GR/StatMech/QM foundations, epistemic views only please

[atyy]

But haven't I already shown that QBism is inconsistent on this point? In QBism, "I am an agent" makes sense, but " We are agents" does not. Yet Fuchs and Schack write "We are agents".

Also, there is no nonlocality problem in the first place, because relativity forbids classical information traveling faster than light. But the nonlocality in quantum mechanics does not lead to classical information traveling faster than light.

So the FMS article is basically wrong. I don't believe this is a matter of interpretation, it is objectively wrong. It is internally inconsistent solving a non-existent problem. I do believe there are viable interpretations in which it can be argued, persuasively, even if I don't agree fully, that if I consider myself classical and everything else is quantum then there is locality. But such an interpretation cannot say "We are agents" without being inconsistent.

 

[RUTA]

RUTA, as I said earlier I really liked the way you put it in this post. I think this puts the epistemic view in a nutshell. It applies to more than just QBism/CBism.


To me the reconciliation does not have to occur in the context of a formal structure (like the "unique spacetime manifold" you offer as example..)
The reconciliation basically amounts to our understanding WHY my (quantum) model of reality differs in some subtle details from yours.
As you suggest "perhaps no unique model is possible".

Indeed this is what I suspect. It seems that forcing A and B at gunpoint to work out a single unique mathematical model of reality leads to "non-locality" or "spooky action at a distance" happening faster than light can travel.

There seems to be a trend towards the epistemic approach (recognizing the subject, the agent or observer as a distinguishing feature of the model) and not just in QM, but also in GR and GR-related areas. Maybe it is another stage in the decline of the "absolute" in physics.

Thanks for your many patient replies, marcus. I think you're right, this is as much as we can glean from this epistemic approach. My complaint/confusion is ontological -- a demand for "absolute" physics, as you say. And, that's exactly what this epistemic approach is telling us to discard. What replaces it? Who knows, but it's not incumbent upon them to say -- at least not at this point.

 

[RUTA]

Also, there is no nonlocality problem in the first place, because relativity forbids classical information traveling faster than light. But the nonlocality in quantum mechanics does not lead to classical information traveling faster than light.

These two facts contribute to the problem, they don't solve it. With these restrictions, how do you explain correlations between space-like separated outcomes that violate Bell's inequality? That's the mystery referred to as "non-locality."

 

[atyy]

These two facts contribute to the problem, they don't solve it. With these restrictions, how do you explain correlations between space-like separated outcomes that violate Bell's inequality? That's the mystery referred to as "non-locality."

But given that there's no logical inconsistency, and good match to experiment, why does it require an explanation?

 

[RUTA]

But given that there's no logical inconsistency, and good match to experiment, why does it require an explanation?

The calculations of physics work and if that's all you care about (a perspective called "instrumentalism"), then no further explanation is needed. But, it's a big problem for those who want to know the nature of reality (ontology) that allows those calculations to work.

 

[atyy]

The calculations of physics work and if that's all you care about (a perspective called "instrumentalism"), then no further explanation is needed. But, it's a big problem for those who want to know the nature of reality (ontology) that allows those calculations to work.

That's an unusual definition. A generalized form of de Broglie-Bohm theory is usually considered to provide a classical ontology but it is nonlocal.

The ontology/instrumentalist divide is not due to nonlocality, but to the measurement problem.

 

[RUTA]

That's an unusual definition. A generalized form of de Broglie-Bohm theory is usually considered to provide a classical ontology but it is nonlocal.

The ontology/instrumentalist divide is not due to nonlocality, but to the measurement problem.

Instrumentalism means that you take your theory to be a computational device that does not necessarily provide any ontological insight. So, an instrumentalist attitude can be applied to any issue in QM.

A nonlocal (superluminal signaling) resolution to the EPR-Bell conundrum is considered problematic by most physicists.

 

[atyy]

Instrumentalism means that you take your theory to be a computational device that does not necessarily provide any ontological insight. So, an instrumentalist attitude can be applied to any issue in QM.

Yes, but that definition holds even for classical physics. For example, why is Newtonian gravity nonlocal? It's generally not considered a problem.

