gdist_local_connectivity asymmetric

[jescab01]

Dear all,

I am using the tvb-gdist package to compute the local_connectivity over a mesh that I created. The algorithm works fine and creates the sparse matrices of local connectivity. However, I have observed that the eventual matrix is not symmetrical.

I am computing the local connectivity using:

cx = cortex.Cortex.from_file(source_file="cortical_surface.zip", region_mapping_file="region_mapping.txt", local_connectivity_file=None)

cx.local_connectivity = local_connectivity.LocalConnectivity(cutoff=40.0, surface=cx.region_mapping_data.surface, equation=equations.Gaussian(parameters=dict(amp=1, sigma=1, midpoint=0, offset=0)))

Do you know how could that be the case? I have tested with two meshes of 8k and 17k vertices. I attach the data.zip with the 8k mesh (cortical_surface.zip and region_mapping.txt) and the generated sparse matrix (local_connectivity-amp1sig1.mat).

Note: the surface is an open mesh that covers only cortical regions, the rest of the mesh (mostly subcortical) was removed.

Thanks in advance,

[jescab01]

Update: I have also tried to compute the local connectivity with a closed (spherical) surface, and the asymmetry is still there.

[liadomide]

Hi @jescab01
Have you seen this older comment about asymmetry #21 (comment) ?
Could it be the same situation in your case also ?

[jescab01]

Hi @liadomide,

Thanks for the link, I had not read it.

Actually, the differences are relatively small, so I guess they can be due to a limited precision in the computation.

Thanks again,

[liadomide]

@jescab01 Thanks for testing and looking into this youself!
Is then ok if we close this issue?

[jescab01]

Yes, thank you.