Neural SPDE solver for uncertainty quantification in high-dimensional space-time dynamics - IMT - Institut Mines-Télécom Accéder directement au contenu
Pré-Publication, Document De Travail Année : 2023

Neural SPDE solver for uncertainty quantification in high-dimensional space-time dynamics

Résumé

Historically, the interpolation of large geophysical datasets has been tackled using methods like Optimal Interpolation (OI) or model-based data assimilation schemes. However, the recent connection between Stochastic Partial Differential Equations (SPDE) and Gaussian Markov Random Fields (GMRF) introduced a novel approach to handle large datasets making use of sparse precision matrices in OI. Recent advancements in deep learning also addressed this issue by incorporating data assimilation into neural architectures: it treats the reconstruction task as a joint learning problem involving both prior model and solver as neural networks. Though, it requires further developments to quantify the associated uncertainties. In our work, we leverage SPDEbased Gaussian Processes to estimate complex prior models capable of handling nonstationary covariances in space and time. We develop a specific architecture able to learn both state and SPDE parameters as a neural SPDE solver, while providing the precisionbased analytical form of the SPDE sampling. The latter is used as a surrogate model along the data assimilation window. Because the prior is stochastic, we can easily draw samples from it and condition the members by our neural solver, allowing flexible estimation of the posterior distribution based on large ensemble. We demonstrate this framework on realistic Sea Surface Height datasets. Our solution improves the OI baseline, aligns with neural prior while enabling uncertainty quantification and online parameter estimation.
Fichier principal
Vignette du fichier
Beauchamp_AISTATS2024.pdf (1.67 Mo) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-04265487 , version 1 (02-11-2023)

Licence

Paternité

Identifiants

Citer

Maxime Beauchamp, Ronan Fablet, Hugo Georgenthum. Neural SPDE solver for uncertainty quantification in high-dimensional space-time dynamics. 2023. ⟨hal-04265487⟩
23 Consultations
20 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More