TY - JOUR

T1 - Dijet Resonance Search with Weak Supervision Using √s=13  TeV pp Collisions in the ATLAS Detector

AU - Collaboration, ATLAS

AU - Barton, A.E.

AU - Bertram, I.A.

AU - Borissov, G.

AU - Bouhova-Thacker, E.V.

AU - Fox, H.

AU - Henderson, R.C.W.

AU - Jones, R.W.L.

AU - Kartvelishvili, V.

AU - Long, R.E.

AU - Love, P.A.

AU - Muenstermann, D.

AU - Parker, A.J.

AU - Sanderswood, Izaac

AU - Smizanska, M.

AU - Tee, A.S.

AU - Walder, J.

AU - Wharton, A.M.

AU - Whitmore, B.W.

AU - Yexley, Melissa

PY - 2020/9/21

Y1 - 2020/9/21

N2 - This Letter describes a search for narrowly resonant new physics using a machine-learning anomaly detection procedure that does not rely on signal simulations for developing the analysis selection. Weakly supervised learning is used to train classifiers directly on data to enhance potential signals. The targeted topology is dijet events and the features used for machine learning are the masses of the two jets. The resulting analysis is essentially a three-dimensional search A→BC, for mA∼O(TeV), mB,mC∼O(100  GeV) and B, C are reconstructed as large-radius jets, without paying a penalty associated with a large trials factor in the scan of the masses of the two jets. The full run 2 √s=13  TeV pp collision dataset of 139  fb-1 recorded by the ATLAS detector at the Large Hadron Collider is used for the search. There is no significant evidence of a localized excess in the dijet invariant mass spectrum between 1.8 and 8.2 TeV. Cross-section limits for narrow-width A, B, and C particles vary with mA, mB, and mC. For example, when mA=3  TeV and mB≳200  GeV, a production cross section between 1 and 5 fb is excluded at 95% confidence level, depending on mC. For certain masses, these limits are up to 10 times more sensitive than those obtained by the inclusive dijet search. These results are complementary to the dedicated searches for the case that B and C are standard model bosons.

AB - This Letter describes a search for narrowly resonant new physics using a machine-learning anomaly detection procedure that does not rely on signal simulations for developing the analysis selection. Weakly supervised learning is used to train classifiers directly on data to enhance potential signals. The targeted topology is dijet events and the features used for machine learning are the masses of the two jets. The resulting analysis is essentially a three-dimensional search A→BC, for mA∼O(TeV), mB,mC∼O(100  GeV) and B, C are reconstructed as large-radius jets, without paying a penalty associated with a large trials factor in the scan of the masses of the two jets. The full run 2 √s=13  TeV pp collision dataset of 139  fb-1 recorded by the ATLAS detector at the Large Hadron Collider is used for the search. There is no significant evidence of a localized excess in the dijet invariant mass spectrum between 1.8 and 8.2 TeV. Cross-section limits for narrow-width A, B, and C particles vary with mA, mB, and mC. For example, when mA=3  TeV and mB≳200  GeV, a production cross section between 1 and 5 fb is excluded at 95% confidence level, depending on mC. For certain masses, these limits are up to 10 times more sensitive than those obtained by the inclusive dijet search. These results are complementary to the dedicated searches for the case that B and C are standard model bosons.

KW - article

KW - body weight

KW - boson

KW - hadron

KW - human

KW - mass spectrometry

KW - punishment

U2 - 10.1103/PhysRevLett.125.131801

DO - 10.1103/PhysRevLett.125.131801

M3 - Journal article

VL - 125

JO - Phys Rev Lett

JF - Phys Rev Lett

SN - 1079-7114

IS - 13

M1 - 131801

ER -