An Adjoint Method for Differentiable Fluid Simulation on Flow Maps 1656 Conference: SIGGRAPH Asia 2025 Conference Papers; December 15–18, 2025; Hong Kong, Hong Kong SIGGRAPH Asia 2025 Conference Papers (SA Conference Papers ’25), December 15–18, 2025, Hong Kong, Hong Kong DOI: 10.1145/3757377.376390
Zhiqi Li, Jinjin He, Barnabás Börcsök, Taiyuan Zhang, Duowen Chen, Tao Du, Ming C. Lin, Greg Turk, Bo Zhu · 2025
Read An Adjoint Method for Differentiable Fluid Simulation on Flow Maps 1656 Conference: SIGGRAPH Asia 2025 Conference Papers; December 15–18, 2025; Hong Kong, Hong Kong SIGGRAPH Asia 2025 Conference Papers (SA Conference Papers ’25), December 15–18, 2025, Hong Kong, Hong Kong DOI: 10.1145/3757377.376390 by Zhiqi Li, Jinjin He, Barnabás Börcsök, Taiyuan Zhang, Duowen Chen, Tao Du, Ming C. Lin, Greg Turk, Bo Zhu (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2511-01259.txt.
Description
Zhiqi Li and colleagues make fluid simulation differentiable on flow maps, so that gradients can flow through a simulation for tasks like identifying, predicting, and controlling vortices. The key observation is that the forward simulation and its backward adjoint twin share the same flow map, which transports fluid quantities from the first frame to the current one.
Their adjoint solver runs directly on that map, propagating gradient information backward without differentiating every intermediate numerical step or storing intermediate states, the twin costs of conventional adjoint methods. A sparse representation mixing long and short maps keeps the cost down. The method uses modest memory even at high resolution while tracking vorticity accurately, opening differentiable tasks that need precise vortex control.
Catalog record
| CREATOR | Zhiqi Li, Jinjin He, Barnabás Börcsök, Taiyuan Zhang, Duowen Chen, Tao Du, Ming C. Lin, Greg Turk, Bo Zhu |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 61 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2511-01259