Exotic particle searches
Despite the success of the Standard Model, the questions it leaves open (e.g. the nature of Dark Matter, the stability of the Higgs mass, the origin of the flavour structure, etc) point to physics beyond it. Searching for that physics at the LHC remains one of the central goals of the field, and it is where the bulk of our ATLAS effort on exotic searches goes. With the full Run-2 and Run-3 datasets in hand, the emphasis is shifting away from the signatures that were easiest to reach first, and towards final states that are experimentally harder: particles that decay far from the collision point, and model-agnostic searches looking for signals that no theorist has thought about yet.
Our work is organised along three directions.
Anomaly detection
A large number of model-specific searches have now returned null results, which makes a complementary, model-agnostic approach worth pursuing. We work with machine-learning models, such as autoencoders and masked-encoder architectures, to learn what Standard Model data looks like, and then flag the events that do not quite fit as "anomalous". Our emphasis is on dilepton final states with additional activity (dilepton + X): because of the relatively low backgrounds, this is one of the most sensitive places to look for something unexpected.
Long-lived particles: dark-photon signatures
Many extensions of the Standard Model contain a dark sector coupled only weakly to ordinary matter. A common realisation is a dark photon, the mediator of a broken dark U(1). If the mixing is small the dark photon is long-lived, and if it is light its decay products are strongly boosted that could produce a collimated cluster of leptons or light hadrons (a "dark-photon jet") reconstructed in the calorimeter or the muon spectrometer. Our most recent result used the full Run-2 dataset to search for these signatures in vector-boson-fusion Higgs production for the first time; combined with earlier ATLAS results, it excludes Higgs branching fractions to dark photons above 10% over a wide range of decay lengths. Work now focuses on extending the reach with Run-3 data and on dedicated triggers.
Long-lived particles decaying to tau leptons
Displaced signatures to taus have not been searched unlike electrons and muons. This is because this is a very challenging experimental signature. Our approach has been to fix the reconstruction first. We have developed track-classification and recurrent-neural-network identification algorithms for displaced hadronic taus, which recover a large fraction of the lost efficiency in a model-independent way, together with dedicated displaced-tau triggers for Run 3 that substantially increase acceptance at high displacement. A Run-3 search now in preparation.
Recent public results include:
- "Search for light long-lived neutral particles from Higgs boson decays via vector-boson-fusion production from *pp* collisions at √s = 13 TeV with the ATLAS detector", Eur. Phys. J. C 84 (2024) 719, arXiv:2311.18298
- "Search for heavy Majorana or Dirac neutrinos and right-handed *W* gauge bosons in final states with charged leptons and jets in *pp* collisions at √s = 13 TeV with the ATLAS detector", Eur. Phys. J. C 83 (2023) 1164
Previous searches by the group focused on heavy diboson resonances (WW/WZ/ZZ), right-handed charged bosons and heavy neutrinos, and dark-matter candidates decaying to boosted jets.
The group consists of Dr Sara Alderweireldt, Jennifer Curran, Dr Yanyan Gao, Prof Sinead Farrington, Afroditi Kyprianou, Prof Christos Leonidopoulos, Dr Alex Sopio, Sana Tabaza, Dr Ben Wynne and Dr Zhongyukun Xu.
The activities are led by Dr Yanyan Gao, Prof Sinead Farrington and Prof Christos Leonidopoulos.