JAX-based differentiable fluid dynamics on GPU and end-to-end optimization

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Wang, Wenkang, Zhang, Xuanwei, Bezgin, Deniz, Buhendwa, Aaron, Chu, Xu, Weigand, Bernhard
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866916304822206464
author Wang, Wenkang
Zhang, Xuanwei
Bezgin, Deniz
Buhendwa, Aaron
Chu, Xu
Weigand, Bernhard
author_facet Wang, Wenkang
Zhang, Xuanwei
Bezgin, Deniz
Buhendwa, Aaron
Chu, Xu
Weigand, Bernhard
contents This project aims to advance differentiable fluid dynamics for hypersonic coupled flow over porous media, demonstrating the potential of automatic differentiation (AD)-based optimization for end-to-end solutions. Leveraging AD efficiently handles high-dimensional optimization problems, offering a flexible alternative to traditional methods. We utilized JAX-Fluids, a newly developed solver based on the JAX framework, which combines autograd and TensorFlow's XLA. Compiled on a HAWK-AI node with NVIDIA A100 GPU, JAX-Fluids showed computational performance comparable to other high-order codes like FLEXI. Validation with a compressible turbulent channel flow DNS case showed excellent agreement, and a new boundary condition for modeling porous media was successfully tested on a laminar boundary layer case. Future steps in our research are anticipated.
format Preprint
id arxiv_https___arxiv_org_abs_2406_19494
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle JAX-based differentiable fluid dynamics on GPU and end-to-end optimization
Wang, Wenkang
Zhang, Xuanwei
Bezgin, Deniz
Buhendwa, Aaron
Chu, Xu
Weigand, Bernhard
Fluid Dynamics
This project aims to advance differentiable fluid dynamics for hypersonic coupled flow over porous media, demonstrating the potential of automatic differentiation (AD)-based optimization for end-to-end solutions. Leveraging AD efficiently handles high-dimensional optimization problems, offering a flexible alternative to traditional methods. We utilized JAX-Fluids, a newly developed solver based on the JAX framework, which combines autograd and TensorFlow's XLA. Compiled on a HAWK-AI node with NVIDIA A100 GPU, JAX-Fluids showed computational performance comparable to other high-order codes like FLEXI. Validation with a compressible turbulent channel flow DNS case showed excellent agreement, and a new boundary condition for modeling porous media was successfully tested on a laminar boundary layer case. Future steps in our research are anticipated.
title JAX-based differentiable fluid dynamics on GPU and end-to-end optimization
topic Fluid Dynamics
url https://arxiv.org/abs/2406.19494