local distances return faulty output

[sina-mansour]

Hi,

I'm using the gdist.local_gdist_matrix function to get the all to all local distances. There are two problems however that seem wrong with the output:

1. The output matrix is not symmetric. However, if a distance from i to j is reported, I don't see why the distance from j to i shouldn't be reported. Am I right to expect a symmetric matrix as an output?

2. Some of the elements of the output matrix are larger than the max_distance provided as an input. Is there something wrong going on in the code?

Thanks.

[liadomide]

@sina-mansour

1. From our tests the symmetry is broken only with a small difference (<1e-6) due to precision issues. Have you observed something different ?
2. Indeed, the max_distance was not considered before. It should be now (after we will merge TVB-2719 Add local_gdist_matrix tests and fix Bug 21 #38 from @ayan-b ). But we still consider acceptable if max_distance will be overcome with a very small Epsilon