People
We build quickly, and we break quickly, and we move on.
We build quickly, and we break quickly, and we move on.
Ashesh Chattopadhyay
I am an assistant professor in the Department of Applied Mathematics at the University of California Santa Cruz. Before I moved back to academia, I was a staff scientist in Silicon Valley at the Palo Alto Research Center (formerly Xerox PARC and currently a part of SRI).
If you've come across any of my papers that are behind a paywall and you want access to it, drop me an email.
The key questions I'm interested in are:
Physical consistencies and interpretability in neural architectures-- How well do neural architectures satisfy physics (by construction) ?
Limit of predictability for neural architectures (on physical problems)
Synergistic integration of neural models with physical models (for problems in turbulence, climate dynamics, and generally chaos). See my talk on this for an overview.
Data assimilation through the eyes of learning algorithms. See my talk on this for an overview.
Theory of deep learning from a dynamical systems perspective (and also the other way around) See Talk 1 and Talk 2 for an overview.
In another life, I used to do:
Reduced-order modeling (data-driven and otherwise) of non-linear systems (usually chaotic and turbulent)
Computer architecture and algorithms (mostly for scientific computing) for exascale-capable hybrid-parallel (MPI+OpenMP/CUDA) physics simulators
Computational geometry (algorithms for optimization) for fluid interface modeling
Google Scholar:link Github: link ResearchGate:link LinkedIn:link Twitter:link
Email: aschatto@ucsc.edu ashesh6810@gmail.com
Outside academia, I am on the advisory board for Vayuh where we are building cool things in the climate tech space.
Pulkit Dubey (postdoc, 2025-)
Leonard Lupin-Jiminez (2023-)
Patrick Wyrod (co-advised with Daniele Venturi) (2024-)
Moein Darman (2024-)
Anish Sambamurthy (2024-)
Niloofar Asefi (2025-) (Department of Electrial and Computer Engineering)
Conrad Ainslie (2025-)
Paul Killam (2025-) (Department of Computer Science and Engineering) co-advised with Xavier Prochaska
Hayley Coyle (SciCAM, thesis: Data-driven modeling of geophysical flows with partial states)
Conrad Ainslie (SciCAM, thesis: A theoretical eigenanalysis framework for neural autoregressive models of multi-scale chaotic dynamics)
A.J. Silberstein (SciCAM, thesis: Data-Driven Observation Models from Invariant Measures for Ensemble-Based Data Assimilation)
Lumina Kinsinger-Dang (SciCAM, thesis: Autoregressive Fourier Neural Operator Networks as Foundation Model for Sub-grid Scale Modeling)