Running Monte Carlo Simulations on a Linux Cluster

During the process of creating a set of CFAST input files with CData, the program generates a Windows and a Linux batch script to assist in running the multiple simulations. This wiki provides guidance on running the Linux script. While it is tailored to the NIST Linux cluster, the software requirements and procedure should be similar and adaptable to other systems.

System Requirements

Pre-compiled Linux version of CFAST are not available. Thus, you will need to clone the CFAST repo and build both CFAST and CData. See Cloning Repos for CFAST and Compiling CFAST for details.

At NIST, we use a bash script that is located in the fds repo called Utilities/Scripts/qfds.sh. This script allows you to run jobs quickly because it automatically writes and submits the PBS job control script. The script for running CFAST Monte Carlo simulations presumes use of this script. For more information on qfds.sh, see Installing and Running FDS on a Linux Cluster.

Once a CData input file is available (see Manuals/Monte_Carlo_Guide/Examples for a few examples), run CData with the -P option to generate the set of CFAST input files and batch scripts to run them. For a CData input file named Example.in with CData in its default location in the repo (typically ~/firemodels/cfast/Build/CData/intel_linux_64), the command looks like

~/firemodels/cfast/Build/CData/intel_linux_64/cdata7__linux_64 Example.in -P

From this command, a file named Example.sh will be created.