merrygoat
commented
5 years ago Fixed in 44baf00f839eaeb3fb296a27000e1038f316d044 Decided to set the unit test to just accept the 3 detections. The issue with the 12E detection was unrelated. Bulk detection of 8K is still under investigation.
The sample 7K cluster is detected multiple times by the unit test.
This is an interesting case where attempting to detect the cluster in isolation is not the same as trying to detect the cluster in bulk.
When an sp3c cluster is detected, the definition of a spindle particle is that is is bonded to all three ring particles. It is common that the two spindle particles are on opposite sides of the ring by geometry but this is not enforced, theoretically if there are two particles on one side of the ring and none on the other we can create and sp3c cluster that looks nothing like the one we imagine. This is very uncommon in bulk since the probability is that the potential spindle candidates (particles bonded to all of the the particles in sp3 ring), which are closest to the ring are on opposite sides.
What to do about this unit test is an interesting question. We can either set up a slightly less minimal system to simulate bulk behavior or we can accept this behavior and write the triple detection into the test.
This is likely the issue that is affecting the rest of the multiple detections (8K, 12E) since they also are also made up from some of the SP3c clusters. It would also be worth checking the FCC and HCP clusters carefully to make sure that they are not affected by this behavior.