MadMiner: Machine Learning-Based Inference for Particle Physics
- PDF / 3,407,093 Bytes
- 25 Pages / 595.276 x 790.866 pts Page_size
- 33 Downloads / 239 Views
(2020) 4:3
ORIGINAL ARTICLE
MadMiner: Machine Learning‑Based Inference for Particle Physics Johann Brehmer1 · Felix Kling2,3 · Irina Espejo1 · Kyle Cranmer1 Received: 8 August 2019 / Accepted: 3 January 2020 © Springer Nature Switzerland AG 2020
Abstract Precision measurements at the LHC often require analyzing high-dimensional event data for subtle kinematic signatures, which is challenging for established analysis methods. Recently, a powerful family of multivariate inference techniques that leverage both matrix element information and machine learning has been developed. This approach neither requires the reduction of high-dimensional data to summary statistics nor any simplifications to the underlying physics or detector response. In this paper, we introduce MadMiner , a Python module that streamlines the steps involved in this procedure. Wrapping around MadGraph5_aMC and Pythia 8, it supports almost any physics process and model. To aid phenomenological studies, the tool also wraps around Delphes 3, though it is extendable to a full Geant4-based detector simulation. We demonstrate the use of MadMiner in an example analysis of dimension-six operators in ttH production, finding that the new techniques substantially increase the sensitivity to new physics.
Introduction Precision measurements at the Large Hadron Collider (LHC) experiments search for direct and indirect signals of physics beyond the Standard Model. Statistically, this requires constraining a typically high-dimensional parameter space, for instance the Wilson coefficients in an effective field theory (EFT) or the couplings and masses in a supersymmetric model. The data going into these analyses consist of a large number of observables, many of which can carry information on the parameters of interest.
* Johann Brehmer [email protected] Felix Kling [email protected] Irina Espejo [email protected] Kyle Cranmer [email protected] 1
Center for Data Science and Center for Cosmology and Particle Physics, New York University, New York, NY 10003, USA
2
Department of Physics and Astronomy, University of California, Irvine, CA 92697, USA
3
SLAC National Accelerator Laboratory, 2575 Sand Hill Road, Menlo Park, CA 94025, USA
The relation between model parameters and observables is typically best described by a suite of computer simulation tools for the hard interaction, parton shower, hadronization, and detector response. These tools take as input assumed parameters of the physics model, for instance a particular value for the Wilson coefficients of an EFT, and use Monte Carlo methods to sample hypothetical observations. Unfortunately, they do not directly let us solve the inverse problem: given a set of observed events, it is not possible to explicitly calculate the likelihood of such a measurement as a function of the theory parameters. This intractability of the likelihood function is a major challenge for particle physics measurements. Particle physicists have developed a range of techniques for this problem of likelihood-fr
Data Loading...
