ddp-batch

This repository offers specialized scripts designed to facilitate the execution of ab initio random structure searching (AIRSS) package and ephemeral data derived potential (EDDP) packages. These scripts are tailored for use with batch queuing systems or job schedulers, specifically the SLURM (Simple Linux Utility for Resource Management) and SGE (Sun Grid Engine), on cluster and high-performance computing (HPC) environments. The provided scripts streamline the process of initializing, managing, and monitoring AIRSS and EDDP tasks, ensuring efficient utilization of the cluster/HPC resources.

Installation

install the package with:

git clone git@github.com:sehunjoo/ddp-batch.git

run setup.sh script:

./setup.sh

Documentation

update-chain

• This script is for generating chain-batch script, which is a chain script for batch queueing system. It basically replaces spawn and despawn with spawn-batch and despawn-batch.

• Usage

  Generate the chain-batch with:

  ./update-chain

spawn-batch

• This script can be used to submit multiple jobs to a cluster using a batch queueing system. This script is for creating a job script, submitting the job script, monitoring the job, and resubimtting/killing the job.

• This script is applicable to all applications, but is specifically designed for use with airss.pl and crud.pl in AIRSS package and forge in EDDP package.

• It creates jobscript-program.sh and despawn-batch files in the working directory. As soon as a job is started, a .spawnpid.* file is created for each subjob. The file contains details of each subjob.

• It automatically resubmits the job if the job is terminated due to the wall clock limit.

• For example,

  nohup spawn-batch -command airss.pl -mpinp 4 -seed Li -max 100 &

• The following command line can be used to stop the spawn-batch script and the running jobs. The spawn-batch script detects the STOP file, runs despawn-batch and stops itself.

  touch STOP

farm-batch

• This script is exactly same as spawn-batch script. This script is merely a counterpart to the farm script that is included in the EDDP package.

References

• Pickard, Ephemeral data derived potentials for random structure search, 2022.
• Salzbrenner et al., Developments and further applications of ephemeral data derived potentials, 2023.