Published July 22, 2026
| Version v1.0.0
Software
Open
Resolution-Agnostic Iterative Generative Data Assimilation
Authors/Creators
Description
A resolution-agnostic framework for sequential data assimilation that decomposes super-resolution and state estimation into a cascade of forecast-then-correct stages using a shared conditional diffusion model across resolution levels. Demonstrated on 1D stochastic Burgers equation and 2D Kraichnan turbulence testbeds.
Notes
Files
dmrigank/Iterative_refinement_DA-v1.0.0.zip
Additional details
Related works
- Is supplement to
- Software: https://github.com/dmrigank/Iterative_refinement_DA/tree/v1.0.0 (URL)
Software
- Repository URL
- https://github.com/dmrigank/Iterative_refinement_DA