Strengthening the quality and user base of the research software ESPResSo for particle-based simulations

• Applicant: Dr. Rudolf Weeber
• Subject Area: Statistical Physics, Nonlinear Dynamics, Complex Systems, Soft and Fluid Matter, Biological Physics
• Term: since 2023
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 528726435

ESPResSo is a versatile research software for particle-based simulations, used by many groups worldwide. The target applications are coarse-grained models in such divers areas as soft matter, biophysics, and process engineering. Its main strengths are the breadth of simulation algorithms provided and its flexible Python interface, allowing detailed control of the simulation process and making it easy to use ESPResSo together with other scientific Python packages. As an open-source project with a transparent development process including code reviews, the FAIR principles are well respected. Within this project, we aim to improve ESPResSo's quality and interoperability with other codes, to enlarge its user and developer base, and to open up opportunities for its use in new fields of research. Long-range solvers are key to simulations in soft matter science but also challenging to understand and implement. In terms of quality assurance, we plan to refactor the relevant long-range code parts such they are more easily readable and clearly reflects the expressions found in the original papers describing the algorithms. Stringent tests will be developed for cases, where no analytical reference solution is available. In terms of extending ESPResSo's scope, we envision ESPResSo as a tool for developing and deploying machine-learned potentials. To this end, we will create a framework which lets users couple ESPResSo's efficient neighbor list generation to their machine-learned models. As a demonstrator, we will couple ESPResSo to popular packages for those potentials. Furthermore, we intend to make ESPResSo more extensible, in particular with regards to Monte Carlo schemes and, using the well-maintained open-source Walberla package, for coupling of lattice-Boltzmann and electrokinetics. We want to enable users without C++ skills to modify and extend these functionalities. Well-documented showcases will serve as starting points. Through package definitions and container technology, we will make it easier to install and run ESPResSo on HPC systems, and to provide stable, tested environments for generating reproducible results. To lower the entry barrier for new developers, we will significantly extend ESPResSo's developer documentation, and prepare video tutorials on setting up a development environment. In developer workshops, we will help developers design and improve their own extension projects. In summary, through a combination of quality assurance, better extensibility and flexibility, and user and developer documentation and training, we will put ESPResSo on a more sustainable footing, and make it an attractive choice for applications for which is has not yet been used.

• DFG Programme: Research Grants
• Co-Investigator: Professor Dr. Christian Holm