I'm starting with this ww=4 a1 = np.random.rand(ww,ww) + 1j*np.random.rand(ww,ww) a2 = Qobj(0.5*(a1 + a1.conjugate().T)) and then running this over and over to see the results: a3, a4 = simdiag([qeye(ww), a2]) a4 What Ive noticed is that for ww=1,2 it gives the same eigenvectors a4 each time I run it, but for larger ww it gives different ones. Am I missing some understanding here? Because I don't see why this would happen Continue reading...