The world’s largest high-energy particle accelerator, the Large Hadron Collider, relies on sub-millisecond processing performed on massive amounts of data coming out of its particle detection system, which is due for a major upgrade in 2024. The old particle detection chips will be upgraded to RD53B chips with faster data transmission, allowing for more complicated data processing. The readout systems that interact with the particle detection chips, YARR and FELIX, need to be tested with the RD53B chips and debugged before the system is put in place. The goal of our research is to create an emulator of the RD53B chip that can produce dynamically generated pseudo-realistic data at the same rates that would be seen in the Large Hadron Collider without the need for heavy radiation. This will allow for readout software to be fully functional and debugged before the actual RD53B chips are fabricated and placed into the Large Hadron Collider. We are using Field-Programmable Gate Arrays (FPGAs) to mimic the hardware inside the real RD53B chips. In place of RD53B’s analog sensors, we have substituted digital logic that generates pseudo-realistic data because FPGAs cannot emulate analog hardware. The alpha version of the RD53B emulator with basic communication and pre-programmed data was completed in February. Recently, the beta version of the emulator with dynamically generated data was completed, and we have been testing communication between the emulator and FELIX. With the beta version of the RD53B emulator tested and verified by us, the developers of YARR and FELIX will use our hardware to help verify that their systems will provide accurate readouts from the real RD53B chips. The next steps for the RD53B emulator include a hardware data decompression accelerator, as well as any additional features requested by the YARR and FELIX teams.

Filtering the data produced by the Large Hadron Collider (LHC) is computationally challenging due to the sheer quantity of the data, on the scale of hundreds of terabytes per second. In the coming years, data production for the LHC is projected to increase by a factor of 15 with the high luminosity upgrade. Machine learning algorithms could provide pattern recognition capable of filtering data produced by the LHC. Specialized hardware could increase the throughput to match the data rates required by the LHC. We analyzed several cloud-based specialized hardware solutions including Amazon Web Service FPGAs, Microsoft’s Brainwave Service, Google’s TPU, and NVIDIA GPUs to compare the performance of each of them for particle physics application. The networks accelerated were trained on a variety of data including: Top vs QCD quark classification, Hadron calorimeter data, and electron energy regression. These experiments demonstrate the feasibility of machine learning algorithms in high throughput required situations such as high energy particle physics.

The 2019 Novel Coronavirus (COVID-19) pandemic has necessitated the implementation of various infectious disease control measures, including the closure of non-essential businesses, social distancing, and the virtualization of schools and universities. As final year nursing students at the University of Washington (UW) adjust to virtual learning and social distancing, certain students working in healthcare may also face the threat of contracting the virus. Little is known about the psychological implications of the COVID-19 pandemic on this population, and there is a need to fill this knowledge gap. This study first aims to capture the perceived stress levels of final year nursing students at the UW amid the COVID-19 pandemic. It secondly aims to explore associations between perceived stress and factors such as COVID-19 testing history, living situation, and healthcare work history. We administered an online survey to final year UW nursing students that inquires about their COVID-19 testing history, living situation, and healthcare work history since March 1, 2020. The survey also includes the 10-item Perceived Stress Scale, which questions students’ feelings and thoughts over the past month. We will analyze the data for associations between scores on the Perceived Stress Scale and students’ testing history, living situation, and work history. Overall, we expect to find moderate to high perceived stress levels among nursing students. We also anticipate that several factors may be associated with higher stress levels among nursing students, including having a history of COVID-19 testing, living with more individuals, and working more hours in healthcare positions. The results of this study may indicate a need for increased psychosocial support in final year nursing students at the UW, as they complete the nursing program and join the nursing workforce to help combat the COVID-19 pandemic.

In the ATLAS detector at the Large Hadron Collider (LHC), high energy quarks and gluons can be produced during proton-proton collisions. Individual quarks and gluons cannot be directly observed, however, when they enter the detector, interactions with the detector create a number or secondary particles called hadrons, which are made up of groups of quarks and gluons, that can be directly observed. A tagger is used to measure the secondary particles and classify them as coming from a quark or a gluon. The tagger uses several variables which are derived from detector data and machine learning algorithms. We analysed data from the detector and Monte Carlo simulation and compared them using the derived variables, which involves calculating ratios, Monte Carlo closure and scale factor, between distributions of the simulation and detector data for each variable. The Monte Carlo closure and scale factor between extracted detector samples and extracted Monte Carlo samples is expected to be close to 1, indicating the simulation models the data, with an uncertainty less than 10%. The results of this study give an analysis on how well these variables are able to classify the initial particle, and allow better calibration of the tagger parameters. Better classification allows for more precise measurements of physics processes at the LHC.

