Theoretical physicists often conjure up the image of peculiar loners who calculate on the blackboard and demand fixed seats on the sofa. Maybe they like people to believe in it, but the reality is that even in theory, these days more and larger collaborations are needed to get anywhere.
It may not be as big as the mega-experiments of the experimental particle physicists with over 3000 researchers, but the largest such collaboration that exists in theoretical physics today. GAMBIT has over 60 active participants. On 2 December this year, Anders Kvellestad from the Department of Physics was elected as the new head of GAMBIT.
In distinguished company
– This is a great honor, and very gratifying for us who work within theoretical physics in Oslo, says Are Raklev, head of the Section for theoretical physics, and himself a member of GAMBIT.
– GAMBIT has researchers from all over the world, from California in the west, to Nanjing in the east and Melbourne in the south, and includes institutions that are leaders in theoretical physics, such as the University of Cambridge and Imperial College London. In this illustrious company, Anders has been unanimously chosen to set the course for GAMBIT over the next three years, says Raklev.
As usual, physicists have fun creating weird acronyms. In addition to being a clever chess move, GAMBIT stands for "a Global And Modular Bsm Inference Tool". An explanation is in order: - The name is quite literally a description of what we do, says Anders Kvellestad.
GAMBIT tests models with a large number of parameters
– We are developing a tool to do statistical inference, i.e. draw conclusions from a data sample. In practice, such a tool looks at physical models of nature to find out how well they describe data from experiments, and what kind of properties these models have. This is done by selecting values for the parameters in the model and testing it against data in a systematic way that makes it possible to make statistically valid comparisons of models.
– The models that we as a collaboration are particularly interested in are models for the very smallest building blocks of the Universe, new physics that can replace the so-called standard model in particle physics, and explain phenomena such as dark matter and dark energy. A significant problem with realistic models is that they have many parameters. This makes it very difficult to explore the models adequately.
137 million CPU hours
Since 2012, GAMBIT has been working to develop a tool to solve this problem. The first publicly available version of the code was released in May 2017 and is available as open source. To exploit the potential of GAMBIT, Kvellestad and Pat Scott at Imperial College London/University of Queensland have received a large allocation from the EU consortium PRACE (Partnership for Advanced Computing in Europe) of a total of 137 million CPU hours over three years.
Can you run GAMBIT on your laptop?
– It is possible to test simpler models on a regular computer, but when we look for traces of dark matter in results from the LHC and astrophysics measurements, it takes millions of CPU hours. A lot of what we do from day to day in GAMBIT is to try to compute results faster, with less computing power. We also have a strict policy of reuse of data. All our physics articles come with an open data set that can be downloaded by other interested researchers, so that they do not have to do the same job again on the same model.
Nuclear physics and infection tracking
Does anything useful for the rest of the world come out of a bunch of theoretical physicists sitting around coding?
– The models we are interested in, which describe the Universe's behavior at its smallest scales, must be considered fundamental basic research. But the methods and tools we develop are generic and modular, and can be used on models and data from other disciplines. We are already involved in a research project in nuclear physics, and are in contact with researchers who are interested in seeing if this can be used on models for the spread of infection, which also have very many free parameters, ends Kvellestad.
The leadership role in GAMBIT is a fixed-term position of three years, and Kvellestad will take over the role from 1 January 2021. He will succeed Pat Scott from Imperial College London / University of Queensland. With him as deputy head he has Martin White from the University of Adelaide.
- See also: The GAMBIT project page at UiO.no.