Closed felixerben closed 2 years ago
This can be merged now. As discussed, no comparison of float objects with 0 is done anymore. It is redundant, as the phase convention works also when applied to a purely real number (will be multiplied by e^{0}=1)
This fixes a bug in the phase convention code. The bug would only have been triggered if an eigenvector would have been computed which has an imaginary part that is 0 to floating point precision, and a negative real part.
IMPORTANT: Another wy to fix this would be to remove the if-statement entirely. If that is preferred, I can do that. Please only merge if you are happy with this particular solution.