In 2024 the Large Hadron Collider will undergo upgrades that will dramatically increase the number of collisions occurring. In order to accommodate the increased bandwidth, the innermost detector as well as the readout hardware and data acquisition software will be upgraded. This readout chip, called the RD53, already has a preliminary model called the RD53a for which a software emulator already exists. The final design of the chip, the RD53b has recently been released and the software emulator for the RD53a needs to be updated in order to reflect the new specifications of the RD53b. Our research is seeking to create a software emulator for the RD53b readout chip. In order to create the software emulator, we are updating the existing software emulator for the RD53a and comparing the emulator's output with that of the physical hardware chip. We are currently working to create robust software tests of the old software emulator by running scans from the data acquisition software using the software emulator. Thus far the analog, digital and threshold scans for the RD53a emulator have been implemented. In this talk we will give an overview of the design of the software emulator, deployment progress and the further development plan. We will show how this software emulator can enable the faster development of new data acquisition software in preparation for the upgrades to the Large Hadron Collider.

Recent developments in Machine Learning have led to a number of different applications across a variety of fields. This rapid progress has been fueled by the increased performance of Graphics Processing Units (GPUs). Similar rapid developments can be seen happening in Quantum Computing Hardware. While still years behind logical computing, certain statistical models indicate that quantum computers will be able to outperform classical computers within years. However, due to the lack of high memory systems, all of the distributions have to be represented in low-dimensional space. Our work focuses on using classical computing to automatically find efficient feature maps that allow users to scale down real-world or established problems into low-dimensional space, which can then be loaded into quantum computers. Additionally, by creating a simple, modular design, we want to allow other researchers to have a simple interface to compare classical and quantum versions of algorithms to investigate if there are any benefits to using quantum computing over classical systems. We expect quantum computers to perform similarly, if not better, than similarly sized classical models, but still be outperformed by larger, more complex classical systems.

Machine learning (ML) is an important tool in analyzing huge data sets. There are various machine learning models in the realm of physics that do everything from identifying subatomic particles to predicting the energy of particle jets. Our project is focused on creating a benchmark that would be used to test different models and compare them with each other. Our goal is to host a service that would evaluate different metrics for a user-provided model and display the results. We have created a Yadage workflow that analyzes different tag taggers (i.e. ML models that identify top quarks). To evaluate the top taggers, we plotted their ROC (Receiver operating characteristic) curves. We then compared the AUC (Area Under the Curve) for each model. Our current goal is to run our workflows on REANA (Reproducible research data analysis platform) servers. We believe this project is not just restricted to the world of physics and can be extended to benchmark models from other disciplines as well such as health sciences, natual language processing, computer vision, etc. Similar Benchmarks could be created for different types of models which can be compared using a common dataset for a better comparison

The number of theoretical models for particle collisions has been steadily increasing, but creating and running a new program for each analysis is time-consuming. Often the individual steps in these analyses are applicable to a wide range of models. We created Recast-workflow, a project that preserves the steps in the analyses for truth-level reinterpretations (theoretically ideal models), as a solution to this issue. Developed as a Python3 package with a command line interface, this program generates runnable yadage workflows, defined by a yaml schema with instructions for running each step of the analysis. There are three main steps in a Recast workflow - generation, selection, and statistics. The generation step or “subworkflow” we implemented used MadGraph with Pythia, which takes a particle collision and uses a given model to generate the parton shower. We used Rivet, a software for validating data produced by Monte Carlo event generators, for the selection step and Contur or pyhf are the final steps we made to be used to find statistical confidence levels. Recast-workflow runs each stage using the yadage workflow engine in a docker encapsulated environment, and the output from each stage is passed to the subsequent one. This project will help researchers gauge the potential of interesting physics results from a region of phase space by running these generated workflows quickly without the computational complexity of a full reinterpretation. In the future, this can be made more accessible through a web interface and powerful through an expanded catalogue of steps.