Graduate A Realization of a Basic Wigner's Friend Type Experiment

[DarMM]

I don't understand what the validity of this inference has to do with the issue of the parties disagreeing about whether Friend is a quantum system or an external observer.

Without the inference you don't have the disagreements in probabilities that form the core of the refutations in the rest of the article.

 

[charters]

Without the inference you don't have the disagreements in probabilities that form the core of the refutations in the rest of the article.

But only by conceding Wigner and Friend's state spaces are inequivalent, which is the underlying criticism anyway.

 

[DarMM]

But only by conceding Wigner and Friend's state spaces are inequivalent, which is the underlying criticism anyway.

I don't understand, could you explain?

 

[charters]

I don't understand, could you explain?

For Wigner |x>Lab = |x>S⊗|I see x>Friend. The inference you reject clearly holds, at least for Wigner. If he opens the box on Friend's measurement basis, he can simply go directly inspect the qubit for himself, before or after Friend reports anything verbally, and he will see what Friend saw.

For the inference to not hold, |0>S must be defined in a separate state space from the above (one where F is not a tensor factor but the external measurer).

 

[DarMM]

For Wigner |x>Lab = |x>S⊗|I see x>Friend. The inference you reject clearly holds, at least for Wigner. If he opens the box on Friend's measurement basis, he can simply go directly inspect the qubit for himself, before or after Friend reports anything verbally, and he will see what Friend saw.

Yes of course after he measures the entire state of the lab he might obtain $|0\rangle_L$ but the friend measuring and observing $|0\rangle_S$ does not imply that Wigner will with probability 1 observe $|0\rangle_L$. That's the invalid part preventing the contradictory probabilities.

The argument in the paper doesn't even hold in classical epistemically restricted theories, so why it "must" hold in QM I do not understand.

This is the reason FR and Brukner took the approach of wrapping CHSH and Hardy's paradox in Wigner's friend, because basic Wigner's friend holds no problem for these views. See Chapter 11 of Richard Healey's The Quantum Revolution in Philosophy for a full account.

 

[charters]

Yes of course after he measures the entire state of the lab he might obtain |0⟩L|0⟩L|0\rangle_L but the friend measuring and observing |0⟩S|0⟩S|0\rangle_S does not imply that Wigner will with probability 1 observe |0⟩L|0⟩L|0\rangle_L. That's the invalid part preventing the contradictory probabilities

So you are claiming if Friend gets |0>S, there is still non zero probability Wigners gets |1>L, wherein Friend will tell Wigner "I saw 1"?

Brukner & Baumamn I think agree with my view, insofar as they say Friend can't reason correctly without agreeing to use Wigner's state space: https://arxiv.org/abs/1901.11274.

 

[DarMM]

So you are claiming if Friend gets |0>S, there is still non zero probability Wigners gets |1>L, wherein Friend will tell Wigner "I saw 1"?

Brukner & Baumamn I think agree with my view, insofar as they say Friend can't reason correctly without agreeing to use Wigner's state space: https://arxiv.org/abs/1901.11274.

Yes indeed but all of this can occur in a classical epistemically restricted model with local variables, so there's nothing shocking or unacceptable about it to me.

 

[charters]

Yes indeed but all of this can occur in a classical epistemically restricted model with local variables, so there's nothing shocking or unacceptable about it to me.

It entails the conclusion that Friend who lives in Friend-world is not the same person as Friend who lives in Wigner-world. These two versions of Friend disagree about the outcome of the experiment.

 

[DarMM]

It entails the conclusion that Friend who lives in Friend-world is not the same person as Friend who lives in Wigner-world. These two versions of Friend disagree about the outcome of the experiment.

No it doesn't as the exact same mathematical structures and relations hold in Spekkens toy model without this conclusion.

 

[charters]

No it doesn't as the exact same mathematical structures and relations hold in Spekkens toy model without this conclusion.

Can you explain this or give a source then? All I see is you agreeing to the proposition that Friend can both measure 0 and still tell Wigner he saw 1. I can't understand how you could possibly explain this without having two totally disjoint versions of Friend, in a way more extreme than MWI.

 

[DarMM]

Can you explain this or give a source then? All I see is you agreeing to the proposition that Friend can both measure 0 and still tell Wigner he saw 1. I can't understand how you could possibly explain this without having two totally disjoint versions of Friend, in a way more extreme than MWI.

An explanation within Spekkens model is here:
https://www.physicsforums.com/threa...ncomplete-comments.966033/page-3#post-6152735

 

[charters]

An explanation within Spekkens model is here:
https://www.physicsforums.com/threa...ncomplete-comments.966033/page-3#post-6152735

I don't see how this is relevant. In that post, you say a state |00> is "compatible with the above use of a superposition by the superobserver" |000> + |111>. This was never in issue. The claim you agreed to here is that |00> is further compatible with |111> alone, or that is at least how I read #127.

