Reverse-mode automatic differentiation and optimization of GPU kernels via Enzyme
Computing derivatives is key to many algorithms in scientific computing and machine
learning such as optimization, uncertainty quantification, and stability analysis. Enzyme is a …
learning such as optimization, uncertainty quantification, and stability analysis. Enzyme is a …
Scalable automatic differentiation of multiple parallel paradigms through compiler augmentation
Derivatives are key to numerous science, engineering, and machine learning applications.
While existing tools generate derivatives of programs in a single language, modern parallel …
While existing tools generate derivatives of programs in a single language, modern parallel …
Source-to-source automatic differentiation of OpenMP parallel loops
J Hückelheim, L Hascoët - ACM Transactions on Mathematical Software …, 2022 - dl.acm.org
This article presents our work toward correct and efficient automatic differentiation of
OpenMP parallel worksharing loops in forward and reverse mode. Automatic differentiation …
OpenMP parallel worksharing loops in forward and reverse mode. Automatic differentiation …
Automatic differentiation for adjoint stencil loops
Stencil loops are a common motif in computations including convolutional neural networks,
structured-mesh solvers for partial differential equations, and image processing. Stencil …
structured-mesh solvers for partial differential equations, and image processing. Stencil …
Hybrid Parallel Discrete Adjoints in SU2
The open-source multiphysics suite SU2 features discrete adjoints by means of operator
overloading automatic differentiation (AD). While both primal and discrete adjoint solvers …
overloading automatic differentiation (AD). While both primal and discrete adjoint solvers …
PARAD: A Work-Efficient Parallel Algorithm for Reverse-Mode Automatic Differentiation∗
Automatic differentiation (AD) is a technique for computing the derivative of function F: R n→
R m defined by a computer program. Modern applications of AD, such as machine learning …
R m defined by a computer program. Modern applications of AD, such as machine learning …
Parallelizable adjoint stencil computations using transposed forward-mode algorithmic differentiation
Algorithmic differentiation (AD) is a tool for generating discrete adjoint solvers, which
efficiently compute gradients of functions with many inputs, for example for use in gradient …
efficiently compute gradients of functions with many inputs, for example for use in gradient …
Event-based automatic differentiation of OpenMP with OpDiLib
J Blühdorn, M Sagebaum, N Gauger - ACM Transactions on …, 2023 - dl.acm.org
We present the new software OpDiLib, a universal add-on for classical operator overloading
AD tools that enables the automatic differentiation (AD) of OpenMP parallelized code. With it …
AD tools that enables the automatic differentiation (AD) of OpenMP parallelized code. With it …
Sense & sensitivities: The path to general-purpose algorithmic differentiation
M Innes - Proceedings of Machine Learning and Systems, 2020 - proceedings.mlsys.org
We present Zygote, an algorithmic differentiation (AD) system for the Julia language. Zygote
is designed to address the needs of both the machine learning and scientific computing …
is designed to address the needs of both the machine learning and scientific computing …
Automatic differentiation of parallel loops with formal methods
J Hückelheim, L Hascoët - … of the 51st International Conference on …, 2022 - dl.acm.org
This paper presents a novel combination of reverse mode automatic differentiation and
formal methods, to enable efficient differentiation of (or backpropagation through) shared …
formal methods, to enable efficient differentiation of (or backpropagation through) shared …