Abstract
The production of dark matter particles from confining dark sectors may lead to many novel experimental signatures. Depending on the details of the theory, dark quark production in proton-proton collisions could result in semivisible jets of particles: collimated sprays of dark hadrons of which only some are detectable by particle collider experiments. The experimental signature is characterised by the presence of reconstructed missing momentum collinear with the visible components of the jets. This complex topology is sensitive to detector inefficiencies and mis-reconstruction that generate artificial missing momentum. With this work, we propose a signal-agnostic strategy to reject ordinary jets and identify semivisible jets via anomaly detection techniques. A deep neural autoencoder network with jet substructure variables as input proves highly useful for analyzing anomalous jets. The study focuses on the semivisible jet signature; however, the technique can apply to any new physics model that predicts signatures with anomalous jets from non-SM particles.
Highlights
SVJ mZ' = TeV, r inv0 50 100 150 200 250 300 350 400 450 500 mj (GeV) pD T 0.1EFP1 fraction of jets fraction of jets fraction of jets −3 −2 −1 η3 j Axis Minor
The study focuses on the semivisible jet signature; the technique can apply to any new physics model that predicts signatures with anomalous jets from non-SM particles
2.3 Feature selection Since the goal of this study is to find anomalies on the basis of tagging individual anomalous jets rather than anomalous events, a set of jet-level and jet substructure variables is determined for the training
Summary
Both multi-jet QCD and dark sector samples are generated using PYTHIA8 [19] with model parameters identical as in [20] and reconstructed with DELPHES [21] providing a detector response approximating that of CMS. To examine the robustness of autoencoder models to varying signal parameters, multiple SV jet samples were generated with values of the fraction of stable jet hadrons rinv ∈ {0.3, 0.5, 0.7} and the Z boson mass mZ ∈ {1.5, 2.0, 2.5, 3.0, 3.5, 4.0}. The selection efficiency was found to be at the level of 0.13% for the QCD events and between 0.2 and 15.3% for SV jets, depending on the signal model parameters. The number of QCD jets after applying selections is ≈100k (split between training, validation, and testing) and varies between 500 and 15k for SV jets, depending on the model parameters. 2.3 Feature selection Since the goal of this study is to find anomalies on the basis of tagging individual anomalous jets rather than anomalous events, a set of jet-level and jet substructure variables is determined for the training. A fraction of them were rejected due to poor signal-background discrimination
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