 

[DarMM]

I don't see how this is relevant. In that post, you say a state |00> is "compatible with the above use of a superposition by the superobserver" |000> + |111>. This was never in issue. The claim you agreed to here is that |00> is further compatible with |111> alone, or that is at least how I read #127.

If you look at my post and know how measurements work in Spekkens model that is also true. If you don't know the details of measurements in Spekkens model I can go through the details here.

 

[charters]

If you look at my post and know how measurements work in Spekkens model that is also true. If you don't know the details of measurements in Spekkens model I can go through the details here.

I reread your post and the measurement subsection the wikipedia entry for Spekkens model, and I can't see how it possibly allows F to get |00> and W to get |111> (absent the solipsist/single user premise Felline discusses). So I'd appreciate the details.

 

[DarMM]

I reread your post and the measurement subsection the wikipedia entry for Spekkens model, and I can't see how it possibly allows F to get |00> and W to get |111> (absent the solipsist/single user premise Felline discusses). So I'd appreciate the details.

No worries I'll try to put it up later today.

 

[DarMM]

I drop normalization here. I think I got confused at one point, so I will start again.

So first do you agree there is no problem with Wigner's measurements in the $\{|000\rangle,|111\rangle\}$ basis and the friend obtaining outcomes like $|00\rangle$?

This is no more a problem than you would have with two states in statistical mechanics being used at two different levels. The statistics match, both obtaining each outcome 50% of the time. One need only reason that each time the friend gets $|00\rangle$ Wigner will get $|000\rangle$ and similarly for $|11\rangle$. Wigners "uncollapsed" probability reflects only his ignorance of what has occurred.

It is in fact measurements along other bases like $\{|000\rangle + |111\rangle, |000\rangle - |111\rangle\}$ demonstrating inteference that are the problem.

 

[charters]

One need only reason that each time the friend gets |00⟩|00⟩|00\rangle Wigner will get |000⟩|000⟩|000\rangle and similarly for |11⟩|11⟩|11\rangle. Wigners "uncollapsed" probability reflects only his ignorance of what has occurred.

Yes I agree with this of course, but this is not what you were suggesting before. You were saying in a run when F gets |00>, W can get |111> and furthermore that this does not imply a single user/solipsism interpretation. This was what you suggested to undermine Felline's argument. So, to be clear, 1) are you no longer making this claim and 2) if not, where is your disagreement with Felline?

I will also mention I spent some time with Spekkens' original paper yesterday, and I would caution against drawing any conclusions about a QBist or Bub type view based on what works in Spekkens toy model. The latter handles Wigner's Friend in the same manner as a hidden variables approach, and it is still an ontological model at heart. The defective interpretations are these fully non-ontological, informational ones.

 

[DarMM]

Yes I agree with this of course, but this is not what you were suggesting before. You were saying in a run when F gets |00>, W can get |111> and furthermore that this does not imply a single user/solipsism interpretation. This was what you suggested to undermine Felline's argument. So, to be clear, 1) are you no longer making this claim and 2) if not, where is your disagreement with Felline?

As I said I got confused about what we were discussing. I'm aware of how Spekkens model differs.

So as that we can focus the discussion and so that I am not discussing the wrong thing, what to you is the point Felline makes that refutes these views?

 

[charters]

As I said I got confused about what we were discussing. I'm aware of how Spekkens model differs.

So as that we can focus the discussion and so that I am not discussing the wrong thing, what to you is the point Felline makes that refutes these views?

These views being Bub's type of view?

 

[DarMM]

These views being Bub's type of view?

Primarily, although if you wish to include QBism that is fine as well.

 

[charters]

Primarily, although if you wish to include QBism that is fine as well.

The problem is F and W don't agree whether F gets entangled with the qubit or causes a collapse. In the former case, the state after F measures the qubit is (4) on pg 7 of Felline. In the latter the state is either (5) or (6). These behave differently "at the second beamsplitter" and affect W's measurement on the +/- or "ok/fail" basis. Observers who share a reality can't disagree about this.

But I stress this is only a provisional problem, not a knockout inconsistency in QT that makes it un-usable. It can be cured in various ways. At least some QBists pay the price of accepting QT as a single user theory, and adopting a general worldview along these lines. Spekkens, Bohmians, TSVF go to hidden variables. MWI and GRW have their obvious answers. But the pied piper has to be paid in some way. The neo-Copenhagen/informational folks seem to deny the existence of price they pay (single user) so I am reluctant to credit them with having a tenable position by imputing one of the acceptable cures.

 

[DarMM]

The problem is F and W don't agree whether F gets entangled with the qubit or causes a collapse. In the former case, the state after F measures the qubit is (4) on pg 7 of Felline. In the latter the state is either (5) or (6). These behave differently "at the second beamsplitter" and affect W's measurement on the +/- or "ok/fail" basis. Observers who share a reality can't disagree about this.

Okay let us stick to Spekkens model for the moment. It can be seen in that model that Wigner uses a state like $|000\rangle + |111\rangle$ but this is not in conflict with the $|00\rangle$ of the friend. I agree with Baumann in the paper you linked with above that the friend would be wrong to conclude that the state of the entire lab is $|000\rangle$.

