XanaduAI/MrMustard

Attenuator not being applied to state

Closed this issue · 2 comments

elib20 commented

Before posting a bug report

  • I have searched exisisting GitHub issues to make sure the issue does not already exist.

Expected behavior

If I apply an Attenuator to a state, I would expect the output state to be mixed and to have had loss applied to it.

Actual behavior

The state does not seem to be affected by the Attenuator.

Reproduces how often

Always.

System information

Mr Mustard: a differentiable bridge between phase space and Fock space.
Copyright 2018-2021 Xanadu Quantum Technologies Inc.

Python version:            3.10.4
Mr Mustard version:        0.3.0-dev
Numpy version:             1.21.6
Numba version:             0.55.1
Scipy version:             1.8.0
The Walrus version:        0.19.0
TensorFlow version:        2.9.1
Torch version:             None

Source code

state = Vacuum(2) >> Sgate(1.15,modes=[0,1]) >> Rgate(np.pi/2,modes=[1]) >> BSgate(np.pi/4,modes=[0,1]) << Fock(4)
state = state >> Attenuator(0.5, modes=[0])
state.is_pure ##returns True but should be False

Tracebacks

No response

Additional information

No response

ziofil commented

oh wow. it seems like channels are not applied correctly when a state is in fock representation:
Fock(10) >> Attenuator(0.5) only affects the norm

ziofil commented

Found the reason: there was a mismatch between is_unitary and _is_unitary. fixing it now