Quantum Machine Learning (QML) has shown early promise over the last few years. From simple AI algorithms to sophisticated neural networks, quantum computers have produced results that are as good as or better than their classical counterparts. However, all of these models have to deal with the memory bottleneck, which is caused by the limited number of qubits in near-term quantum devices. We instead propose a hybrid neural network that works by sandwiching any QML algorithm between two classical neural networks, using PyTorch. The design allows for an automatic scaling of quantum algorithms to inputs and outputs of any size, addressing the bottleneck issue, but it also provides an easy way of comparing classical algorithms to quantum ones and an expandability to other, more advanced classical scenarios. Additionally, the software supports the usage of configuration files, which allow for fast-paced testing of basic hypotheses, without the need of writing custom code.

The Large Hadron Collider at CERN is built to accelerate particles to high speeds and make them collide. The curved trajectory of these particles is recorded by detecting electrical charges deposited on multiple layers of tracking equipment as a particle travels through the detector. The resulting patterns are then used to reconstruct the path traveled by the particle. The more the collisions, the higher the number of charge depositions to detect. This exponential increase in data is predicted to be the main issue with the next run of the LHC experiment, where we are set to test higher energy interactions. As traditional tracking algorithms do not scale well with this increase in data, supplementing them with machine learning provides a promising solution. Scaling high-energy particle tracking in the LHC to process petabytes of data is the focus of the Exa.TrkX project, which our study is part of. In our research, we study the robustness of Exa.Trkx models and algorithms against noise and misalignment. Robustness is judged by analyzing performance metrics like purity and efficiency of pairs of charge deposits or “doublets''. Purity is defined as the ratio of true-positives over positives, and efficiency is defined as the ratio of true positives over the number of true values. A true deposit belongs to the same trajectory as the one we are comparing it with. In this presentation, I will discuss how we proved robustness against noise by observing that the change in doublet purity and efficiency was a trivial decrease of 0.3 and 0.2 percent respectively. Our research makes sure that the Exa.TrkX models can be applied to actual LHC data. We do this by proving that the models are not affected by real-life impurities in the data like noise.

In recent years, as machine learning algorithms develop, particle physics have started to use machine learning algorism to solve physics problems. One of the most complicated problems is to find BSM (Physics beyond the Standard Model) signal in the standard model background. GAN (generative adversarial network) is one of the most interesting ideas in computer science today. Two models are trained simultaneously by an adversarial process: A generator uses to generate fake data and a discriminator uses to distinguish between real data and fake data. My goal is to train a conditional GAN algorithm to generate specific fake data and apply a classifier with LHC Olympic dataset in order to find anomalies (BSM). The dataset has two regions, signal region (might have anomalies) and sideband region (only background). The final stage is to generate the fake data in the sideband region and use a binary classifier with the signal region to find if there are any anomalies in the signal region.

Many beyond the Standard Model (BSM) theories suggest the existence of more fundamental scalar fields and associated Higgs bosons than previously thought, the standard model Higgs being the lightest and most easily discovered. As independently testing each of the many heavy Higgs theories would be inefficient, in this talk, I describe the use of an alternative, model-independent, generic approach to model exclusion and validation in the search for a generic heavy Higgs boson theorized to have both 4-dimensional (dim-4) and effective 6-dimensional (dim-6) interactions with the Standard Model particles. If the generic heavy Higgs is connected with BSM physics at the scale of a few teraelectronvolts (TeV), we will see an excess beyond Standard Model predictions in several observables at high transverse momentum. I will discuss the role of the dim-4 and dim-6 operators at play, before expounding upon the simulations used to characterize generic heavy Higgs production in proton-proton collisions and the corresponding Large Hadron Collider (LHC) data. Channels, signal regions, and control regions are defined within an event data model analysis framework to maximize the significance of any potential result. In addition, cuts are made on key variables, such as the invariant mass of the hadronic W boson in the same-sign dilepton signal region and the invariant mass of the heavy Higgs in the trilepton signal region in order to separate the known physics from any potential new physics. If discovered, a generic heavy Higgs would validate a key part of many BSM models and help to focus such theoretical work, while also founding an entirely new area of research for experimentalists.