A nonlocal (superluminal signaling) resolution to the EPR-Bell conundrum is considered problematic by most physicists.

But that would be true of Newtonian gravity too.

 

[RUTA]

Yes, but that definition holds even for classical physics. For example, why is Newtonian gravity nonlocal? It's generally not considered a problem.

Actually, it is considered a problem that is resolved by GR.

 

[atyy]

Actually, it is considered a problem that is resolved by GR.

But if you take that to be a problem, then one can always have instrumentalist problems. For example, in GR, what enforces minimal coupling?

Sure we can explain that by taking GR to be quantum spin 2, and using energy conservation.

But then, if GR is quantum spin 2, it is not renormalizable, which means we effectively have a cut off.

Resolutions may be a fundamental lack of Lorentz invariance, in which case locality (in the sense of spacelike operators commuting) is destroyed again. Or maybe we have something like AdS/CFT, which is also nonlocal, and a picture in which spacetime fundamentally does not exist, so we cannot even define locality.

So yes this is a problem if one considers Newtonian gravity problematic.

But it is reasonable not to consider this a fundamental problem.

The fundamental problem in orthodox QM is that one starts by divding the universe into classical and quantum realms. The classical realm includes the measurement apparatus in textbook QM (say Landau and Lifshitz) or the rational agent (in QBism). But if presumably the apparatus or QBist agent is also physical, then shouldn't there be one set of laws covering the quantum and classical realms? The measurement problem is the fundamental problem in QM, not the problem of locality.

 

[marcus]

...The fundamental problem in orthodox QM is that one starts by diving the universe into classical and quantum realms. The classical realm includes the measurement apparatus in textbook QM (say Landau and Lifshitz) or the rational agent (in QBism). But if presumably the apparatus or QBist agent is also physical, then shouldn't there be one set of laws covering the quantum and classical realms? The measurement problem is the fundamental problem in QM, not the problem of locality.

Hi Atyy, don't you think this is getting a bit off topic? I think the aim in this thread is to get a better understanding of these epistemic approaches on their own terms.

BTW I don't recall any passage by Mermin where he says the observer/agent is "classical".
I don't recall the statement of any dichotomy such as you suggest, dividing the world into classical+quantum "realms". You get that dichotomy in historical earlier discussions earlier interpretations like Copenhagen.
In CB, by contrast, agent is neither classical nor quantum. It is something the agent itself is not trying to model.

I don't want this thread to descend to the level of philosophical opinions.I want us to LEARN more about some of these epistemic developments not just in QM but also Thermo & GR on their own terms! So how about we PROVISIONALLY adopt the following uncritical attitude? See'f this'd be temporarily acceptable:
Let's imagine that the "agent" is not trying to understand his own understanding, or include his own knowledge in the physical world he is measuring and modeling. So there is not supposed to be "one set of laws, covering" everything (including even the abstract information and decisions and curiosity that infest the abstract mind of the agent).

I realize this is a bit ridiculous. But let's table the question of what could possibly constitute the non-physical identity of the subject that is trying to understand the physical object.

Let's also see if this is acceptable to RUTA.

RUTA, again I really like the patient clarity with which you have been summarizing the epistemic viewpoint, and thoughtfully weighing it in your estimation. I also sympathize with the ontological hunger you clearly express. The feeling that something is missing, when one is told not to hope for a comprehensive absolute physical model of everything. Or not to count on there being one, anyway. Maybe one should always keep hoping. You actually said this better, I think, a few posts back.

 

[atyy]

Hi Atyy, don't you think this is getting a bit off topic? I think the aim in this thread is to get a better understanding of these epistemic approaches on their own terms.

No, because the measurement problem is the fundamental problem, and collapse of the wave function is part of that problem. One general feature of epistemic views, not particular to QBism, is that the collapse of the wave function as a natural interpretation as a form of Bayesian inference. I do not challenge the coherence and beauty of subjective Bayesian thought.

