co_autocorrs NAN issue

[Moonzyyy]

I was testing what happened when you insert an array of just 0's into pycatch22 and its local C file. Functions that use co_autocorrs, such as Co_firstmin and co_f1, will face a division by zero error and therefore return NaN as a complex number, this happens when dividing F by the divisor.

However, what I found interesting is that running these functions on pycatch22 will return 0, while if you run them locally on C it returns the length of the array. Is this discrepancy happening due to the python wrapper (python_wrapper_int)?

[Moonzyyy]

Actually, coming back to this, it also happens on some time series values that running the C code locally gives different results than pycatch22.

For example:

As you can see the output running it locally in C is -0.1409879

However on pycatch22 the output is different, this one is:

-0.7025243149147881

Is there a reason for this difference?