- #1
- 45,830
- 22,746
I have been reading up on the PBR theorem, and it seems to me that there might be an additional assumption that is not discussed in the treatments I have read. The most comprehensive treatment of the general subject of theorems of which PBR is an example (theorems which attempt to show that the quantum state ##\psi## must be ontic) is this paper by Leifer:
https://arxiv.org/pdf/1409.1570.pdf
I'll briefly sketch my understanding of the theorem in order to point out the additional assumption that seems to me to be made, but which I don't see discussed. In Example 7.9 in Leifer's paper, we have two copies of a qubit, each of which we can prepare in one of two quantum states ##|0\rangle## and ##|+\rangle##. The preparations are assumed to be independent, so the quantum state of the two-qubit system after both preparations is one of the four product states ##|0\rangle |0\rangle##, ##|0\rangle |+\rangle##, ##|+\rangle |0\rangle##, or ##|+\rangle |+\rangle##. We then make an entangled measurement of the two-qubit system that projects onto one of four possible orthonormal quantum states ##|\phi_{00} \rangle##, ##|\phi_{01} \rangle##, ##|\phi_{10} \rangle##, or ##|\phi_{11} \rangle## (which are written down in the two-qubit basis in the paper).
Everything so far is just a description of a realizable scenario in terms of standard quantum mechanics. However, we now consider a generic model in which there is an ontic state denoted by ##\lambda##, and the set of all possible ontic states is denoted by ##\Lambda##. A quantum state ##\psi## in this model corresponds to a probability distribution ##\mu_\psi (\lambda)## over the possible ontic states. If we have multiple copies of a system, each with some ontic state, each of which is prepared independently, then the probability distribution ##\mu## corresponding to the quantum state of the full system is just the product of the distributions corresponding to the quantum states of each of the individual systems (this is called the Preparation Independence Principle, or PIP). We then say that the model is ##\psi## epistemic if there are some pairs of quantum states ##\psi##, ##\phi## such that the probability distributions ##\mu_\psi## and ##\mu_\phi## overlap. (We assume in what follows that any such pair of states is not orthogonal, so ##\langle \psi | \phi \rangle > 0##.)
The key point in the proof of the theorem is that each of the four possible outcome quantum states of the measurement described above is orthogonal to one of the four possible two-qubit product states, so if that outcome is obtained, it rules out the two-qubit product state that it is orthogonal to. However, since the individual qubit states ##|0\rangle## and ##|+\rangle## are not orthogonal, if our model is ##\psi## epistemic, then the probability distributions corresponding to them must overlap, so there must be some ontic state ##\lambda## that is contained in both, i.e., that is compatible with both quantum states. That means that the ontic state ##(\lambda, \lambda)## for the two-qubit system must be compatible with all four of the possible outcome quantum states. But that means that if the two-qubit system is in that ontic state, which is a possible ontic state for all four of the possible two-qubit product states after the preparation procedure, it gives a nonzero probability for all four of the possible outcome states. This contradicts the QM prediction that each of the possible outcome states rules out one of the four possible two-qubit product states.
What appears to me to be an additional assumption in all this is the assumption that the entangled measurement procedure does not change the ontic state. The theorem relies on properties of the ontic state that follow from it being contained in the probability distribution of one of the four two-qubit product quantum states. But the quantum state of the system after measurement is not any of those states; it is one of the four entangled outcome states. How do we know that this change in the quantum state is not accompanied by a change in the ontic state? And if the ontic state changes during the measurement, clearly we cannot prove the theorem, because we cannot assume that the ontic state after measurement is still one that is contained in the probability distribution of one of the four product states. All we can assume is that the ontic state after measurement is contained in the probability distribution of the entangled state corresponding to the outcome we measured. And that is not enough to establish the theorem. So it seems like the assumption that the ontic state does not change during the measurement is necessary to prove the theorem.
Am I missing something?
https://arxiv.org/pdf/1409.1570.pdf
I'll briefly sketch my understanding of the theorem in order to point out the additional assumption that seems to me to be made, but which I don't see discussed. In Example 7.9 in Leifer's paper, we have two copies of a qubit, each of which we can prepare in one of two quantum states ##|0\rangle## and ##|+\rangle##. The preparations are assumed to be independent, so the quantum state of the two-qubit system after both preparations is one of the four product states ##|0\rangle |0\rangle##, ##|0\rangle |+\rangle##, ##|+\rangle |0\rangle##, or ##|+\rangle |+\rangle##. We then make an entangled measurement of the two-qubit system that projects onto one of four possible orthonormal quantum states ##|\phi_{00} \rangle##, ##|\phi_{01} \rangle##, ##|\phi_{10} \rangle##, or ##|\phi_{11} \rangle## (which are written down in the two-qubit basis in the paper).
Everything so far is just a description of a realizable scenario in terms of standard quantum mechanics. However, we now consider a generic model in which there is an ontic state denoted by ##\lambda##, and the set of all possible ontic states is denoted by ##\Lambda##. A quantum state ##\psi## in this model corresponds to a probability distribution ##\mu_\psi (\lambda)## over the possible ontic states. If we have multiple copies of a system, each with some ontic state, each of which is prepared independently, then the probability distribution ##\mu## corresponding to the quantum state of the full system is just the product of the distributions corresponding to the quantum states of each of the individual systems (this is called the Preparation Independence Principle, or PIP). We then say that the model is ##\psi## epistemic if there are some pairs of quantum states ##\psi##, ##\phi## such that the probability distributions ##\mu_\psi## and ##\mu_\phi## overlap. (We assume in what follows that any such pair of states is not orthogonal, so ##\langle \psi | \phi \rangle > 0##.)
The key point in the proof of the theorem is that each of the four possible outcome quantum states of the measurement described above is orthogonal to one of the four possible two-qubit product states, so if that outcome is obtained, it rules out the two-qubit product state that it is orthogonal to. However, since the individual qubit states ##|0\rangle## and ##|+\rangle## are not orthogonal, if our model is ##\psi## epistemic, then the probability distributions corresponding to them must overlap, so there must be some ontic state ##\lambda## that is contained in both, i.e., that is compatible with both quantum states. That means that the ontic state ##(\lambda, \lambda)## for the two-qubit system must be compatible with all four of the possible outcome quantum states. But that means that if the two-qubit system is in that ontic state, which is a possible ontic state for all four of the possible two-qubit product states after the preparation procedure, it gives a nonzero probability for all four of the possible outcome states. This contradicts the QM prediction that each of the possible outcome states rules out one of the four possible two-qubit product states.
What appears to me to be an additional assumption in all this is the assumption that the entangled measurement procedure does not change the ontic state. The theorem relies on properties of the ontic state that follow from it being contained in the probability distribution of one of the four two-qubit product quantum states. But the quantum state of the system after measurement is not any of those states; it is one of the four entangled outcome states. How do we know that this change in the quantum state is not accompanied by a change in the ontic state? And if the ontic state changes during the measurement, clearly we cannot prove the theorem, because we cannot assume that the ontic state after measurement is still one that is contained in the probability distribution of one of the four product states. All we can assume is that the ontic state after measurement is contained in the probability distribution of the entangled state corresponding to the outcome we measured. And that is not enough to establish the theorem. So it seems like the assumption that the ontic state does not change during the measurement is necessary to prove the theorem.
Am I missing something?
Last edited: