An automation toolkit for complex simulation tasks¶
FabSim3 is a Python-based automation toolkit for scientific simulation and data processing workflows, licensed under the BSD 3-clause license. It is developed as part of the SEAVEA project http://www.seavea-project.org, and is part of the VECMA Toolkit http://www.vecma-toolkit.eu.
Among other things, FabSim3 supports the use of simple one-liner commands to:
- Enable execution of simulation and analysis tasks on supercomputers.
- Establish and run coupled models using the workflow automation functionalities.
- Organize input, output and environment information, creating a consistent log and making it possible by default to repeat/reproduce runs.
- Perform large ensemble simulations (or replicated ones) using a one-liner command.
Users can perform complex remote tasks from a local command-line, and run single jobs, ensembles of multiple jobs, and dynamic workflows through schedulers such as SLURM, PBSPro, LoadLeveller and QCG. FabSim3 stores machine-specific configurations in the repository, and applies it to all applications run on that machine. These configurations are updated by any contributor who feels that a fix or improvement is required.
FabSim3 relies strongly on Fabric (http://www.fabfile.org, shown to work with versions 1.5.3 and 1.10.0) and PyYAML. It has been used to run simulation workflows on supercomputers such as ARCHER, SuperMUC, Carthesius, Eagle, as well as local clusters and desktops.
FabSim3 is publicly available at: http://www.github.com/djgroen/FabSim3.
The accompanying software paper can be found here: https://doi.org/10.1016/j.cpc.2016.05.020.
The main plugins for FabSim3 include:
- FabMD, focused on molecular dynamics.
- FabFlee, focused on agent-based modelling.
- FabUQCampaign, focused on UQ ensemble sampling.
- FabDummy, a dummy plugin used for testing the toolkit.
- FabMUSCLE, a preliminary integration with the MUSCLE3 coupling toolkit.