Weekly Seminars
The AMSC program is excited to introduce a weekly seminar series starting this Spring. The seminars aim to foster collaboration between faculty and students, showcase research, and encourage student recruitment. Held every Monday at 4:00 PM in MATH 3206 (or virtually via Zoom), the series offers both synchronous and asynchronous presentation options. Join us for the first seminar on February 24 and stay tuned for ongoing weekly sessions. Recordings of the seminars can be accessed by the UMD community here.
This Week
Date: Monday, April 7, 2025
Time: 4:15 PM
Place: MATH 3206 (Colloquium Room)
Speaker: Bill Fagan (Biology)
Title: Finite Expression Method: A Symbolic Approach for Scientific Machine Learning
Abstract: This seminar will provide an overview of how continuous stochastic processes have been applied to the study of animal movement ecology using data from GPS tracking devices. I will present the mathematical foundations of these applications and discuss how we statistically fit the stochastic process models to diverse biological datasets. I will then give an overview of the wide range of applications that my colleagues and I have found for these approaches, including such biological topics as:
1) animal home ranges, migration, and space use
2) behavioral evidence for learning and disease states
3) route-based movement by carnivores
4) consumer-resource interactions
Movement data from GPS tracking devices typically feature a high degree of temporal autocorrelation, often at multiple scales. Over the years, our work has dealt with such data in a variety of statistical contexts, including:
1) timeseries analysis
2) kernel density estimation
3) path estimation via kriging
4) estimation of probability ridges
5) comparative (i.e., phylogenetically controlled) analyses
The talk will present results from joint work with mathematicians Leonid Koralov and Mark Lewis; past-postdocs Christen Fleming, Eliezer Gurarie, and Michael Noonan; past-PhD students Justin Calabrese and Nicole Barbour; current PhD students Frank McBride, Marron McConnell, Gayatri Anand, Stephanie Chia, Qianru Liao, and Phillip Koshute; current undergraduate Zachary Tomares; and hundreds of biologists. Open questions abound and span a wide range of difficulty. I have access to mountains of animal movement data and am eager for collaborators.
Spring Semester Schedule
- February 24: Maria Cameron (Mathematics)
- March 3: Steven Gabriel (Mechanical Engineering)
- March 10: Elana Fertig (School of Medicine)
- March 17: Spring break - No Seminar
- March 24: AMSC Open House - No Seminar
- March 31: Haizhao Yang (Mathematics)
- April 7: Bill Fagan (Biology)
- April 14: Antony Jose (Cell Biology & Molecular Genetics)
- April 21: Harry Dankowicz (Mechanical Engineering)
- April 28: Ricardo Nochetto (Mathematics)
- May 5: Alexander Estes (School of Business)
Past Seminar Details
Maria Cameron (Math)
Date: Monday, February 24, 2025
Time: 4:15 PM
Place: MATH 3206 (Colloquium Room)
Speaker: Maria Cameron (MATH)
Title: Complex Dynamics of Nonlinear Oscillators and Their Applications
Abstract: Nonlinear oscillators have a broad range of applications in engineering, including rotors, energy harvesters, sensors, and precision timing devices. The dynamics of a single oscillator with cubic nonlinearity and external periodic forcing is surprisingly rich. Depending on parameters, it may admit multiple attractors that may be periodic or chaotic. Their basin boundaries may be fractal. Linking oscillators into arrays and adding noise further complicates their dynamics. I will discuss a method for finding the most probable escape paths from the basins of attractors of noisy oscillators, sensor design, and a few open mathematical problems related to nonlinear oscillators.
Steven Gabriel (Mechanical Engineering)
Date: Monday, March 3, 2025
Time: 4:15 PM
Place: MATH 3206 (Colloquium Room)
Speaker: Steven Gabriel (Mechanical Engineering)
Title: Optimization/Equilibrium Modeling & Algorithm Development for Infrastructure Planning: Focus on Energy, Water and Transport Summary of Research and Teaching
Abstract: Professor Gabriel’s research group develops models, theory, and algorithms for solving problems that arise in infrastructure planning such as: energy, water, transport. These models are typified by a set of autonomous agents (i.e., energy market participants, vehicles) that share a common network. The equilibrium aspects arise since each of the players or subsets of the players compete non-cooperatively with each other for the infrastructure network’s resources. The concatenation of all these optimization problems as well as any system-level constraints results in what is known as an equilibrium problem; typically called a mixed complementarity problem (MCP) or a variation inequality (VI). Such problems generalize the Karush-Kuhn-Tucker (KKT) conditions of nonlinear programs, Nash-Cournot games, as well as many other problems in operations research, engineering and economic systems. These equilibrium problems can also be single-level, wherein all the agents are at the same level or such problems can be multi-level. In the latter case, some famous paradigms include: bilevel optimization (e.g., Stackelberg leader-follower games), attacker-defender interdiction problems and trilevel optimization. Please see Professor Gabriel’s website for further details: http://www.stevenagabriel.umd.edu/ or email him directly at with any questions you might have.
Elana Fertig (School of Medicine)
Date: Monday, March 10, 2025
Time: 4:15 PM
Place: MATH 3206 (Colloquium Room)
Speaker: Elana Fertig (School of Medicine)
Title: Forecasting carcinogenesis
Abstract: This talk presents a hybrid mathematical modeling and bioinformatics strategy to uncover interactions between neoplastic cells and the microenvironment during carcinogenesis and therapeutic response. As pancreatic cancer develops, it forms a complex microenvironment of multiple interacting cells. The microenvironment of advanced cancer includes a dense composition of cells, such as macrophages and fibroblasts, that are associated with immunosuppression. New single-cell and spatial molecular profiling technologies enable unprecedented characterization of the cellular and molecular composition of the microenvironment. These technologies provide the potential to identify candidate therapeutics to intercept immunosuppression. Inventing new mathematical approaches in computational biology are essential to uncover mechanistic insights from high-throughput data for these precision interception strategies. Here, we demonstrate how converging technology development, machine learning, and mathematical modeling can relate the tumor microenvironment to carcinogenesis and therapeutic response. Combining genomics with mathematical modeling provides a forecast system that can yield computational predictions to anticipate when and how the cancer is progressing for therapeutic selection. This mathematical forecast system will empower a new predictive oncology paradigm, which selects therapeutics to intercept the pathways that would otherwise cause future cancer progression.
Haizhao Yang (Mathematics)
Date: Monday, March 31, 2025
Time: 4:15 PM
Place: MATH 3206 (Colloquium Room)
Speaker: Haizhao Yang (Mathematics)
Title: Finite Expression Method: A Symbolic Approach for Scientific Machine Learning
Abstract: Machine learning has revolutionized computational science and engineering with impressive breakthroughs, e.g., making the efficient solution of high-dimensional computational tasks feasible and advancing domain knowledge via scientific data mining. This leads to an emerging field called scientific machine learning. In this talk, we introduce a new method for a symbolic approach to solving scientific machine learning problems. This method seeks interpretable learning outcomes via combinatorial optimization in the space of functions with finitely many analytic expressions and, hence, this methodology is named the finite expression method (FEX). It is proved in approximation theory that FEX can efficiently learn high-dimensional complex functions. As a proof of concept, a deep reinforcement learning method is proposed to implement FEX for learning the solution of high-dimensional PDEs and learning the governing equations of raw data.