Hello, :grin: I need to calculate the gradient of a 3 dimensional matrix. Is there today in 2023 an equivalent of numpy.gradient in Julia?( numpy.gradient — NumPy v1.24 Manual ) This question had already been asked, but the answer was not really satisfactory I think. Differentiation without explici…

I did not find any related package on JuliaHub . Perhaps it would be worth to create a new package with such a function? Any comments?

I added a gradient function to CoupledFields.jl (]add CoupledFields#master to try, see Example 3 in the docs ). It uses a different method than numpy.gradient, but has the advantage of working for regular grids or irregular gridpoints. Let me know if it works for your case!

@Mattriks I’m confused by your syntax. If I try using CoupledFields: ndgrid, I get the error ERROR: UndefVarError: ndgrid not defined I also get errors for i = range(-2, 2, 6). How do I run your example?

Can you please tell which exactly method do you use for gradient calculation in case of irregular grid points.

For now, the new function is available on master branch: ]add CoupledFields#master

The fact that this is hard to find may be because in Julia you often don’t actually need to store the gradient as an array (or tuple-of-arrays), instead you compute it pointwise on the fly as an input to whatever other computation you’re actually doing. In other words, the gradient does not need to …

The OP mentions “the gradient of a 3d matrix” (which I assume means 3d positions, and a 1d variable at those positions), can imgradients be used?

Interpolations.jl works for this, too, although with a bit more overhead than in-place calculation: julia> using Interpolations julia> v = rand(3, 3, 3); julia> itp = linear_interpolation(axes(v), v); julia> [Interpolations.gradient(itp, idx...) for idx in knots(itp)] 3×3×3 Array{StaticArraysCor…

Yes, it’s arbitrary-dimensional.

Equivalent of numpy.gradient in Julia

General Usage

Mattriks January 24, 2023, 7:25pm 11

I’m going to add a “Theory” page to the CoupledFields documentation, but for now there is a brief explanation here: Smooth delta functions to compute Jacobian and Hessian from samples - #10 by Mattriks

Topic		Replies	Views
Is there a central difference/gradient function somewhere? General Usage	12	5150	January 10, 2019
Differentiation without explicit function (np.gradient) New to Julia differentiation , gradient	8	5435	June 12, 2021
Q: FiniteDifferences \| gradient of a 3D scalar array Numerics question	5	1045	August 13, 2022
Types and gradients, including Forward.gradient General Usage	20	5640	December 20, 2016
Gradient vector of a matrix of geographical extent General Usage differentiation , calculus , gradient	3	519	September 19, 2022

Equivalent of numpy.gradient in Julia

Related topics