2019

DiffTaichi: Differentiable Programming for Physical Simulation

Hu, Yuanming, Anderson, Luke, Li, Tzu-Mao et al.

Understand

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 the performance and productivity of our language in gradient-based learning and optimization tasks on 10 different physical simulators.

Built on

A simple automatic derivative evaluation program
R. E. Wengert · 1964
Earlier work this paper cites.
Algebraic reconstruction techniques (art) for three-dimensional electron microscopy and x-ray photography
Richard Gordon, Robert Bender, and Gabor T Herman · 1970
Earlier work this paper cites.
What is an adjoint model?
Ronald M Errico · 1997
Earlier work this paper cites.
Stable fluids
Jos Stam · 1999
Earlier work this paper cites.
Interactive manipulation of rigid body simulations
Jovan Popović, Steven M Seitz, Michael Erdmann, Zoran Popović, and Andrew Witkin · 2000
Earlier work this paper cites.
Fast continuous collision detection between rigid bodies
Stéphane Redon, Abderrahmane Kheddar, and Sabine Coquillart · 2002
Earlier work this paper cites.
Evaluating derivatives: principles and techniques of algorithmic differentiation , volume 105
Andreas Griewank and Andrea Walther · 2008
Earlier work this paper cites.
Partial differential equations with numerical methods , volume 45
Stig Larsson and Vidar Thomée · 2008
Earlier work this paper cites.
Reverse-mode AD in a functional framework: Lambda the ultimate backpropagator
Barak A. Pearlmutter and Jeffrey Mark Siskind · 2008
Earlier work this paper cites.
Openad/f: A modular open-source tool for automatic differentiation of fortran codes
Jean Utke, Uwe Naumann, Mike Fagan, Nathan Tallent, Michelle Strout, Patrick Heimbach, Chris Hill, and Carl Wunsch · 2008
Earlier work this paper cites.
Modeling and solving constraints
Erin Catto · 2009
Earlier work this paper cites.
Theano: A cpu and gpu math compiler in python
James Bergstra, Olivier Breuleux, Frédéric Bastien, Pascal Lamblin, Razvan Pascanu, Guillaume Desjardins, Joseph Turian, David Warde-Farley, and Yoshua Bengio · 2010
Earlier work this paper cites.
Whole-function vectorization
Ralf Karrenberg and Sebastian Hack · 2011
Earlier work this paper cites.

Similar

Computing the singular value decomposition of 3x3 matrices with minimal branching and elementary floating point operations
Aleka McAdams, Andrew Selle, Rasmus Tamstorf, Joseph Teran, and Eftychios Sifakis · 2011
Cited alongside, same era.
ispc: A spmd compiler for high-performance cpu programming
Matt Pharr and William R Mark · 2012
Cited alongside, same era.
The Tapenade automatic differentiation tool: Principles, model, and specification
Laurent Hascoet and Valérie Pascual · 2013
Cited alongside, same era.
Halide: A language and compiler for optimizing parallelism, locality, and recomputation in image processing pipelines
Jonathan Ragan-Kelley, Connelly Barnes, Andrew Adams, Sylvain Paris, Frédo Durand, and Saman Amarasinghe · 2013
Cited alongside, same era.
Opendr: An approximate differentiable renderer
Matthew M Loper and Michael J Black · 2014
Cited alongside, same era.
The material point method for the physics-based simulation of solids and fluids
Chenfanfu Jiang · 2015
Cited alongside, same era.
Autograd: Effortless gradients in numpy
Dougal Maclaurin, David Duvenaud, and Ryan P Adams · 2015
Cited alongside, same era.
Tensorflow: A system for large-scale machine learning
Martín Abadi, Paul Barham, Jianmin Chen, Zhifeng Chen, Andy Davis, Jeffrey Dean, Matthieu Devin, Sanjay Ghemawat, Geoffrey Irving, Michael Isard, et al · 2016
Cited alongside, same era.
Interaction networks for learning about objects, relations and physics
Peter W. Battaglia, Razvan Pascanu, Matthew Lai, Danilo Rezende, and Koray Kavukcuoglu · 2016
Cited alongside, same era.
A compositional object-based approach to learning physical dynamics
Michael B Chang, Tomer Ullman, Antonio Torralba, and Joshua B Tenenbaum · 2016
Cited alongside, same era.
A differentiable physics engine for deep learning in robotics
Original
Jonas Degrave, Michiel Hermans, Joni Dambre, et al · 2016
Cited alongside, same era.
Automatic differentiation in pytorch
Adam Paszke, Sam Gross, Soumith Chintala, Gregory Chanan, Edward Yang, Zachary DeVito, Zeming Lin, Alban Desmaison, Luca Antiga, and Adam Lerer · 2017
Cited alongside, same era.

Then

JAX: composable transformations of Python+NumPy programs, 2018
James Bradbury, Roy Frostig, Peter Hawkins, Matthew James Johnson, Chris Leary, Dougal Maclaurin, and Skye Wanderman-Milne · 2018
Later among the works it cites.
A moving least squares material point method with displacement discontinuity and two-way rigid body coupling
Yuanming Hu, Yu Fang, Ziheng Ge, Ziyin Qu, Yixin Zhu, Andre Pradhana, and Chenfanfu Jiang · 2018
Later among the works it cites.
Neural 3d mesh renderer
Hiroharu Kato, Yoshitaka Ushiku, and Tatsuya Harada · 2018
Later among the works it cites.
Differentiable monte carlo ray tracing through edge sampling
Tzu-Mao Li, Miika Aittala, Frédo Durand, and Jaakko Lehtinen · 2018
Later among the works it cites.
Flexible neural representation for physics prediction
Damian Mrowca, Chengxu Zhuang, Elias Wang, Nick Haber, Li Fei-Fei, Joshua B Tenenbaum, and Daniel LK Yamins · 2018
Later among the works it cites.
Spnets: Differentiable fluid dynamics for deep neural networks
Original
Connor Schenck and Dieter Fox · 2018
Later among the works it cites.
Toward wave-based sound synthesis for computer animation
Jui-Hsien Wang, Ante Qu, Timothy R Langlois, and Doug L James · 2018
Later among the works it cites.
Taichi: A language for high-performance computation on spatially sparse data structures
Yuanming Hu, Tzu-Mao Li, Luke Anderson, Jonathan Ragan-Kelley, and Frédo Durand · 2019
Closest in time.
Zygote: A differentiable programming system to bridge machine learning and scientific computing
Original
Mike Innes, Alan Edelman, Keno Fischer, Chris Rackauckus, Elliot Saba, Viral B Shah, and Will Tebbutt · 2019
Closest in time.
Enoki: structured vectorization and differentiation on modern processor architectures, 2019
Wenzel Jakob · 2019
Closest in time.
Differentiable cloth simulation for inverse problems
Junbang Liang, Ming C Lin, and Vladlen Koltun · 2019
Closest in time.
Mitsuba 2: A retargetable forward and inverse renderer
Merlin Nimier-David, Delio Vicini, Tizian Zeltner, and Wenzel Jakob · 2019
Closest in time.
Differentiable programming mega-proposal
Richard Wei, Marc Rasi Dan Zheng, and Bart Chrzaszcz · 2019
Closest in time.

Beyond the bibliography

alphaXiv searches the wider corpus for related work and actual follow-ups.

Open on alphaXiv

alphaXiv is searching for related work…

Reading the bibliography…