Now how does the use of $|000\rangle + |111\rangle$ by Wigner and $|00\rangle$ by the friend imply solpsism of some form?

 

[charters]

Now how does the use of |000⟩+|111⟩|000⟩+|111⟩|000\rangle + |111\rangle by Wigner and |00⟩|00⟩|00\rangle by the friend imply solpsism of some form?

It doesn't. The problem is when Wigner measures on the +/- basis. The state after F measures, but before the path recombination is either

W says: |00>+|11> ⊗|ready to measure >

or

F says: |00>⊗|ready to measure >

After (attempted) recombination, these become:

W says: |0+1>⊗|F's brain erased>⊗|+>

or

F says: |00+>+|00->

W says he never gets |->, F says W does. F rejects the premise that (properly) wiping his brain restores the original state of the qubit, W assumes it does. If both these people are correct, they don't share a reality.

 

[DarMM]

Sorry perhaps I'm missing something. Wigner and the Friend have the same number of qubits there. Wigner should have three (system, device, lab) and the friend two (system, device).

 

[charters]

Sorry perhaps I'm missing something. Wigner and the Friend have the same number of qubits there. Wigner should have three (system, device, lab) and the friend two (system, device).

The tension arises when Friend makes a prediction about what Wigner will measure. Are you claiming this is impossible?

 

[DarMM]

The tension arises when Friend makes a prediction about what Wigner will measure. Are you claiming this is impossible?

For the friend to make predictions about what Wigner will measure? No, but what is the tension? He would reason his own experiences permit the use of $|00\rangle$ in future predictions about the system and the device. However he would reason the superposed state is the correct one for Wigner to use rather than $|000\rangle$. Neo-Copenhagenists such as Richard Healey and others agree that the superposed state is correct for measurements on the entire lab that Wigner is capable of. How is this in tension with the $|00\rangle$ for the device and system?

 

[charters]

For the friend to make predictions about what Wigner will measure? No, but what is the tension? He would reason his own experiences permit the use of $|00\rangle$ in future predictions about the system and the device. However he would reason the superposed state is the correct one for Wigner to use rather than $|000\rangle$. Neo-Copenhagenists such as Richard Healey and others agree that the superposed state is correct for measurements on the entire lab that Wigner is capable of. How is this in tension with the $|00\rangle$ for the device and system?

If the device-system is in |00>, it does not behave the same under recombination as |00>+|11>. Friend can't have it both ways. Either he thinks the state collapsed to |00> in which case he predicts no interference effects from the |11> term under recombination, or he thinks the device/system got entangled, in which case he does predict interference effects. Friend can even pass a note to Wigner with this prediction on it as it does not depend on whether Friend saw 0 or 1.

Or, if Friend's state is only valid up to Wigner's choice to do a recombination experiment, at which point Friend has to switch to using Wigner's state (this being one of Baumann & Brukner's ideas to resolve the tension), then Friend is admitting they need MWI or HV type reasoning, at least in some cases/for some predictions. At this point, they might as well just apply this full time, since it will work in all cases.

 

[DarMM]

If the device-system is in |00>, it does not behave the same under recombination as |00>+|11>

You're dropping a system here which is crucial. The friend is using $|00\rangle$ (note two systems) and Wigner is using $|000\rangle + |111\rangle$ (note three systems). The question concerns what is the conflict between the $|00\rangle$ used by the friend and the $|000\rangle + |111\rangle$ used by Wigner.

Wigner does not use $|00\rangle + |11\rangle$

 

[charters]

You're dropping a system here which is crucial. The friend is using $|00\rangle$ (note two systems) and Wigner is using $|000\rangle + |111\rangle$ (note three systems). The question concerns what is the conflict between the $|00\rangle$ used by the friend and the $|000\rangle + |111\rangle$ used by Wigner.

Wigner does not use $|00\rangle + |11\rangle$

This notational difference doesn't matter. The physical experiment is equivalent to Wigner sending Friend's entire lab through an interferometer. The interferometer has 2 exit ports labelled |+> and |->. The question to both parties is simply the statistics for when the lab arrives at each exit. W says it is 100% at |+> when he does a recombination. F says it will be 50/50, if he genuinely believes his measurement collapsed the state. So either someone discovers he is wrong, or they must live in disjoint realities. Friend is wrong = many worlds/hidden variables. Wigner is wrong = GRW. Disjoint realities = Copenhagen/QBism

 

[DarMM]

This notational difference doesn't matter...F says it will be 50/50, if he genuinely believes his measurement collapsed the state

It does, because it is central to how QBism and Neo-Copenhagenism resolve this point. Why would the friend using the $|00\rangle$ state after his measurement believe the outcome of such an interferometry experiment on the entire lab to have 50:50 odds? To obtain these odds for the entire lab he'd have to ascribe it the state $|000\rangle$, but why would he do this on the basis of the $|00\rangle$ state for the device and system alone? He didn't observe the lab down to the atomic level.