Workshop on orchestrating multiphysics simulations with 4C & QUEENS

Our colleagues Dr.-Ing. Matthias Mayr and Dr.-Ing. Sebastian Brandstäter organized and delivered a two-day workshop on 4C Multiphysics & QUEENS to introduce fellow researchers in computational science and engineering to these open-source research software frameworks. The workshop sessions gave the opportunity for hands-on experience in both tools as well as in their combined application to leap beyond predictive simulation of multiphysics systems.

Together with collaborators from the Technical University of Munich and the Helmholtz-Zentrum Hereon, the workshop was organized as a two-day event: the first day focused on the 4C multiphysics simulation framework, the second day introduced the solver-independent multi-query analysis tool QUEENS and its interplay with 4C. 

If you have missed the workshop, you can work through the course material by yourself. Material and instructions are available in this GitHub repository.

The Event

The workshop has been embedded into the 2025 UKACM GACM Autumn School on "Open-Source Codes for High-Performance Computing" with a particular focus on "Enabling Technologies for Computational Engineering and Open-Source Research Software". The autumn school has been organized by Prof. Jelena Ninic, Dr. Hoang-Giang Bui and Dr. Hassan Liravi (University of Birmingham) as well as Prof. Alexander Popp and Dr.-Ing. Matthias Mayr (University of the Bundeswehr Munich) under the auspices of the UK Association for Computational Mechanics (UKACM) and the German Association for Computational Mechanics (GACM). It covered both theoretical and practical aspects of open-source initiatives, including Nektar++ (delivered by researchers from Imperial College London, Newcastle University, and King's College London), 4C & QUEENS (delivered by researchers form the University of the Bundeswehr Munich, Technical University of Munich, and Helmholtz-Zentrum Hereon), and OpenFOAM (delivered by researchers from Aston University and the University of Birmingham).

The Software Frameworks

• 4C (Website: https://4c-multiphysics.org, GitHub: https://github.com/4C-multiphysics/4C) is a parallel multiphysics research code to analyze and solve a plethora of physical problems described by ordinary or partial differential equations. Its development is driven by challenging research questions and real-world problems, for which existing tools do not suffice, either due to the lack of capabilities or due to falling short of accuracy or performance. 4C offers ready-to-use simulation capabilities for a variety of physical models, encompassing single-field systems such as solids, fluids, and scalar transport phenomena, as well as coupled multiphysics systems. Its modular architecture also supports research in mathematical modeling and the development of novel numerical methods.
• QUEENS (Website: https://queens-py.org, GitHub: https://github.com/queens-py/queens) is a Python framework for solver-independent multi-query analyses of large-scale computational models. It fulfills two major objectives: first, it provides a broad suite of analysis techniques, ranging from parameter studies to Bayesian inverse problems, with an emphasis on probabilistic approaches; and second, it ensures robust management of multi-query settings in high-performance computing (HPC) environments, including simulation scheduling, error handling, and communication management, all in a solver-agnostic fashion. The application of modern probabilistic methods to advanced physics-based models is enabled through a high-level Python abstraction, thereby facilitating comprehensive analyses of complex systems within HPC workflows.
• The 4C & QUEENS Ecosystem: Physical and probabilistic modeling are distinct but complementary domains, each requiring specialized expertise to address their respective challenges. The 4C and QUEENS frameworks provide an interoperable ecosystem that bridges these domains: 4C enables high-fidelity physical modeling of single- and multiphysics systems, while QUEENS offers capabilities for simulation analytics and probabilistic modeling. Their co-development and interoperability allow researchers to develop models in their respective domains independently, while facilitating the seamless fusion of multiphysics simulations with probabilistic analysis techniques. The workshop will offer tutorials demonstrating how computational mechanics researchers can exploit this interoperability to enhance predictive capabilities and accelerate scientific discovery. Topics will range from simulation analytics to Bayesian uncertainty quantification, illustrated with practical examples based on 4C multiphysics models of various complexity to lay a solid foundation for future innovative research.

A heartfelt thank you also to our PhD students Regina Bühler and Bishr Maradni for their energetic support before and during the workshop!