Pierre Yger

At the very least, maybe I should harmonize the way we are handling the notations of merges. In MergeUnitSorting the signature is different, since it is given as units_to_merge, new_unit_ids....

The branch is now almost working, with all the needed features. Everything can be recomputed, even the metrics. I just need to modify a bit the quality_metric_calculator to only recompute...

Actually, we should have a meeting, because there are some corner issues that should be discussed. In fact, if when we do a merge if we change the sparsity mask...

This is good to go, everything is working, with a soft and a hard mode @alejoe91 @samuelgarcia . Several observations: - could be good to have a recompute mode (instead...

I mean that currently, we have a soft mode which does not recompute anything and a hard mode that recompute everything. But there could be an in-between, where merges that...

The waveforms was quite a headache, but ready to read on my side, and all extensions are working! In the soft mode, no recomputation based on the traces is triggered...

Actually, While everything is working, there is a remaining question and issue for PC. I will reslice them as I did for the waveforms, but should I do with the...

PCA is also handled, so now everything is good to go. I've simplified a bit the waveforms, assuming that in soft merges, we can only have reduction of the sparsity...

What are these metrics ? The similarity is handled

Cool, thanks: However, the merge for CCGs is not working anymore because of the update of the compute methods for CCG I guess. I'll try to have a look later