Session B15: Undergraduate Research I

Live

We report on a design of a computationally efficient parallelized framework to analyze the large data volume generated by our muon tomography detector. The detector consists of plastic scintillator counters with photomultiplier tubes and a CAMAC data acquisition system with analog-to-digital, time-to-digital, and scaler modules. Our system relies on a delay-line technique which reduces the number of readout channels significantly. We discuss applications of machine-learning inspired techniques for generating higher fidelity images compared to what is possible by traditional image reconstruction algorithms. The sequential emission file generated by the detector makes it an ideal platform for training Recurrent Neural Networks (RNNs) which recast the problem of predicting ''next hits'' as an optimization problem. Long Short-Term Memory (LSTMs) networks to contextualize the entire data frame are also utilized which provide an additional constraint on the training policy for the RNNs making the regime even more robust. Image Segmentation (IS) is finally used to generate a pixel map that extracts shape information of target object which constitutes the final constraint layer for the RNN training routine.   

Not Participating

The primary motive of our research project is to use muon tomography in order to find hidden cavities within archeological excavation sites. Our target excavation site is Limyra, which is located in Southern Turkey. Muon tomography, the focus of our research, is a technique that is used to create three-dimensional maps of the dig sites using Coulomb scattering of the muon. This technique is used to find cavities inside volcanos, mountains, and archeological sites. We have been developing and testing prototypes of portable muon telescope for several years. The first prototype was built in 2019 with four trays containing scintillator bars and silicon photomultipliers. This prototype successfully produced images of a large water tank. A second prototype was built in 2020 which tested a new technique for readout signals. We are currently working on a third prototype. Our goal is to design and construct a muon telescope which can detect much smaller objects than the water tank and that will be operational at the Limyra hillside. Using GEANT4, we developed a Monte Carlo simulation program to optimize the design of these prototypes. In this presentation, we will report the performance of the third prototype muon telescope estimated from our Monte Carlo studies.   

Live

Very-High-Energy (VHE) neutrinos are expected to be produced by cosmic-ray interactions with the Cosmic Microwave Background (CMB). In these photo-hadronic interactions, $\nu_{\mu}$ and $\nu_e$ are produced. As these neutrinos traverse the cosmic void, they morph from one flavor to another, yielding, in the standard scenario, a democratic flavor composition at their arrival on Earth. This so-called cosmogenic flux of VHE neutrinos is a target of the next generation neutrino observatories: IceCube-Gen2, TAMBO, RNO, GRAND, POEMMA, and CHANT. In a recent publication, a novel detection strategy for these neutrinos has been put forward. This new technique relies on the observation of Earth-throughgoing $\nu_{\tau}$ at PeV energies. By measuring the flux at this energy, we can indirectly observe the flux at EeV energies since these two are related by the cascading down of the neutrinos. However, such a link demands an accurate simulation of the VHE $\nu_{\tau}$ transport. TauRunner is a Python Monte Carlo (MC) package specialized in EeV $\nu_{\tau}$ transport with the limitation of not accounting for secondary flavors produced in some $\tau$ decay channels. In this contribution, I will present the newest version of this MC, which now incorporates all neutrino flavors in the propagation   

Live

For the High-Luminosity LHC program, the ATLAS experiment will be upgraded with a new tracking detector, the Inner Tracker (ITk). The ITk is the innermost detector of the ATLAS experiment and is composed of concentric cylindrical layers of silicon detectors: five layers of pixel detectors surrounded by four layers of strip detectors. To validate the design of the ITk and quantify its expected performance, a detailed simulation of the new detector needs to be created. The simulation is built with a C++ framework that allows for a simple description of the detector elements and its properties using flexible XML markup language. The geometry of the detector will be virtually created using the "GeoModelXML" software package by building up each component in a hierarchical manner. This then allows us to simulate particles passing through sensors, providing a detailed estimation of the signals induced on the readout electronics as well as the particles' interaction with the passive detector material. A portion of the ITk is already described in this framework. Before the ITk detector simulation can be used in the mainstream ATLAS software release, the framework must be extended to include the full detector. The Brandeis team has built a portion of the ITk geometry through GeoModel.   

