works on

From the 1 of 28 linked papers with an AI index.

activity
20242026
collaborators

28 papers

cs.LG2026

Multimodal Spatiotemporal Atmospheric Data Assimilation with Latent Video Flow-matching

Dibyajyoti Chakraborty, Romit Maulik

Data assimilation (DA) uses Bayesian inference to update the state of a numerical forecast model with observed data. In this study, we propose a fundamentally different, unified ap…

cs.LG2026

A Weak Penalty Neural ODE for Learning Chaotic Dynamics from Noisy Time Series

Xuyang Li, John Harlim, Dibyajyoti Chakraborty +1

The paper introduces a weak-form loss function for training Neural ODEs that improves learning of chaotic dynamics from noisy time‑series data, yielding more stable and accurate sh…

cs.LG2026

Scale-Aware Learning of Chaotic Dynamics on Unstructured Meshes via Binned Spectral Losses

Kanad Sen, Romit Maulik

Surrogate modeling for high-dimensional nonlinear dynamical systems that exhibit chaos requires mechanisms that preserve not only pointwise accuracy but also the scale-dependent st…

cs.LG2026

High-Resolution Climate Projections Using Diffusion-Based Downscaling of a Lightweight Climate Emulator

Haiwen Guan, Dibyajyoti Chakraborty, Moein Darman +3

The proliferation of data-driven models in weather and climate sciences has marked a significant paradigm shift, with advanced models demonstrating exceptional skill in medium-rang…

physics.flu-dyn2026

Multiscale Hypersonic Boundary Layer Reconstruction via Spectral Binning and Subdomain-wise Conditional Diffusion

Hojin Kim, Dibyajyoti Chakraborty, Takahiko Toki +2

We propose a multiscale probabilistic reconstruction framework for hypersonic Couette flow, where near-wall states are inferred from limited top-wall observations using conditional…

physics.flu-dyn2026

Deep Learning of Solver-Aware Turbulence Closures from Nudged LES Dynamics

Ashwin Suriyanarayanan, Dibyajyoti Chakraborty, Romit Maulik

The differentiable physics paradigm may be leveraged as an a-posteriori approach for discovering turbulence closure models by embedding a neural network parameterization directly i…