Accelerating Multiphase Simulations With Denoising Diffusion Model Driven Initializations

Abstract This study introduces a hybrid fluid simulation approach that integrates generative diffusion models with physics‐based simulations, aiming at reducing the computational costs of flow simulations while still honoring all the physical properties of interest. Pore‐scale simulations enhance ou...

Descrizione completa

Dettagli Bibliografici

Autori principali:	Jaehong Chung, Agnese Marcato, Eric J. Guiltinan, Tapan Mukerji, Hari Viswanathan, Yen Ting Lin, Javier E. Santos
Natura:	Artigo
Lingua:	Inglês
Pubblicazione:	Wiley 2024-12-01
Serie:	Journal of Geophysical Research: Machine Learning and Computation
Soggetti:	diffusion models lattice‐Boltzmann pore‐scale generative modeling physics
Accesso online:	https://doi.org/10.1029/2024JH000293
Tags:	Aggiungi Tag Nessun Tag, puoi essere il primo ad aggiungerne!!