Classification of particle jets that originate from light or heavy flavor quarks is an important task in determining the nature of particles created in collisions. This data can be first preprocessed into a list of particle tracks, which then can be processed sequentially. Recurrent Neural Networks (RNNs) are a powerful tool that is used to process sequential information, and we develop several RNNs such as the Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) networks to classify the data into background and signal jets. The transformer is also a new, powerful model that is used to process sequential information, and is said to be more powerful and more efficient to train than the LSTM. We study the performance of the transformer compared to RNN models by training the transformer model on the jet flavor dataset, and then comparing the AUC values of each respective model. We find that the transformer outperforms the LSTM models in classifying light and heavy flavor quarks, and that it does so with less training parameters.

In the ATLAS detector, high energy quarks and gluons can be produced in proton-proton collisions (p-p collisions). Quarks and gluons, elementary particles of the standard model of particle physics, have a quantity called color charge which subjects them to color confinement; only color neutral groups, called hadrons, can be observed and thus a single quark or gluon cannot be observed. This poses a challenge in determining whether a quark or a gluon was produced in a collision. Each quark or gluon produced will create a multitude of hadrons within several femtometers of the collision, which are then grouped together to form jets. This analysis uses five variables from the jet: transverse momentum, energy correlation, track multiplicity, and jet width, to differentiate quark-initiated jets and gluon-initiated jets. Previous calibrations only used track multiplicity based tagger, whereas this calibration will use a boosted decision tree based tagger. My task is to provide an analysis of the scale factor between simulation and detector data (with a desired value of 1) and systematic uncertainties. Preliminary results show that the scale factor is about 0.8-1.2 with systematics around 8%.

The success of the Standard Model of particle physics has incentivized attempts to find new theories that go beyond the Standard Model. Computer simulations of the particle colliders and their detectors are required to evaluate the validity of these new theories for experimental research. RECAST is a framework for reinterpreting Large Hadron Collider analyses using Yadage computational workflows. RECAST-workflow builds on RECAST in order to run truth-level reinterpretations which achieve much faster results by sacrificing complexity. It also allows for workflows to be modularized through subworkflows which encapsulate each step (generation, selection, analysis). The aim of our work is to improve upon the existing integration of the event generator MadGraph to support custom models, as well as adding the additional generators Sherpa and Herwig. This is done through modifying the existing python code base and creating portable Docker containers to encapsulate the programs in each step. This tool was applied to the SVJ model for both t-channel and s-channel and the results compared to published simulations which we anticipate to match. In this talk we will demonstrate the utility of RECAST for fast and modular particle simulations, highlighting the new generators and how they can be applied to study new interesting models.

The ATLAS detector is the largest general-purpose particle detector at the Large Hadron Collider, surrounding a site where protons collide at near-light speed and recording the resulting expulsion of particles and energy with an array of sub-detectors. Particles travelling outward from the collision deposit charge in clusters of cells (‘cluster images’) along the electromagnetic and hadronic calorimeters of the ATLAS detector. A convolutional neural network is used to analyze cluster images generated by incident Pions and classify whether the incident Pions are neutral or charged. For each type of Pion, a dense neural network is used to analyze cluster images and predict the energy of the incident Pions. In this project, I implemented a mixture density network in place of the dense neural network to analyze cluster images and predict the energy of incident Pions as well as the associated uncertainty of the energy for each cluster. The energy resolution of each cluster contains important information for the purposes of tracking particles’ trajectories throughout the detector, especially as collisions become more energetic and particles with overlapping tracks become more numerous; propagating the uncertainty from each cluster to the particle tracks would result in more accurate measurements by the ATLAS detector, allowing the standard model of physics to be studied under greater scrutiny.