One of the main bottlenecks when searching for new physics at the Large Hadron Collider (LHC) at CERN is the simulation of collisions, and in particular the reconstruction of jets, meaning the showers of hadrons that come from the production of high-energy quarks and/or gluons in the final state. This reconstruction is done by using so-called jet-algorithms, where the current state-of-the-art implementation is in the FastJet code (fastjet.fr). Today the relative slowness of this jet-reconstruction is what is limiting us from investigating new physics models with LHC data in the full detail we would like to.
We have a project focused on investigating speed-ups of these algorithms, in particular the anti-kT algorithm which is the one most used by the LHC experiements, using modern computational techniques on multi-core CPU or GPU set-ups with parallelisation using OpenMP. In the project the student will need to acquire both a good understanding of relevant computational techniques using C++, and an understanding of the particle physics basis for the calculation.
If you are interested and can't wait for a spot in the master programme to try yourself, you can find a succinct description of this jet-challenge in the following document.