Physically-Motivated Machine Learning Models for Lagrangian Fluid Mechanics

Practical lessons on building machine learning models for Lagrangian fluid mechanics that respect the physics.

An overview of what we have learned building machine learning models for Lagrangian fluid mechanics — where physical structure helps, where it merely constrains, and what CFD and ML practice each have to offer the other.

(Winchenbach & Thuerey, 2024)

Presented orally at SPHERIC 2024 in Berlin, Germany. The full paper is available in this repository.

References

2024

  1. Physically-Motivated Machine Learning Models for Lagrangian Fluid Mechanics
    Rene Winchenbach, and Nils Thuerey
    In 18th International SPHERIC Workshop, Berlin, Germany, 2024
    Code