Why an automated materials simulation management?
Most of the scientific simulations, especially those involving atomistic and electronic structure calculations usually remotely run on a HPC cluster. However, the necessary steps until one finally gives the execute/run command or submits the job (which will not stay too much in the queue) through the so-called batch job scripts are all time consuming. But even after submitting the job, one needs to check if the calculation is running, if it failed or finished successfully. Usually, depending on the commands inside the batch job script, the failure or successful termination of a calculation can be signalled back to the user even through email. But even so the whole procedure is fragmented and time-consuming at certain steps.
However, a collaboration between the Theory and Simulation of Materials (THEOS), the National Centre for Computational Design and Discovery of Novel Materials (MARVEL) at Ecole Polytechnique Federale de Lausanne and the Research and Technology Center at Robert Bosch LLC in Cambridge has succeeded to develop a ” flexible and scalable informatics’ infrastructure to manage, preserve, and disseminate the simulations, data, and workflows of modern-day computational science “.
They created the Automated Interactive Infrastructure and Database for Computational Science, AIIDA, a platform based on Python which helps with the hurdles and pain of job submissions. Instead of performing fragmented time consuming tasks the user will use one interface to control all the steps. All connections with the HPC clusters is automatic through the use of the so-called Aiida demon. As a consequence the user will have remotely access to the state of the calculations on any HPC cluster.
A truly interesting platform released under an open source MIT license worth to be tried out. Based on its capabilities for sure would be a great add-on for any computational materials science lab.
Check out more details in their published article .