Session N11: Analysis Tools and Techniques

FETA: Flow-Enhanced Transportation for Anomaly Detection

Resonant anomaly detection is a promising framework for model-independent searches for new particles. Weakly supervised resonant anomaly detection methods compare data with a potential signal against a template of the Standard Model (SM) background inferred from sideband regions. We propose a means to generate this background template that uses a normalizing flow to create a mapping between high-fidelity SM simulations and the data. The flow is trained in sideband regions with the signal region blinded, and the flow is conditioned on the resonant feature (mass) such that it can be interpolated into the signal region. To illustrate this approach, we use simulated collisions from the Large Hadron Collider (LHC) Olympics Dataset. We find that our flow-constructed background method has competitive sensitivity with other recent proposals and can therefore provide complementary information to improve future searches.   

Parallelizing Event Generation in PYTHIA with Python

High energy particle physics requires event generation for Monte Carlo simulation studies, often using the PYTHIA library. Depending on the process being studied, generating these events sequentially is adequate both in terms of number of events for the required statistics as well as the time required to generate these events. However, it is advantageous for rare processes to generate these events in parallel. Pythia has the ability to do this using openMP and C Pragmas. In this study we moved to python to simplify the code development process. We look at the unique challenges when doing event parallelization with python and how to structure the code to resolve them.   

Insitu Jet Energy Scale Calibration using the ATLAS Detector

When a highly energetic gluon or quark flies out of an LHC collision, it will produce a collimated shower of high energy hadrons called a jet. Such objects are extremely prevalent in searches involving the decay of heavy particles such as Higgs, top, and more exotic particles. Therefore, it is critical to model the detector response and p_T of jets accurately.  The jet energy can be calibrated using events in which a jet is balanced against a photon or Z boson, but only up to moderate p_T.  The multi-jet balance method allows for the calibration of high p_T jets using events in which a single high-p_T jet is recoiling against a system already calibrated jets. The in-situ calibration is a technique for measuring the difference in data and MC jet response and applying a calibration factor until the modeling reaches good agreement.   

A Review of NEST Models, and Their Application to Improvement of Particle Identification in Liquid Xenon Experiments

The Noble Element Simulation Technique (NEST) is a widely adopted and reliable simulation toolkit for rare-event physics searching experiments using liquid xenon. The key part of signals and backgrounds simulation in liquid xenon is accurate modeling of charge and light production. To achieve this, based on various experimental data, NEST has its signal and background production models fit using comprehensive and empirical formulae for the average charge and light yields and their variations. NEST also simulates the final scintillation pulses and exhibits the correct energy resolution as a function of the particle type, the energy, and the electric fields. After vetting of NEST against raw data, with several specific examples pulled from XENON, ZEPLIN, LUX / LZ, and PandaX, we interpolate and extrapolate its models to draw new conclusions on the properties of future detectors (e.g., XLZD's). It turns out that for the future detectors, the discrimination power of electronic recoil backgrounds from a potential nuclear recoil signal (especially WIMP dark matter) can be improved by another order of magnitude (99.95% discrimination) from the oft-quoted one, with a reasonably high photon detection efficiency and achievable drift fields.   

Data Production at ICARUS

ICARUS is a liquid argon time projection chamber detector which serves as the far detector for the Short Baseline Neutrino (SBN) program at the Fermilab campus. In a normal year of data-taking, ICARUS runs at about 0.8 Hz when running with beam and at about 0.5 Hz for off-beam events. The effective annual rate is 0.725 Hz, producing about 23 million raw data events. In terms of data space occupancy, given an event size of about 162 MB/event, the data can run upward of about 4 PB in disk space. In its first run, ICARUS produced around 0.5 PB worth of data in the span of about a month and is expected to take more than 3 PB worth of data in three months for its second run. All these data need to be processed via the decoding and filtering stages to ensure data quality and to pass through event reconstruction algorithms before being used for physics analysis. This talk will discuss all the steps taken at ICARUS to turn raw data into analysis worthy data.   

Developing Multidimensional Voronoi Histograms for NOvA's Physics Program

In instances where computational time and complexity scale with the number of bins in a histogram, the curse of dimensionality can make it difficult to choose a binning for data in more than a few dimensions. In one dimension, variable-width bins preserve resolution in regions of interest while reducing resolution in other regions to limit the total number of bins used to capture the data. In this talk, I will show a novel technique to extend the concept of variable bin widths to higher dimensions using Voronoi diagrams. I will also show applications of this technique to produce histograms useful for tuning cross-section models in NOvA and for the NOvA multidetector oscillation fit program.   

MWD Algorithm Optimization in Mu2e's Stopping Target Monitor

The Mu2e experiment will search for coherent neutrino-less conversion of muons to electrons in muonic aluminum. The Mu2e Stopping Target Monitor employs two detectors optimized to detect photons from 100 keV up to a few MeV. One is a high-purity Germanium diode, and the other is a Lanthanum Bromide (LaBr) scintillator detector; they will measure x-rays and gamma rays ejected from the Mu2e stopping target to count the number of stopped muons. The waveform data from these detectors will be put through zero-suppression and moving-window-deconvolution (MWD) algorithms to obtain an energy histogram. In this presentation, we present the MWD parameters that optimize the energy resolution for LaBr using data from the ELBE test beam, in particular the Y-88 decay line at 1836 keV, as a surrogate for the 1809 keV line from muon capture on Al, which has a rate proportional to the number of stopped muons.