Friday, February 27, 2026
Location: Fariborz Maseeh Hall (FMH), room 462
Time: 3:15pm - 4:15pm
Speaker: Ashesh Chattopadhyay, PhD
Title: Earth on a chip: AI-based autoregressive models of our Earth system
Abstract: Recent advances in machine learning have revolutionized dynamical modeling, yet AI weather and climate models often suffer from instability and unphysical drift when integrated over long timescales. This talk unifies three complementary works addressing this challenge. First, we present a theoretical eigenanalysis of neural autoregressive models that establishes a semi-empirical framework linking inference-time stability to the spectrum of the model’s Jacobian. This analysis reveals how integration-constrained architectures suppress unstable eigenmodes and enable predictable error growth. Building on this foundation, we identify spectral bias—a universal tendency of deep networks to under-represent high-wavenumber dynamics—as the root cause of instability in AI weather models. We demonstrate how higher-order integration schemes and spectral regularization, implemented in the FouRKS framework, mitigate this bias and produce century-scale stable emulations of turbulent flows. Finally, we translate these theoretical insights into practice with LUCIE-3D, a data-driven climate emulator trained on reanalysis data that captures forced responses to CO₂, reproduces stratospheric cooling and surface warming, and remains computationally efficient. Together, these results chart a rigorous pathway from mathematical theory to physically consistent AI climate models capable of stable, interpretable, and trustworthy long-term Earth-system emulation.
Biography: Ashesh is an Alfred. P. Sloan fellow and assistant professor in the department of applied mathematics at the University of California Santa Cruz. His interests lie at the intersection of theoretical deep learning, dynamical systems, and computational physics. Ashesh did his PhD from Rice University, Houston and spent a year at Xerox PARC and then at SRI as a staff research scientist before moving to UCSC.
The faculty host of this speaker is Professor Safa Mote
Friday, January 30, 2026
Location: Fariborz Maseeh Hall (FMH), room 462
Time: 3:15pm - 4:15pm
Speaker: Nina Unsheim
Title: Errors of Students’ Reasoning in the Context of Evaluating Media Claims
Abstract: The evaluation of media claims using mathematics and data is important, because of an increasing number and influence of mis- and disinformation today. However, results of public surveys lead to the assumption that the population has problems in evaluating media claims and accompanying reasoning processes. For this reason, we identify method-related and content-related errors, which occur in the reasoning processes in the context of evaluating media claims. Ten different errors are identified in a qualitative content analysis of answers from students of a German selective upper secondary school from year 9 and the corresponding year 10 working on evaluating media claims relating to sustainability topics. The existence of errors shows the necessity and importance of implementing the topics in school.
Biography: Nina Unshelm is a research assistant and PhD Student at the University of Würzburg, working at the Chair of Mathematics Education. She studied secondary school teaching with a focus on mathematics and chemistry. Her research interests include mathematics in the media, particularly the evaluation of media claims using mathematics and data and related reasoning processes, critical thinking, data literacy, as well as mathematics education with real-world connections and interdisciplinary learning.
The faculty host of this speaker is Dr. Eva Thanheiser
Friday, November 21, 2025
Location: Fariborz Maseeh Hall (FMH), room 462
Time: 3:15pm - 4:15pm
Speaker: Daniel Clark Orey, Ph.D.
Title: From an Imprecise Precision: Reflections on the Connections Between Culture and Modelling
Abstract: In this talk, I will share my thoughts in relation to culture and modelling. This awareness gradually emerged over time, beginning while growing up on the West Coast of North America, and later working in Guatemala, then New Mexico, California, and Nepal, to now where I currently live in Brasil. In the mid 80´s, as a young master’s student in New Mexico, I participated in a National Science Foundation (NSF) grant, which allowed a group of researchers to travel throughout the Southwestern United States. As part of my master’s research, I used this model in Guatemala, and then with doctoral research in México. By using LOGO, we looked at first contacts with computers, this was when I began to see interactions between humans, culture, and mathematics. These elements became the foundations of the dialogical processes of ethnomodelling with researchers in Brasil, Ecuador, Colombia, the United States, Ghana, México, Costa Rica, Nepal, and Indonesia.
Biography: Daniel Clark Orey, Ph.D. is Professor Emeritus of Mathematics and Multicultural Education at California State University, Sacramento. He has taught and lived in Oregon, Brazil, Guatemala, Mexico, Nepal and the United States. He is a Fulbright Senior specialist with experiences at the Pontifical Catholic University of Campinas in Brazil (1998) and at Kathmandu University in Nepal (2007). He is currently professor of mathematics education in the Department of Mathematics Education and serves in the Post-Doctoral Academic Masters and Doctoral Program in Mathematics Education at the Universidade Federal de Ouro Preto, Brazil.
The faculty host of this speaker is Dr. Eva Thanheiser