SparseDiffTools.jl
SparseDiffTools.jl copied to clipboard
JuliaDiff
Fast jacobian computation through sparsity exploitation and matrix coloring
The following example works fine for sparse `sparsity`, but `sparsity::Diagonal` does not appear to be supported. ```julia function fscalar(x) return dot(x, x) end x = randn(5) sparsity = hessian_sparsity(fscalar, x)...
1. `auto_jacvec!` and `num_jacvec!` only work when the output and input dimensions are the same, i.e., the jacobian is square. The issue is this line (and the equivalent for `num_jacvec!`),...
Hello! I can't seem to use `autoback_hesvec` when the function includes a sparse CSC matrix. ```julia using Zygote, SparseArrays, SparseDiffTools x, t = rand(Float32, 5), rand(Float32, 5) A = sprand(Float32,...
Using the existing hvp operator and the forward over back AD hvp function in this repository I have made the following modified hvp operator. ```julia mutable struct HvpOperator{F, T, I}...
It would be good to port over some of the PolyesterForwardDiff threading support onto the loops here. It wouldn't be too hard but the caches would need to be extended...
In some cases, it is quite handy to use `OffsetArrays` within the residual routine. Yet it seems that the sparsity pattern cannot be detected in that case. Following the example...
Would be great to define a struct that wraps a vector-valued function with a vector input such that when constructing an instance, the colouring is done once using the user-input...
The examples shown in the README.md rely on [SparsityDetection.jl](https://github.com/SciML/SparsityDetection.jl), which has been deprecated: > Note: This repo has been deprecated in favor of ModelingToolkit.jl which can similarly inspect code and...
