[Feature] Synchrofact Detection

[Kleinjohann]

This is a method to detect highly synchronous spikes (at the level of sampling rate precision with an option to extend this to jittered synchrony) and annotate or optionally remove them.

The detection of complexity intervals (time intervals with at least n spikes separated by strictly fewer than spread - 1 empty bins) is particularly suited for artefact detection and analysis, but can be used to get a (time-resolved) overview of complexities at any time-scale. This is different from statistics.time_histogram and statistics.complexity_pdf since it uses the amount of empty bins separating spikes, not a simple bincount and not a moving window with a fixed width while still providing time-resolved output.

I started from an institute-internal script and adapted it to use a BinnedSpikeTrain in the histogramming procedure and I adapted a set of basic tests I wrote for said internal script.

[pep8speaks]

Hello @Kleinjohann! Thanks for updating this PR. We checked the lines you've touched for PEP 8 issues, and found:

There are currently no PEP 8 issues detected in this Pull Request. Cheers! 🍻

Comment last updated at 2020-12-03 15:57:25 UTC

[coveralls]

Coverage increased (+0.2%) to 90.137% when pulling 4103045 on INM-6:synchrofact_detection into 8e465fd on NeuralEnsemble:master.

[morales-gregorio]

Solves #326

From my side this PR is ready for review.

[dizcza]

There are many small remarks, but two major concerns are:

• consider switching to OOP;
• consider out-of-place processing instead of in-place spike removal.

We'll discuss this tomorrow anyway.

---

Also, I don't see any references. Is this completely your own implementation and ideas?

[dizcza]

I find this PR ready for the final review and merge.
Since it depends on another ongoing PR (in terms of the file location of your methods) about the Spike Contrast method, let's wait for the latter to be merged first.

[dizcza]

Ready for the final review.

[Kleinjohann]

I have fixed everything we discussed in our meeting on this and addressed all comments in here.

I spotted some double backticks in docstrings, are those a problem?

[dizcza]

Thanks a lot!

I spotted some double backticks in docstrings, are those a problem?

No, it's not a problem. Double vs single backticks is a matter of choice.

I'll merge it as soon as the tests pass.

P.S. Consider also adding into viziphant some of the nice synchroplots that you showed to us a month ago.