Pol Febrer Calabozo

I don't understand how you would search for a band gap easily using this. Since each eigenvalue object corresponds to a K point, you'd have to use the method for...

Conceptually it would be something like this, right? ```python def DM_from_eigenstates(eigenstates): # Initialize a density matrix DM = sisl.DensityMatrix(eigenstates[0].parent) # Calculate and store all vectors between atoms Rij = DM.Rij("atom")...

Ok, thanks for the clarifications! > Also one would generally not want a dense matrix as you do it. In can be rather quick in dense form, but will scale...

> Will your finding neighbours be more efficient than close? I know that kdtree's should perform much better, I just never have had the time to look into it! So...

What's the most efficient way to store values in a sparse matrix? I'm giving this a try and I have a loop like: ```python DM = sisl.DensityMatrix(geometry, nnzpr=1000) for i,...

Great, thanks! My intuition was that I could initialize a sparse density matrix from an already existing sparsity pattern, but I couldn't find a way to do that. Is there...

Changing this line https://github.com/zerothi/sisl/blob/2d976d3ceb4424e57904e67890cbd1adf1559715/sisl/physics/sparse.py#L152 to ```python if isspmatrix(P) or isinstance(P, SparseOrbital): ``` works. But probably there should be a type dispatcher, like `Geometry.new`? Because if you provide a `SparseOrbital` it...

Ok, I understood how to access the csr data array directly and now I have an efficient working code to compute the DM (not for NC and SO yet). I've...

Nevermind, I already solved the doubts with discussions at ICN2 :sweat_smile:

Yes, I had also thought this. However standardized methods like `read_geometry` to get a `Geometry` directly still seemed like a good idea 🤔 You don't think so?