Incorrect sparsity pattern on simple example

[thowell]

Hello! Awesome package.

I'm trying to extend the examples and have found a simple case that fails:

function f!(dx,x)
dx[1:5] = x[1:5] + x[1:5]
nothing
end
output = zeros(5)
input = ones(5)
sparsity!(f!,output,input)

The above function 'f' should return the same sparsity pattern as the function 'g' below:

function g!(dx,x)
dx[1:5] = x[1:5]
nothing
end

but when called on 'f' the sparsity pattern returned is empty.

Is my usage incorrect?

[ChrisRackauckas]

Hey, this package will soon be supersceded by ModelingToolkit's sparsity detection algorithms, which are documented here: https://mtk.sciml.ai/dev/highlevel/#Sparsity-Detection-1 . You might want to try those since it's more robust. See https://mtk.sciml.ai/dev/tutorials/auto_parallel/ as an example with automated sparse Jacobians, or you can replace the actual symbolic Jacobians with just jacobian_sparsity to get the sparsity pattern and it's effectively the same algorithm as what's done here, except using MTK for the abstract interpretation.