Physical Sciences

A search for heavy right-handed Majorana neutrinos is performed with the ATLAS detector at the CERN Large Hadron Collider, using the 140  fb−1 of proton–proton collision data at s=13  TeV collected during Run 2. This search targets tt¯ production, in which both top quarks decay into a bottom quark and a W boson, where one of the W bosons decays hadronically and the other decays into an electron or muon and a heavy neutral lepton. The heavy neutral lepton is identified through a decay into an electron or muon and another W boson, resulting in a pair of same-charge same-flavor leptons in the final state. This paper presents the first search for heavy neutral leptons in the mass range of 15–75 GeV using tt¯ events. No significant excess is observed over the background expectation, and upper limits are placed on the signal cross sections. Assuming a benchmark scenario of the phenomenological type-I seesaw model, these cross section limits are then translated into upper limits on the mixing parameters of the heavy Majorana neutrino with Standard Model neutrinos. © 2024 CERN, for the ATLAS Collaboration 2024 CERN

Weakly interacting massive particles (WIMPs) may interact with a virtual pion that is exchanged between nucleons. This interaction channel is important to consider in models where the spin-independent isoscalar channel is suppressed. Using data from the first science run of the LUX-ZEPLIN dark matter experiment, containing 60 live days of data in a 5.5 tonne fiducial mass of liquid xenon, we report the results on a search for WIMP-pion interactions. We observe no significant excess and set an upper limit of 1.5 × 10−46 cm2 at a 90% confidence level for a WIMP mass of 33 GeV/c2 for this interaction.

Abstract: Recently, combinations of generative and Bayesian deep learning have been introduced in particle physics for both fast detector simulation and inference tasks. These neural networks aim to quantify the uncertainty on the generated distribution originating from limited training statistics. The interpretation of a distribution-wide uncertainty however remains ill-defined. We show a clear scheme for quantifying the calibration of Bayesian generative machine learning models. For a Continuous Normalizing Flow applied to a low-dimensional toy example, we evaluate the calibration of Bayesian uncertainties from either a mean-field Gaussian weight posterior, or Monte Carlo sampling network weights, to gauge their behaviour on unsteady distribution edges. Well calibrated uncertainties can then be used to roughly estimate the number of uncorrelated truth samples that are equivalent to the generated sample and clearly indicate data amplification for smooth features of the distribution.