search

NSAPH-Projects / topological-equivariant-networks

E(n)-Equivariant Topological Neural Networks
MIT License
24 stars 1 forks source link

`TopoNetX` causes `scipy` to raise `SparseEfficiencyWarning` #4

Open ekarais opened 9 months ago

ekarais commented 9 months ago

See the code block below:

cc = toponetx.classes.CombinatorialComplex()
# ...add some cells to the cc...
cc.adjacency_matrix(rank=i, via_rank=j)

The call to cc.adjacency_matrix can cause scipy to raise a SparseEfficiencyWarning.

We can try using the index argument of adjacency_matrix to potentially fix this issue, but it may also not help. I could not find an issue on their repository, but I can try to create one with a minimal example. We should try using the index argument anyways as it will remove the need for some manually written functions.

I do not see this as a critical issue, I mainly wanted to document it here to avoid future confusion.

ekarais commented 9 months ago

It turns out that I misunderstood the purpose of index and using it is not helpful for either purpose. It simply returns the cell corresponding to each row/column.