Lab Home | Phone | Search
	Center for Nonlinear Studies
	Home  ▶ People   CNLS Staff Members   Executive Committee   Postdocs   Visitors   Students   Research   Publications  ▶ Conferences   Workshops   Sponsorship  ▶ Talks   Seminars   Postdoc Seminars Archive   Quantum Lunch   Quantum Lunch Archive   P/T Colloquia   Archive   Ulam Scholar     Postdoc Nominations  ▶ Student Requests   Student Program  ▶ Visitor Requests   Description   Past Visitors  ▶ Services   General     History of CNLS     Maps, Directions   T-Division   LANL				Monday, June 16, 20251:00 PM - 2:00 PMCNLS Conference Room (TA-3, Bldg 1690) Seminar Koopman mode decomposition with learned distances David AristoffColorado State University We introduce Featurized Koopman Mode Decomposition (FKMD), an integrated method for choosing KMD features using a learned Mahalanobis distance on a delay embedded space. The method is inspired by the recent observation that the outerproduct of the weights of a fully trained neural network agree with the average gradient outerproduct (AGOP) of the underlying interpolant (Mechanism for feature learning in neural networks and backpropagation-free machine learning models, Radhakrishnan et al., 2024). The Mahalanobis distance aids in featurizing KMD in cases where good features are not a priori known. We show that FKMD improves predictions for a high-dimensional linear oscillator, a high-dimensional Lorenz attractor that is partially observed, and a cell signaling problem from cancer research. Bio: David Aristoff is an associate professor of mathematics at Colorado State University, having joined the faculty in 2014. Prior to that, he was a Dunham Jackson Assistant Professor at the University of Minnesota, and earned his Ph.D. in mathematics from UT Austin in 2011. His research is in applied math and scientific computing. Join the meeting in Teams Meeting ID: 258 219 958 587 Passcode: 24cq2oZ7 Host: Danny Perez, T-1