Unveiling dark matter with soft displaced vertices at LHC

The progress of mankind depends on its understanding of nature. The knowledge about elementary particles and fundamental forces gained during the 20th century has shaped our society today. With the discovery of the Higgs boson 10 years ago at CERN`s LHC collider, our current theory, the Standard Model, was experimentally confirmed. Yet many questions remain unanswered, suggesting that the Standard Model needs to be extended. Astrophysical observations suggest that about 85% of the matter in the universe, dark matter, consists of unknown particles. These particles are very difficult to detect. Various techniques have been developed for their direct detection, and many attempts have been made to find the dark matter in collisions at the LHC, but so far without success. This may indicate some complexity of the dark matter content. In our project, we focus on models that predict at least two new massive particles with a small mass difference. In this configuration, dark matter would elude our previous attempts to find it. Nevertheless, such scenarios lead to the observed relic density of dark matter via the mechanism of co-annihilation. The standard technique for searching for dark matter at the LHC is based on conservation of energy, with undetected dark matter particles manifesting themselves through missing energy. In the models targeted by our project, the slightly heavier co- annihilation partner produced in the collision would decay into the dark matter particle and a few known and detectable particles. However, due to the small mass difference between the two new particles, the visible particles are produced with very little energy, typically ignored in previous searches. We will analyze data taken and being collected by the CMS experiment from 2016 to 2025, taking advantage of the fact that the heavier particle is often predicted to be long-lived. Therefore, we will attempt to reconstruct its decays away from the original collision point. Such a goal is very challenging because the visible particles that form the displaced vertex have very low energy and they must be found in an overwhelming background of simultaneous collisions. To accomplish this task, we plan to use advanced machine learning methods.