Improving CMS Contact Interaction Limits using Bayesian Statistics

Nikita Elkin, Illinois Mathematics and Science Academy
Kaushal Gumpula, Illinois Mathematics and Science Academy
Timothy Mou, Illinois Mathematics and Science Academy

Abstract

In the Standard Model, quarks and leptons are understood to be fundamental particles. However, they have been theorized to be composite, made up of constituent particles called preons. These constituent particles would interact in contact interactions according to an energy scale Λ, which would result a non-resonant enhancement in the dilepton invariant mass spectrum when compared to current Standard Model predictions. Data from the CMS detector is used to set lower limits on Λ.

Our analysis aims to improve the current method of limit-setting. The problem with the current method is that in cases of destructive interference, the yield in the signal bins can be less than the Standard Model prediction, making it impossible to interpret contact interactions as a signal process. In our approach, we combine the yields of the signal and Drell-Yan processes, as the total number of events will always be positive. We do this by parameterizing the combined signal and Drell-Yan yields as a function of 1/Λ^2 and setting a limit directly on Λ. We will present the promising progress we have made on this method and discuss the challenges still remaining with calculating expected limits.

 
Apr 26th, 10:25 AM Apr 26th, 10:40 AM

Improving CMS Contact Interaction Limits using Bayesian Statistics

Room A147

In the Standard Model, quarks and leptons are understood to be fundamental particles. However, they have been theorized to be composite, made up of constituent particles called preons. These constituent particles would interact in contact interactions according to an energy scale Λ, which would result a non-resonant enhancement in the dilepton invariant mass spectrum when compared to current Standard Model predictions. Data from the CMS detector is used to set lower limits on Λ.

Our analysis aims to improve the current method of limit-setting. The problem with the current method is that in cases of destructive interference, the yield in the signal bins can be less than the Standard Model prediction, making it impossible to interpret contact interactions as a signal process. In our approach, we combine the yields of the signal and Drell-Yan processes, as the total number of events will always be positive. We do this by parameterizing the combined signal and Drell-Yan yields as a function of 1/Λ^2 and setting a limit directly on Λ. We will present the promising progress we have made on this method and discuss the challenges still remaining with calculating expected limits.