ForwardDiff.jl
autodiff
Our great sponsors
ForwardDiff.jl | autodiff | |
---|---|---|
4 | 7 | |
854 | 1,532 | |
1.4% | 2.5% | |
5.7 | 7.5 | |
22 days ago | 22 days ago | |
Julia | C++ | |
GNU General Public License v3.0 or later | MIT License |
Stars - the number of stars that a project has on GitHub. Growth - month over month growth in stars.
Activity is a relative number indicating how actively a project is being developed. Recent commits have higher weight than older ones.
For example, an activity of 9.0 indicates that a project is amongst the top 10% of the most actively developed projects that we are tracking.
ForwardDiff.jl
-
The Elements of Differentiable Programming
You seem somewhat obsessed with the idea that reverse-mode autodiff is not the same technique as forward-mode autodiff. It makes you,,, angry? Seems like such a trivial thing to act a complete fool over.
What's up with that?
Anyway, here's a forward differentiation package with a file that might interest you
https://github.com/JuliaDiff/ForwardDiff.jl/blob/master/src/...
-
Excited for Julia v1.9
Just so you know, v1.9 doesn't solve the load problems. What it does it gives package authors the tools to solve the problems, specifically precompilation as binaries and package extensions. It won't actually solve the load problems until the packages are updated to effectively make use of these features. This is already underway, https://sciml.ai/news/2022/09/21/compile_time/ with things like and https://github.com/JuliaDiff/ForwardDiff.jl/pull/625, but it is a fairly heavy lift to ensure things aren't invalidating and that everything that's necessary is precompiling.
-
Looking for numerical/iterative approach for determining a value
As a quick way to do it, you can use ForwardDiff.jl to determine the partial with respect to h. Then use a Newton-Raphson algorithm to solve for the value of h. I'm not familiar with the actual problem you're solving so there may be more appropriate ways to solve this based on the shape of your function, but this is my knee-jerk reaction to a problem like this. You could also calculate the partial derivative analytically if that is something that you want.
-
Question About Numerical Derivatives/Gradients: Why has no one yet implemented a gradient function in Julia that is similar to the gradient function in MATLAB and NumPy?
In these discussions, which are the only ones I could find that are the most pertinent and similar to what I'm talking about, https://github.com/JuliaDiff/ForwardDiff.jl/issues/390 and https://discourse.julialang.org/t/differentiation-without-explicit-function-np-gradient/57784 , nobody suggested or answered FiniteDiff.jl's finite differencing gradient for getting the numerical derivatives/gradients of an array of values. The answer is either the diff() function or Interpolations.jl, which I already explained in the post why I would want an alternative to those two options to exist, without having to call NumPy's gradient function.
autodiff
- The Elements of Differentiable Programming
-
Astray: A performance-portable geodesic ray tracing library.
I completely agree. Specifying the metric rather than the Christoffel symbols would make it much easier for the users. Something like https://github.com/autodiff/autodiff might just work as the metric tensor is made up of primitives.
-
Point-to-Point Distance Constraint: Gradient of Forward Kinematics
Old username :D So far I have been using Eigen for linear algebra and NLOPT for optimization algorithms. I have found "autodiff" that hopefully looks easy to use: https://github.com/autodiff/autodiff
- Autodiff: Simple C++17 library for Automatic Differentiation
-
Gradients Without Backpropagation
Forward-mode differentiation is easy to implement in C++ with templates, operator overloading, and dual numbers (https://en.wikipedia.org/wiki/Automatic_differentiation#Auto...). Some libraries such as autodiff (https://github.com/autodiff/autodiff) and CppAD (https://github.com/coin-or/CppAD) use this method.
- Ensmallen: A C++ Library for Efficient Numerical Optimization
-
I am creating a fast, header-only, C++ library for control algorithms
I was thinking of adding [autodiff](https://github.com/autodiff/autodiff) in the future, mainly because it works seamlessly with *Eigen*. One big advantage would be that I could use it for AD for NonlinearSystems as well.
What are some alternatives?
Zygote.jl - 21st century AD
CppAD - A C++ Algorithmic Differentiation Package: Home Page
FiniteDiff.jl - Fast non-allocating calculations of gradients, Jacobians, and Hessians with sparsity support
FastAD - FastAD is a C++ implementation of automatic differentiation both forward and reverse mode.
Enzyme.jl - Julia bindings for the Enzyme automatic differentiator
MathNet - Math.NET Numerics
ChainRules.jl - forward and reverse mode automatic differentiation primitives for Julia Base + StdLibs
MKL.NET - A simple cross platform .NET API for Intel MKL
NBodySimulator.jl - A differentiable simulator for scientific machine learning (SciML) with N-body problems, including astrophysical and molecular dynamics
CppRobotics - Header-only C++ library for robotics, control, and path planning algorithms. Work in progress, contributions are welcome!
Tullio.jl - ⅀
PythonRobotics - Python sample codes for robotics algorithms.