Python based workflows for model construction

NAISS 2023/7-22



Principal Investigator:

Paul Erhart


Chalmers tekniska högskola

Start Date:


End Date:


Primary Classification:

10304: Condensed Matter Physics

Secondary Classification:

10403: Materials Chemistry



The present proposal is seeking cloud resources to maintain (1) a web server that is based on the Django framework ( as well as (2) two jupyterhub instances to be used for teaching and research, respectively. Web server ---------- During the last years we have developed several web applications that are served via and allow * accessing the properties of atoms under pressure (a collaboration with Assoc. Prof. Martin Rahm at the Department of Chemistry and Chemical Engineergin at Chalmers) * accessing the dielectric functions of binary alloys * constructing special-quasi random structures for alloys using the icet code ( as a backend * constructing nanoparticle (so-called Wulff) shapes using the wulffpack code ( as a backend We are planning on adding more applications in the coming years. These applications are commonly based on the Django framework (, which requires modest server-side resources. At the moment the server is run on ssc.large instance. Assuming 365 day-24 hour operation this amounts to 4380 coins. Jupyterhub servers ----------------- Our method development and analysis workflow heavily relies on Python and jupyter notebooks. We also heavily use this setup, e.g., in the TIF345 "Advanced Simulation and Machine Learning" course that is part of the Master's program in Physics. To support this workflow we have for many years run a jupyterhub server on a ssc.xlarge instance, which has been heavily used. Since this has caused problems due to overload, e.g., during lab sessions, we would like to split the servers for group internal and teaching use. Assuming 365 day-24 hour operation this amounts to 17,520 coins.