Unexpected behaviour for ixpeobssim.evt.kislat2015.xStokesAnalysis.calculate_polarization

[doroshv]

Due to masking one can get unexpected results for PA for the said function when arbitrary intensity/stokes parameters are assumed (and I would expect masking to take care of such cases, not make problem invisible). For instance, calling

In [22]: ixpeobssim.evt.kislat2015.xStokesAnalysis.calculate_polarization(array(
...: [500]),array([50.0]),array([-50.0]),array([1]))
Out[22]:
(array([0.14142136]),
array([0.06299155]),
array([-0.39269908]),
array([0.22383074]))

gives correct values, whereas

In [23]: ixpeobssim.evt.kislat2015.xStokesAnalysis.calculate_polarization(array(
...: [50]),array([50.0]),array([-50.0]),array([1]))
Out[23]: (array([1.41421356]), array([0.]), array([0.]), array([0.]))

gives something which is clearly wrong (i.e. PA shall not depend on I). Initialising PA with nans rather than 0 would make it more obvious that something is going wrong.