Matthew Epland

Portfolio Projects:

Exploring Interdisciplinary Research at Duke via Ph.D. Committees
1st Place - 2018 Scholars@Duke Visualization Challenge

Poster
Paper
Interactive Network
Repository

2018 Scholars@Duke Visualization Challenge
Duke Scholars Bridge Disciplines to Tackle Big Questions

Description: By combining Duke Ph.D. committee membership data with the faculty appointments directory, connections between academic organizations were found and used to construct an undirected, weighted network (or graph). From this network communities of closely linked organizations were created via the Louvain method. Additionally, the level of interdisciplinary activity in each organization was measured by comparing the relative weights of their external and self connections. Analysis won 1st place in the competition.

Methods: Network analysis, Louvain method
Software: networkx, pandas, jupyter

ATLAS TRT Particle ID Machine Learning R&D Studies

Slides
Repository

Description: R&D studies of particle identification in the ATLAS Transition Radiation Tracker (TRT) where conducted utilizing machine learning techniques, with the goal of separating electron tracks from muons. Developed with fellow Duke graduate students Doug Davis and Sourav Sen, and continued by Davis and others within the TRT group. Support Vector Machines (SVM) and Boosted Decision Trees (BDT) from the scikit-learn library were tested, as well as Neural Networks (NN) in constructed in Keras+TensorFlow.

Methods: SVM, BDT, NN, k-fold Cross-Validation
Software: scikit-learn, keras, tensorflow

Numerical Methods and the Dampened, Driven Pendulum

Paper
Repository

Description: A computational study of the dampened, driven pendulum using the Euler-Cromer and Rung-Kutta numerical methods to investigate resonance, nonlinear behavior, and chaos. Numerical simulations compared to theory where possible.

Methods: Euler-Cromer, Rung-Kutta, Ordinary Differential Equations, Chaos and Lyapunov Exponents
Software: numpy, matplotlib

ATLAS Experiment

My research interest is experimental particle physics at the LHC, specifically the ATLAS experiment. Based at Duke University I analyzed data and wrote code, focusing on jet physics with my advisor Prof. Ayana Arce.

Multi-b Jet Search:

For my PhD dissertation I researched physics beyond the standard model with the multi-b search team. We looked for supersymmetric (SUSY) particles, gluinos, which decay to multiple b-quark jets plus missing energy. I applied machine learning in the form of Boosted Decison Trees (BDT) to improve the search's sensitive area by ∼30%.

• Defense Summary Slides - 2019
• Dissertation - 2019
• Search for supersymmetry in final states with missing transverse momentum and multiple b-jets in proton-proton collisions at sqrt(s) = 13 TeV with the ATLAS detector - 2018

Large-Radius Jet Calibration:

For my qualification task to earn ATLAS authorship I developed a data driven jet energy scale (in-situ JES) calibration for large-radius jets. I was responsible for the photon-jet direct balance portion of the combined calibration, which used momentum conservation to calibrate the large-radius jet by balancing it with a well-measured photon.

• In situ calibration of large-radius jet energy and mass in 13 TeV proton–proton collisions with the ATLAS detector - 2019
• Large-R Photon+Jet In Situ Calibration - BOOST 2018 Poster
• In-situ measurements of the ATLAS large-radius jet response in 13 TeV pp collisions - 2017
• In-situ jet energy scale calibration of large-radius jets - US LUA Lightning Round Talk - 2016

Links:

• ATLAS collaboration
• Prof. Ayana Arce

Teaching Fellow for Data Science Pipeline and Critical Thinking
2019
Part of the Harvard Business Analytics program
Course Overview: Ultimately, business analytics is about using data, analytics, and algorithms to make prescriptive predictions about future events and decisions. This course will take a holistic approach to helping participants understand the key factors involved, from data collection to analysis to prediction and insight. Projects will give students hands-on experience developing and running a data science pipeline to ensure that the correct business predictions are being made. Emphasis will be on merging technical skills with critical thinking to ensure that robust data science pipelines are being created for business benefit.

Teaching Assistant for DECISION 522Q: Data Visualization
Fall 2017
Course Description: This course explores techniques to effectively communicate information about data using graphical means, in popular data visualization tools such as Tableau.

Laboratory Teaching Assistant for PHYSICS 271L: Electronics
Spring 2016
Course Description: Elements of electronics including circuits, transfer functions, solid-state devices, transistor circuits, operational amplifier applications, digital circuits, and computer interfaces.

Laboratory Teaching Assistant for PHYSICS 152L: Introductory Electricity, Magnetism, and Optics
Fall 2015
Course Description: Intended principally for students in engineering and the physical sciences. Topics include: electric charge, electric fields, Gauss's Law, potential, capacitance, electrical current, resistance, circuit concepts, magnetic fields, magnetic and electric forces, Ampere's Law, magnetic induction, Faraday's Law, inductance, Maxwell's Equations, electromagnetic waves, elementary geometric optics, wave interference, and diffraction.