Corpus ID: 203626832

DiffTaichi: Differentiable Programming for Physical Simulation

@article{Hu2020DiffTaichiDP,
  title={DiffTaichi: Differentiable Programming for Physical Simulation},
  author={Y. Hu and L. Anderson and Tzu-Mao Li and Q. Sun and N. Carr and Jonathan Ragan-Kelley and F. Durand},
  journal={ArXiv},
  year={2020},
  volume={abs/1910.00935}
}
  • Y. Hu, L. Anderson, +4 authors F. Durand
  • Published 2020
  • Computer Science, Physics, Mathematics
  • ArXiv
  • We present DiffTaichi, a new differentiable programming language tailored for building high-performance differentiable physical simulators. Based on an imperative programming language, DiffTaichi generates gradients of simulation steps using source code transformations that preserve arithmetic intensity and parallelism. A light-weight tape is used to record the whole simulation program structure and replay the gradient kernels in a reversed order, for end-to-end backpropagation. We demonstrate… CONTINUE READING
    35 Citations
    JAX MD: A Framework for Differentiable Physics
    • 1
    • Highly Influenced
    • PDF
    Scalable Differentiable Physics for Learning and Control
    • 6
    • Highly Influenced
    • PDF
    Differentiable programming and its applications to dynamical systems
    • 1
    • PDF
    Solver-in-the-Loop: Learning from Differentiable Physics to Interact with Iterative PDE-Solvers
    • 3
    • PDF
    Sparse-Input Neural Network Augmentations for Differentiable Simulators
    Differentiable Molecular Simulations for Control and Learning
    • 5
    • PDF
    Instead of Rewriting Foreign Code for Machine Learning, Automatically Synthesize Fast Gradients
    • PDF

    References

    SHOWING 1-10 OF 47 REFERENCES
    Differentiable programming for image processing and deep learning in halide
    • 33
    • PDF
    End-to-End Differentiable Physics for Learning and Control
    • 109
    • PDF
    Reverse-mode AD in a functional framework: Lambda the ultimate backpropagator
    • 68
    • PDF
    Taichi: a language for high-performance computation on spatially sparse data structures
    • 15
    • PDF
    SPNets: Differentiable Fluid Dynamics for Deep Neural Networks
    • 47
    • PDF
    ChainQueen: A Real-Time Differentiable Physical Simulator for Soft Robotics
    • 44
    • PDF
    Theano: A CPU and GPU Math Compiler in Python
    • 777
    • Highly Influential
    • PDF
    Automatic differentiation in PyTorch
    • 7,360
    • Highly Influential
    Mitsuba 2
    • 22
    • PDF