Same Physics, Different Gradients: A State-of-the-Art Report on Differentiable Physics Simulation
cs.GR, cs.CV, cs.LG, cs.NA, cs.RO, math.NA
Submitted: 2026-11-09
Updated: 2026-11-09
Code: https://github.com/jax-ml/jax
Terminology
Sources
- Lagrangian Neural Networks
- Dojo: A Differentiable Physics Engine for Robotics
- A Differentiable Programming System to Bridge Machine Learning and Scientific Computing
- Differentiable Rendering: A Survey
- PAT3D: Physics-Augmented Text-to-3D Scene Generation
- Highly-Efficient Differentiable Simulation for Robotics
- Gradients are Not All You Need
- Does "Do Differentiable Simulators Give Better Policy Gradients?" Give Better Policy Gradients?
- Differentiable Simulation of Hard Contacts with Soft Gradients for Learning and Control
- Differentiable Physics: A Position Piece
- Differentiate the Solver, Not the Equation: Reverse-Sweep Adjoints for Block Implicit Simulation
- Learning Deployable Locomotion Control via Differentiable Simulation
- Learning Quadruped Locomotion Using Differentiable Simulation
- An Exact Bitwise Reversible Integrator
- Physics-based Deep Learning
- Differentiable Physics Simulations with Contacts: Do They Have Correct Gradients w.r.t. Position, Velocity and Control?
Related papers
- AtlasLC: Fast Codec-Ready Compression of Object-Centric 3D Gaussian Splatting
- Velocity-Space 3D Asset Editing
- Physics-Driven Extraction of Dynamic Fluid Asset from Monocular Video with Generative 3DGS
- SCRIPT: Scalable Diffusion Policy with Multi-stage Training for Language-driven Physics-Based Humanoid Control
- Delaunay-Based Nearest Neighbor Search with Grid-Accelerated Queries and Parallel Construction
- Macrofacet Theory for Gaussian Process Statistical Surfaces