
      Table to write the metrics from simulations to

Advanced options:
  --application-id [string] (optional)
      Specifies the application id to check the network against, it will default to a random uuid

  --client-id [string] (optional)
      Specifies the client id to use for receiving and sending messages, will override to 'controller'
      when the controller role is specified otherwhise it will default to a random uuid

  --worker
      Manualy creates a worker that  conncets to the controller. You will not need this during normal use.
      
  --controller
      Default mode, creates a controller node first, then spawns the workers.

--connect [uri] or --listen [uri]
      Connect or Bind to a URI in the style of 'tcp://[ip]:[port]'

 Scheduler arguments:
  --concurrent [number of allowed tasks] (optional)
      Specifies the amount of tasks that are allowed to be concurrently deployed on the cluster, it defaults to eight tasks.