A Bayesian Framework for Dark Photon Searches at LDMX

Sammanfattning

Despite decades of experimental searches, dark matter, a hypothetical form of matterwith highly suppressed electromagnetic interactions, has yet to be detected. The absenceof any detection in the 10 GeV–TeV mass range has motivated searches in lower massranges, reaching sub-GeV levels. Due to the Lee-Weinberg bound, a thermal-relic darkmatter particle at this mass scale requires a new force mediator to bring the annihilationcross-section into agreement with the observed relic density. This force mediator is dubbedthe dark photon, and several experiments designed to produce and detect it have beenbuilt or are currently under construction.The main objective of this thesis is to develop a Bayesian framework for inferring thecoupling strength and mass of the dark photon using data from the upcoming Light DarkMatter eXperiment (LDMX), a fixed target experiment where dark photons are expectedto be produced in electron-tungsten collisions. The framework is validated on MadGraph5-simulated data sets to quantify LDMX’s (Phase II) detection capabilities and parameterinferencepower. This task is accomplished using dynamic nested sampling, designed forestimating posteriors and calculating marginal likelihoods. The secondary objective is toinvestigate whether the degeneracy between the dark photon mass and interaction modelcan be resolved through a combined analysis of the recoil-electron’s transverse momentum|pT | and total energy E. The degeneracy is explored by comparing Bayes factors computedfrom the marginal likelihoods of the different models. The framework recovers the darkphoton mass and coupling accurately for benchmarks corresponding to thermal relic targetsfor complex scalar dark matter with R ≡ mA′/mχ = 2.5, while identifying benchmarks withR = 2.2, producing fewer expected signal events, as being below detection threshold. Usingtwo-dimensional (E, |pT |) kinematics, the Bayes factor comparison breaks the mass-modeldegeneracy between the interaction model groups ([KM, C, A] and [M, E]), with maximumBayes factors of lnK = 591. One-dimensional analysis of E and |pT | alone still separatesthe model groups, though less decisively, with maximum Bayes factors of lnK ≈ 42 for Ealone and lnK ≈ 22 for |pT | alone. The introduction of a relic-target prior on gf furtherseparates KM from [C, A], at the cost of additional assumptions.