GAMBIT-light — particle physics power to the people
Since the first public release in 2017, GAMBIT has become the leading tool for large-scale parameter estimation studies within Beyond-the-Standard-Model physics, a research field studying new theories of physics at the subatomic level.
However, numerical exploration of models with many free parameters is a computational challenge common across many research fields. So through the support of the Advanced User Support framework, Senior Engineers Maiken Pedersen and Marcin Krotkiewski (the University Centre for Information Technology, University of Oslo), and GAMBIT Leader Anders Kvellestad (the Section for Theoretical Physics, University of Oslo), have developed GAMBIT-light — a lightweight version of GAMBIT for use in fields outside particle physics.
All physics-related code stripped away
Building on the modular design of the original GAMBIT code, in GAMBIT-light, all physics-related code has been stripped away while the core machinery of GAMBIT remains. Among other things, this includes features such as two-level parallelisation, a collection of highly efficient sampling and optimisation algorithms, logging, file output to multiple formats, and continuation of aborted jobs.
Users can easily connect their domain-specific target functions, in the form of C, C++, Fortran or Python libraries, to GAMBIT-light in a diskless manner and with no need for modification or recompilation of GAMBIT-light itself. This allows users to focus on their code while leaving the heavy optimisation task and bookkeeping to GAMBIT-light. Further, the GAMBIT-light code repository has been set up such that relevant developments in the main GAMBIT repository automatically propagate to GAMBIT-light, allowing the GAMBIT Collaboration to maintain and further develop GAMBIT-light with minimal overhead.
The first public version of GAMBIT-light will soon be available on GitHub. Among the first research projects to use GAMBIT-light will be the interdisciplinary project Pandemics, particles and parameter (s)paces, recently financed by the Faculty of Mathematics and Natural Sciences at the University of Oslo. This project is a collaboration between researchers at the National Institute for Public Health (NIPH/FHI) and the Section for Theoretical Physics at the University of Oslo. The aim is to tackle computational challenges common to both particle physics simulations and simulations of disease spread in populations by combining the capabilities of GAMBIT-light with the development of tailored regression algorithms.