DAWIS: Data Assimilation with Windowed Inverse Sampling via Multitask Interpolants

stat.ML, cs.LG, physics.ao-ph

Submitted: 2026-10-02

Updated: 2026-10-02

Code: https://github.com/Erik-Wikingsson/DAWIS

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Sources

• Stochastic Interpolants: A Unifying Framework for Flows and Diffusions
• Accuracy of the Ensemble Kalman Filter in the Near-Linear Setting
• A Survey on Diffusion Models for Inverse Problems
• Exact Global MCMC with Denoising Diffusion
• Sampling Data with Chains of Forward-Backward Diffusion Steps
• Ensemble score filter with image inpainting for data assimilation in tracking surface quasi-geostrophic dynamics with partial observations
• Decoupled Weight Decay Regularization

Related papers

• Behavior of prediction performance metrics with rare events
• Optimal Estimation of Generic Dynamics by Path-Dependent Neural Jump ODEs
• A Posterior-Dynamics Framework for Imaging Inverse Problems with Pretrained Diffusion Priors
• Efficient quadratic entropy with distance sketches
• Embedding-Bias in Conditional Independence Testing
• From Log-Odds to Shapley Values: An Explanatory Geometry for the Weighted Naive Bayes Classifier