Fix dropped state collapse in FreeQuantumState.measure_multiple

[HasanTasdiq]

Summary

FreeQuantumState.measure_multiple (sequence/kernel/quantum_state.py) computes the post-measurement state correctly but then assigns it to attributes that do not exist on FreeQuantumState:

new_state = new_states[res]
for state in entangled_list:
    state.quantum_state = new_state          # FreeQuantumState has no `quantum_state`
    state.entangled_photons = entangled_list # FreeQuantumState has no `entangled_photons`
return res

FreeQuantumState stores its vector in self.state and its entangled partners in self.entangled_states (see the single-qubit measure method, which does s.state = new_state). Writing to quantum_state / entangled_photons silently creates unused attributes and leaves the real state uncollapsed: after measure_multiple, the entangled FreeQuantumState objects keep their pre-measurement superposition.

Demonstration

Two qubits each in |+>, combined, then measured in the computational basis:

res = FreeQuantumState.measure_multiple(basis, [a, b], rng)   # e.g. res == 2  (=> |10>)
a.state   # buggy: still (0.5, 0.5, 0.5, 0.5); expected (0, 0, 1, 0)

This is reached in practice by Photon.measure_multiple, used by PolarizationBSM and TimeBinBSM. The returned index res (which drives detector firing) is correct, so detector outcomes are unaffected; but the measured photons' tracked state is left wrong, which is incorrect for any consumer that inspects photon state after a multi-qubit measurement.

Fix

Assign the collapsed vector to the real fields, mirroring the single-qubit measure method:

new_state = tuple(new_states[res])
for qs in entangled_list:
    if qs is not None:
        qs.state = new_state
        qs.entangled_states = entangled_list
return res

Test plan

• Added test_measure_multiple_collapses_state, which combines two |+> photons, measures in the computational basis, and asserts the shared state collapses to the measured basis vector (and stays shared).
• Confirmed the new test fails on the current code and passes with the fix.
• Existing test_measure_multiple and other photon/kernel tests still pass.

🤖 Generated with Claude Code

[hayekr]

Thank you for your contribution. Please see my suggestions before we merge this to RnD.

[hayekr]

please call this quantum_state. Avoid abbreviations when able.

[hayekr]

It would be good to add assertions preventing these invalid class attributes from being used.

assert not hasattr(photon1.quantum_state, "quantum_state")
assert not hasattr(photon1.quantum_state, "entangled_photons")
assert not hasattr(photon2.quantum_state, "quantum_state")
assert not hasattr(photon2.quantum_state, "entangled_photons")

[hayekr]

You check that state is shared but we should also check that entanglement_states is also shared
assert photon1.quantum_state.entangled_states is photon2.quantum_state.entangled_states