Search for Higgsinos in Compressed Mass Spectra using Neural Networks in $sqrt{s} = SI{13}{TeV}$ $pp$ Collisions with the ATLAS Detector

Super-symmetric extensions of the Standard Model featuring Higgsinos with compressed mass spectra and masses near the electroweak scale may solve the hierarchy problem. These models provide a higgsino-like neutralino as a Dark Matter candidate that can also be consistent with current cosmological evidence.

This analysis looks for the pair-production of higgsino-like charginos (${ ilde{chi}}_1^pm$) and neutralinos (${ ilde{chi}}_2^0$) in $pp$ collisions at the LHC in association with a jet from initial state radiation (ISR). The search attempt to extend the current limit set by LEP at $m_{{ ilde{chi}}_1^pm} leq 95$ GeV, focusing on the mass splitting $Delta m=m_{{ ilde{chi}}_2^0}-m_{{ ilde{chi}}_1^0}$ near 0.3--1.5 GeV, where the chargino is semi-long-lived (with $c au sim SI{1}{mm}$).

We use two fully-connected neural networks (FCNN) to exploit the event-level signatures associated with the ISR jet and the undetected ${ ilde{chi}}_1^0$, and the track-level signatures of soft displaced pions $pi^pm$ from ${ ilde{chi}}_1^pm/{ ilde{chi}}_2^0$ decay (aligning with the significant $E_mathrm{T}^mathrm{miss}$).

The estimated exclusion limit based on Monte Carlo and blinded data in the full ATLAS Run 2 dataset (139 $mathrm{fb}^{-1}$) will be presented.