Live

Within Beyond the Standard Model (BSM) theories, interest has been growing in the proposed dark sector, containing particles that are not charged under Standard Model (SM) gauge groups. Dark photons may interact via the vector portal as a result of kinetic mixing, so the LHC could be a viable tool for dark photon production. We consider the $ZH$ production mechanism to search for a newly-predicted decay of the Higgs boson into a photon ($\gamma$) and a massless dark photon ($\gamma_D$) with a target final state of $Z (\rightarrow \ell^{+}\ell^{-}) H (\rightarrow \gamma \gamma_D)$. We use Monte-Carlo simulation samples corresponding to the total integrated luminosity of $pp$ collisions at $\sqrt{s}=$ 13 TeV collected by the ATLAS detector in 2015-2018. First, we implement rectangular cut optimization to obtain a baseline signal region (SR) selection. Next, we explore the use of boosted decision trees (BDTs) and neural networks to train a classifier for signal-background discrimination. We find that BDTs using the Gradient Boosting algorithm yield the best performance with $93 \%$ accuracy and an area under the Receiver Operating Characteristic (ROC) curve (AUC) of 0.96. We finally present preliminary studies of using the BDT output scores for signal-background discrimination.   

Live

High-energy neutrinos and gamma rays are two important messengers of extreme astrophysical environments, and individual flaring $\gamma $-ray blazars offer a promising opportunity for the identification of neutrino emitters. The neutrino event IC201114A was detected by IceCube on 2020-11-14 in the vicinity of the known $\gamma $-ray source 4FGL J0658.6$+$0636. To better understand the $\gamma $-ray variability of this source and to search for any temporal correlation between the IceCube event and $\gamma $-ray emissions, we analyzed $\gamma $-ray observations of 4FGL J0658.6$+$0636 over the timescales of 1 day, 1 month, 6 months, 1 year, 5 years and 10 years prior to the event with the Large Area Telescope (LAT) onboard NASA's Fermi Gamma-ray Space Telescope. We found neither evidence for strong $\gamma $-ray variability nor a significant detection in the time windows up to 1 year prior to the IceCube event. The lack of temporal correlation between gamma rays from 4FGL J0658.6$+$0636 and the neutrino event could suggest that the neutrino event is not from this source or that the gamma rays are absorbed in the emitting region. Regular monitoring of this source and timely followup observations of future IceCube events will continue to search for multi-messenger emission from potential cosmic accelerators.   

Live

The measurement of $\phi$ mesons provides a unique and complementary method for exploring properties of the quark-gluon plasma (QGP). The $\phi$ meson has a relatively small hadronic interaction cross section and is sensitive to the increase of strangeness in the QGP (strangeness enhancement), a phenomenon associated with soft particles in bulk matter. Measurements in the dilepton channels are especially useful since leptons interact only electromagnetically, thus carrying the information about their production directly to the detector. Measurements in different nucleus-nucleus collisions allow us to perform a systematic study of the nuclear medium effects on $\phi$ meson production and gain better insight to the formation of the QGP. With data taken by the PHENIX detector in 2014 and 2015, we measure the $\phi$ meson production in a wide range of transverse momenta and rapidity. In this talk, we present the status of $\phi$ meson invariant yield and nuclear modification ($R_{AA}$) measurements in a variety of small collision systems, including p+Al and p+Au at $\sqrt{s_{NN}}$ = 200 GeV.   

Live

The HAYSTAC experiment is a quantum-enhanced search for cold dark matter axions at Yale University. By coupling the HAYSTAC cavity to a squeezed state receiver, this experiment has surpassed the quantum limit. A large component of the accompanying data analysis is the search for a weak putative axion signal buried in noise. As such, computationally efficient signal processing and filtering techniques that can easily operate in the time and frequency domains are a crucial analysis tool. We present simulations of several different maximum likelihood estimation algorithms and least squares analyses applied to data corrupted by additive white Gaussian noise. These methods aim to compute the original signal parameters and reconstruct clean sinusoidal signals. We compare these algorithms against each other to determine which method offers the optimal experimental sensitivity and examine their associated limits and constraints.   

Live

Prior to obtaining elastic and inelastic scattering cross sections from magnetic spectrograph data from the Research Center for Nuclear Physics (RCNP), one must first apply particle identification, shape correction, background subtraction, and state selection to several runs of data. Currently, these processes are completed sequentially by a researcher and verified independently by another researcher. This process is simple yet time-consuming, and the human judgement inherent in these processes has the potential to add non-negligible errors to the final results. We propose to improve the generalized speed and accuracy of these operations in RCNP data using artificial intelligence techniques. To do this, we will use established image processing methods to isolate potential regions of interest and then select the proper region through an exploration of their physical properties. In doing so, we will create a generalized cross section data reduction routine with greater speed, accuracy, and ease of use than the previous approach.