Cross-correlation process standard deviation computed incorrectly
Summary
In the cross-correlation process, the standard deviation is computed at each time point.
Steps to reproduce
Invoke the processes.core.multisignal.crosscorr post-process.
What is the current bug behavior?
The standard deviation is computed at each time point. If auto-correlation is computed and the (population) autocorrelation function plotted, the user will see:
- The function not starting at 1. In theory, it should always start at 1.
- Flat function if the input is a population of artificial sinusoids, with noise. In theory, the autocorrelation function in this case should be oscillatory with damping.
What is the expected correct behavior?
It should be computed over time for each replicate.
Logs/Traceback
Not applicable.
Possible fixes
See attachment (credit: @pswain) in relation to the attachment in issue #7 (closed). In particular, line 66.
I...
- Will implement the equivalent in
processes/core/multisignal/crosscorr.py - Will not implement the added plotting functionality here, in favour of creating plotting routines (derived from this) that users should use downstream. One function should do one thing -- computing and plotting should be separate.
Edited by Arin Wongprommoon