Matmul overloaded for correlator class.

[fjosw]

I overloaded the matmul class which should now allow for the multiplication of matrix valued Corr objects with matrices. I will write a few tests and look into edge cases but this solution seems to work.

[fjosw]

I know extended the matmul method to also work on two correlator objects.

Additionally, I added a trace method for matrix correlators.

[fjosw]

I now also implemented a __rmatmul__ method for the multiplication of an array with a Corr. For this to work I had to set __array_priority__ to a high value (see https://numpy.org/doc/stable/reference/arrays.classes.html#numpy.class.__array_priority__). I hope this does not cause any unintended effects.

[fjosw]

Another bug in Corr.roll for matrix correlators fixed.

[s-kuberski]

Strictly speaking, it is not necessary to catch this, since the trace of a 1x1 matrix should be well defined. But of course it does not make much sense.

[fjosw]

I just checked and numpy does not like the trace of an array with a single entry:

np.trace(np.array([1.1]))
> ValueError: diag requires an array of at least two dimensions

[s-kuberski]

Hi,

I did not (yet?) check the code by running it, but so-far everything looks fine. Thanks for taking care of this.

While you are at it: Currently, it is not possible to initialize a Corr object from a one-dimensional array:

li = [pe.cov_Obs(o, 0, '') for o in [1, 2, 3, 4]]
ar = np.array(li)
pe.Corr(li)
pe.Corr(ar)

----> 4 pe.Corr(ar)

     46             # This only works, if the array fulfills the conditions below
---> 47             if not len(data_input.shape) == 2 and data_input.shape[0] == data_input.shape[1]:
     48                 raise Exception("Incompatible array shape")
     49             if not all([isinstance(item, Corr) for item in data_input.flatten()]):

IndexError: tuple index out of range

would you mind looking into this?