Wednesday 10:00 -11:00, https://ucr.zoom.us/j/97606227247
Organizers : Weitao Chen / Yat Tin Chow / Qixuan Wang / Jia Gou / Mykhailo Potomkin / Yiwei Wang / Maziar Raissi / Siting Liu / Thomas Bury
Past Organizers : Mark Alber / James Kelliher / Amir Moradifam / Heyrim Cho
Archives: Fall 2019 / Winter 2020 / Spring 2020 / Fall 2020 / Winter 2021 / Spring 2021 / Fall 2021 / Winter 2022 / Spring 2022 / Fall 2022 / Winter 2023 / Spring 2023 / Fall 2023 / Winter 2024 / Spring 2024 /Fall 2024 / Winter 2025 / Spring 2025 / Fall 2025 / Winter 2026/ Spring_2026
Links : Interdisciplinary Center for Quantitative Modeling in Biology (ICQMB) seminar
In Fall 2026, the PDE & Applied Math Seminar will be offered in hybrid format: speakers may present either via Zoom or in person in Skye 285 (Computer Lab), and a Zoom option will be available for all talks.
Specific information about the format of each talk will be provided in the email announcement and posted below. If you're interested in attending the seminar, please contact Dr. Yat Tin Chow (yattinc@ucr.edu) and Dr. Yiwei Wang (yiweiw@ucr.edu).
Fall 2026 Schedule
Sep 30, Yesom Park (UCLA)
Oct 07, Kunlun Qi (MSU)
Oct 14, Oliver P. Waite (UCR)
Oct 21, Sebastian Munoz (UCLA)
Oct 28, Sichen Yang (JHU)
Nov 03, (joint wil ICQMB seminar) Therese Basa Landry (UCSB)
Nov 04, Bohan Zhou (UCSB)
Nov 25, Kai Fung Kan (UCLA)
Dec 02, Di Qi (Purdue)
Upcoming Talks:
Oct 14th, 10:00am-10:50am, hybrid
Oliver P. Waite (UCR)
Title: Inverse Wave Problems on Graphs
Abstract: Recovering both the source and speed of a wave from a single measurement is a difficult inverse problem. In this joint work with Amir Moradifam, we consider a finite graph analogue to this continuous problem. We assume the graph geometry is known and the unknowns are an initial signal and positive vertex masses. I will first discuss uniqueness and stability for recovery of either unknown separately using zero forcing. I will then present three sufficient conditions for simultaneous recovery. The common link between these mechanisms is leveraging the known geometry to choose observation sets that allow redundant information to propagate through the graph. The resulting theory gives global uniqueness under generic spectral and signal hypotheses together with Lipschitz stability on compact classes. Time permitting, I will conclude with some discussion of the ideas these results have given us for simultaneous recovery in the continuous case.
Oct 21th, 10:00am-10:50am, Zoom
Dr. Sebastian Munoz (UCLA)
Title: The optimal rate of convergence in mean field control
Abstract: I will discuss the sharp rate of convergence of the value functions of N-particle stochastic control problems to their mean field limit, for merely Lipschitz mean field costs. For d ≥ 2, the optimal rate is that of the empirical measures of i.i.d. samples in the 1-Wasserstein distance, as conjectured by Daudin, Delarue, and Jackson, and it persists with additive common noise. In dimension one, this benchmark can surprisingly be beaten: cooperating particles outperform independent samples, and the optimal exponent is 4/7, strictly between the accuracy of independent sampling (N^{-1/2}) and that of quantization (N^{-1}). The proofs rely on a new control-theoretic technique of recoupled shadow flows and, in dimension one, on a Gibbs law implementing the cooperation and a Schrödinger ground-state estimate.
Past Talks:
Sep 30th, 10:00am-10:50am, zoom
Dr. Yesom Park, University of California, Los Angeles
Title: Controlling Generative Flows with Minimal Intervention
Abstract: This talk discusses how a control-theoretic perspective provides a principled framework for controlling modern generative models. Flow matching (FM) models have emerged as effective generative models, while many downstream applications require their samples to satisfy prescribed constraints. We formulate constrained generation as a control problem in which a pre-trained FM is modified only as much as necessary, while keeping its learned flow unchanged. This leads to a training-free framework that computes the minimal perturbation of the flow state efficiently through an adjoint equation and adaptively determines when to intervene. The resulting formulation provides theoretical control over both constraint satisfaction and deviation from the pre-trained distribution. We demonstrate its broad applicability across functional, latent, and text-to-image FM backbones, with experiments spanning physics-informed generation, image inverse problems, and text-guided image editing.
Oct 07th, 10:00am-10:50am, zoom
Dr. Kunlun Qi, Michigan State University
Title: Fast Fourier spectral method for kinetic equations: from particle to wave
Abstract: In this talk, I will present our recent progress on developing fast Fourier spectral methods for both classical kinetic equations and wave kinetic equations (WKE).
In the first part, we focus on the Boltzmann equation, a fundamental model bridging many-particle system with macroscopic fluid dynamics. We introduce an efficient spectral formulation that approximates the non-cutoff Boltzmann collision operator and accelerates its computation through fast Fourier transforms (FFT), achieving accuracy and efficiency comparable to the cutoff case. Despite the practical success of the approaches, stability remains a significant challenge due to the lack of positivity and conservation. To address this, we develop a new analysis framework establishing spectral accuracy via a refined $L^2$ estimate of the negative part of the numerical solution. In the second part, we extend the fast spectral methodology to the WKE, the core equation in wave turbulence theory. By reformulating the high-dimensional nonlinear wave kinetic operator into a spherical integral — mirroring the structure of the Boltzmann operator — we reveal a double-convolution structure in Fourier space. This structure enables an efficient FFT-based evaluation of the wave interaction integral. Throughout the talk, I will demonstrate the accuracy, efficiency, and robustness of the proposed methods through a variety of 2D and 3D numerical experiments, and I will highlight several interesting phenomena and conjectures that emerge from these simulations.
.