arXiv subjects

Prasanth Shyamsundar

Publications and source records attributed to Prasanth Shyamsundar.

24 records · Page 2Linked to original sources

arXiv2020

OASIS: Optimal Analysis-Specific Importance Sampling for event generation

We propose a technique called Optimal Analysis-Specific Importance Sampling (OASIS) to reduce the number of simulated events required for a high-energy experimental analysis to reach a target sensitivity. We provide recipes to obtain the optimal sampling distributions which preferentially focus the event generation on the regions of phase space with high utility to the experimental analyses. OASIS leads to a conservation of resources at all stages of the Monte Carlo pipeline, including full-detector simulation, and is complementary to approaches which seek to speed-up the simulation pipeline.

hep-ph↗

arXiv2020

InClass Nets: Independent Classifier Networks for Nonparametric Estimation of Conditional Independence Mixture Models and Unsupervised Classification

We introduce a new machine-learning-based approach, which we call the Independent Classifier networks (InClass nets) technique, for the nonparameteric estimation of conditional independence mixture models (CIMMs). We approach the estimation of a CIMM as a multi-class classification problem, since dividing the dataset into different categories naturally leads to the estimation of the mixture model. InClass nets consist of multiple independent classifier neural networks (NNs), each of which handles one of the variates of the CIMM. Fitting the CIMM to the data is performed by simultaneously training the individual NNs using suitable cost functions. The ability of NNs to approximate arbitrary functions makes our technique nonparametric. Further leveraging the power of NNs, we allow the conditionally independent variates of the model to be individually high-dimensional, which is the main advantage of our technique over existing non-machine-learning-based approaches. We derive some new results on the nonparametric identifiability of bivariate CIMMs, in the form of a necessary and a (different) sufficient condition for a bivariate CIMM to be identifiable. We provide a public implementation of InClass nets as a Python package called RainDancesVI and validate our InClass nets technique with several worked out examples. Our method also has applications in unsupervised and semi-supervised classification problems.

stat.ML↗

arXiv2020

Singularity Variables for Missing Energy Event Kinematics

We discuss singularity variables which are properly suited for analyzing the kinematics of events with missing transverse energy at the LHC. We consider six of the simplest event topologies encountered in studies of leptonic W-bosons and top quarks, as well as in SUSY-like searches for new physics with dark matter particles. In each case, we illustrate the general prescription for finding the relevant singularity variable, which in turn helps delineate the visible parameter subspace on which the singularities are located. Our results can be used in two different ways - first, as a guide for targeting the signal-rich regions of parameter space during the stage of discovery, and second, as a sensitive focus point method for measuring the particle mass spectrum after the initial discovery.

hep-ph↗

arXiv2019

Optimal event selection and categorization in high energy physics, Part 1: Signal discovery

We provide a prescription to train optimal machine-learning-based event selectors and categorizers that maximize the statistical significance of a potential signal excess in high energy physics (HEP) experiments, as quantified by any of six different performance measures. For analyses where the signal search is performed in the distribution of some event variables, our prescription ensures that only the information complementary to those event variables is used in event selection and categorization. This eliminates a major misalignment with the physics goals of the analysis (maximizing the significance of an excess) that exists in the training of typical ML-based event selectors and categorizers. In addition, this decorrelation of event selectors from the relevant event variables prevents the background distribution from becoming peaked in the signal region as a result of event selection, thereby ameliorating the challenges imposed on signal searches by systematic uncertainties. Our event selectors (categorizers) use the output of machine-learning-based classifiers as input and apply optimal selection cutoffs (categorization thresholds) that depend on the event variables being analyzed, as opposed to flat cutoffs (thresholds). These optimal cutoffs and thresholds are learned iteratively, using a novel approach with connections to Lloyd's k-means clustering algorithm. We provide a public, Python 3 implementation of our prescription called ThickBrick, along with usage examples.

physics.data-an↗

arXiv2019

Kinematic Focus Point Method for Particle Mass Measurements in Missing Energy Events

We investigate the solvability of the event kinematics in missing energy events at hadron colliders, as a function of the particle mass ansatz. To be specific, we reconstruct the neutrino momenta in dilepton $t\bar{t}$-like events, without assuming any prior knowledge of the mass spectrum. We identify a class of events, which we call extreme events, with the property that the kinematic boundary of their allowed region in mass parameter space passes through the true mass point. We develop techniques for recognizing extreme events in the data and demonstrate that they are abundant in a realistic data sample, due to expected singularities in phase space. We propose a new method for mass measurement whereby we obtain the true values of the mass parameters as the focus point of the kinematic boundaries for all events in the data sample. Since the masses are determined from a relatively sharp peak structure (the density of kinematic boundary curves), the method avoids some of the systematic errors associated with other techniques. We show that this new approach is complementary to previously considered methods in the literature where one studies the solvability of the kinematic constraints throughout the mass parameter space. In particular, we identify a problematic direction in mass space of nearly 100% solvability, and then show that the focus point method is effective in lifting the degeneracy.

hep-ph↗

arXiv2019

Measuring the mass, width, and couplings of semi-invisible resonances with the Matrix Element Method

We demonstrate the use of the Matrix Element Method (MEM) for the measurement of masses, widths, and couplings in the case of single or pair production of semi-invisibly decaying resonances. For definiteness, we consider the two-body decay of a generic resonance to a visible particle from the Standard Model (SM) and a massive invisible particle. It is well known that the mass difference can be extracted from the endpoint of a transverse kinematic variable like the transverse mass, $M_T$, or the Cambridge $M_{T2}$ variable, but measuring the overall mass scale is a very difficult problem. We show that the MEM can be used to obtain not only the absolute mass scale, but also the width of the resonance and the tensor structure of its couplings. Apart from new physics searches, our results can be readily applied to the case of SM $W$ boson production at the CERN Large Hadron Collider (LHC), where one can repeat the measurements of the $W$ properties in a general and model-independent framework.

hep-ph↗