The hypersphere_overlap_of_subset method failed if the data had any redundant feature. This is a pretty common case, e.g. finding a diverse set of alcohols (all binary chains will have a common feature). This was making grid_partitioning methods to fail.
The fix removes redundant coordinates before normalizing and computing diversity. Now it only fails if all binary chains in the data are the same.
The hypersphere_overlap_of_subset method failed if the data had any redundant feature. This is a pretty common case, e.g. finding a diverse set of alcohols (all binary chains will have a common feature). This was making grid_partitioning methods to fail.
The fix removes redundant coordinates before normalizing and computing diversity. Now it only fails if all binary chains in the data are the same.