tomalrussell / openmole

Workflow engine for exploration of simulation models using high throughput computing

OpenMOLE (Open MOdeL Experiment) has been developed since 2008 as a free and open-source platform. It offers tools to run, explore, diagnose and optimize your numerical model, taking advantage of distributed computing environments. With OpenMOLE you can explore your already developed model, in any language (Java, Binary exe, NetLogo, R, SciLab, Python, C++...).

• The stable version is available on openmole.org.
• A fresh build of the developement version is available on next.openmole.org.

OpenMOLE is distributed under the AGPLv3 free software license.

OpenMOLE by example

Imagine that you want to understand and/or optimize some parameters of an executable that you generally set in an empirical or arbitrary manner. Then embed the executable in OpenMOLE (5 minutes), use one of the distributed genetic algorithms provided by OpenMOLE (5 minutes) and launch the workflow on a distributed execution environment with thousands of machines (1 minute).

To summarize, you can design large scale distributed programs reusing legacy code and advanced numeric methods in approximately 10 minutes.

Try it!

To checkout OpenMOLE you can play with to the demo site (this site is wiped out every few hours). You should click on the little cart and try out some of the market place examples.

OpenMOLE Features:

• Expressive syntax – A Domain Specific Language to describe your exploration processes,
• Transparent distributed computing – Zero-deployment (no installation step) approach to distribute the workload transparently on your multi-core machines, desktop-grids, clusters, grids, ...
• Works with your programs – Embed user’s executables (Java, Binary exe, NetLogo, R, Scilab, Python, C++, ...),
• Scalable – Handles millions of tasks and TB of data,
• Advanced methods – Advanced numerical experiments (design of experiments, optimization, calibration, sensitivity analysis, ...).

OpenMOLE Avanced Features:

• Workflow plateform – Design scientific workflows that may use legacy code,
• Distributed genetic algorithms - Distribute the computation of your fitness functions,
• Distributed computing - A high level aproach to distributed computing.