CASA SDL NQFT
The
"CASA"
SDL Numerical Quantum Field Theory is a joint endeavor by
JSC, and IAS-4,
IAS_6. Our focus is on numerical simulations of
quantum field theories.
Quantum field theories play a central role in describing phenomena from elementary particle physics to solid state physics, as well as in models of neuronal systems. In the vast majority of cases, the complexity and sometimes non-linearity of the theories and models make simulations the only promising means for predicting experimental outcomes or material properties. Such simulations not only represent important applications on the HPC systems installed and planned at JSC. Rather, they are often the most demanding calculations that fully utilize the systems in terms of parallelism and the performance of the individual computing nodes. Simulations in numerical field theory face a threefold challenge:
- the increasing complexity of the hardware makes porting of existing and developing new efficient software significantly harder and more complex,
- the advancing frontier of target observables increase e.g. the complexity of the operators that need to be implemented or require new methods to be developed, and
- the standard methods have advanced significantly, e.g. from simple linear solvers like e.g. BiCGstab to involved multi-grid algorithms like DD-αAMG
These challenges imply a new level of complication in the development and use of HPC systems at the Forschungszentrum Jülich (FZJ) and any HPC system worldwide. Consequently, significant effort is required in order to enable efficient simulations in the future.
Such efforts pay off in multiple ways. Firstly, they enable the scientific community to perform simulations essential for their scientific needs. Secondly, they provide excellent feedback on the architectures that can be used in a co-design effort to improve their effectiveness for these applications and, due to similarities in the models or methods, also for a broader range of users.
The footprint of simulation data generated by the scientific community and stored at the Jülich Supercomputing Centre’s (JSC) file servers is rapidly growing. Efficiently managing and sharing simulation data is, therefore, another important subject. Here, the existing grid structure needs to be urgently updated, both to deal with the size of the data available now but also to enable the management of new and systematically different simulation results, ensuring reproducibility of published works.
Team
Our team consists of a subset of JSC's SDL, complemented by staff members of the participating IAS institutes.
Prof. Dr. Stefan KriegHead of Division HPC for Quantum Systems
PI in Helmholtz Information Program 1, Topic 1
Building 07.1 / Room 243c+49 2461/61-8715
Last Modified: 05.04.2025