Extracting more information from LHC data through a Bayesian framework

by Ezequiel Alvarez

Thursday 20 Jun 2024, 11:30 → 12:30 Europe/Rome

281

Abstract:
How much information is encoded in the LHC data? How much of this information are we currently extracting?

Throughout this talk we will revolve around these questions by studying Bayesian techniques to extract information from the data.  Frameworks designed for this purpose leverage information beyond mere data by incorporating mathematical and statistical tools. Bayesian statistics, in particular, excels at modeling the inner structure of data, thus allowing a better inclusion of prior information and its associated uncertainties as a means to improve information extraction. This approach also includes mechanisms for assessing the unbiasedness and consistency of the results relative to the data. We show with examples the application of Bayesian techniques to efficiently extract information about the underlying physics at the LHC. Potential applications in improving data-driven methods and analyses for processes such as pp > hh > bbbb are discussed, although the principles can be applied to re-evaluate a variety of observables. This methodology can take advantage of correlations at the event-by-event level, as well as previously underutilized attributes such as continuity and unimodality of certain distributions.  We propose the concept of the Information Frontier, arguing that it will play a central role in the upcoming years.