BTW I don't recall any passage by Mermin where he says the observer/agent is "classical".
I don't recall the statement of any dichotomy such as you suggest, dividing the world into classical+quantum "realms". You get that dichotomy in earlier discussions earlier interpretations.
In CB, by contrast, agent is something the agent is not trying to model.

http://arxiv.org/abs/1311.5253 "There is a vestigial remnant in QBism of the Copenhagen classical domain, but the vestige of this “classical domain” varies from one agent to another and is limited to that agent’s directly perceived personal experience."

I don't want this thread to descend to the level of philosophical opinions.I want us to LEARN more about some of these epistemic developments not just in QM but also Thermo & GR on their own terms! So how about we PROVISIONALLY adopt the following uncritical attitude? See'f this'd be temporarily acceptable:
Let's imagine that the "agent" is not trying to understand his own understanding, or include his own knowledge in the physical world he is measuring and modeling. So there is not supposed to be "one set of laws, covering" everything (including even the abstract information and decisions and curiosity that infest the abstract mind of the agent).

It is not personal opinion if a logical contradiction is demonstrated. I do believe there can be objective discussion of QM interpretation. We judge on whether the predictions of QM are reproduced, and whether the interpretation is internally consistent. Where personal opinion would enter is if given two internally consistent interpretations of QM, one chooses one over the other. My claim is that QBism in saying "We are agents" is inconsistent, since there should be only "I am an agent". Thus "We are agents" as a QBist statement, implies a different level of QBist reality from the QBist reality in which only "I am an agent" makes sense. There is nothing inconsistent with two levels of reality within QBism. But then locality can only be established in one plane of QBism, not both.

To make it clear, I do not challenge that it is reasonable and consistent for a QBist agent not to include himself in the wave function - in fact, I am fond of such a view, which is why I consider this settled and not up for discussion, unless someone wants to talk about it. I do challenge any claim that a theory in which there is no classical/quantum cut cannot exist. And I do assert that "I am an agent", "We are agents", and no nonlocality at all is inconsistent.

 

[RUTA]

But if you take that to be a problem, then one can always have instrumentalist problems.

Again, an instrumentalist solution is to ignore any ontological inference (problematic or not) suggested by an otherwise successful theory. The problem with Newtonian gravity being nonlocal isn't ontological, it's a theoretical problem for those who believe all theories of physics should be mutually consistent. In that case, SR suggests all theories should be local and Newtonian gravity is not. A new theory of gravity, GR, then resolves the theoretical conflict.

 

[RUTA]

atyy's last point might relate to what I don't understand about QBism. The problem I pointed out was that Alice and Bob see "nonlocality" when they construct the M4 depiction of the experiment after exchanging measurement outcomes. That problem exists in the realm "we are agents." If Bob, say, rather sticks to "I am an agent" and records only what he observes (to include his observation of Alice's results sent to him in null or time-like fashion), and he doesn't bother to put her results in an M4 depiction to give credence to her as an agent, then he has no ontological basis for nonlocality. Am I on the right track?

 

[atyy]

Again, an instrumentalist solution is to ignore any ontological inference (problematic or not) suggested by an otherwise successful theory. The problem with Newtonian gravity being nonlocal isn't ontological, it's a theoretical problem for those who believe all theories of physics should be mutually consistent. In that case, SR suggests all theories should be local and Newtonian gravity is not. A new theory of gravity, GR, then resolves the theoretical conflict.

But then can you consistently say why the nonlocality in GR is not an ontological problem while that in quantum mechanics or de Broglie-Bohm theory is?

 

[atyy]

atyy's last point might relate to what I don't understand about QBism. The problem I pointed out was that Alice and Bob see "nonlocality" when they construct the M4 depiction of the experiment after exchanging measurement outcomes. That problem exists in the realm "we are agents." If Bob, say, rather sticks to "I am an agent" and records only what he observes (to include his observation of Alice's results sent to him in null or time-like fashion), and he doesn't bother to put her results in an M4 depiction to give credence to her as an agent, then he has no ontological basis for nonlocality. Am I on the right track?

Yes. I believe that QBism can in some sense consistently assert locality, as long as it expunges statements like "We are agents". In the technical sense, Bell nonlocality has to do with P(a,b|λ), where a and b are classical outcomes of measurements. If one denies that b "exists" far away, then one cannot form the Bell inequality. Now one also cannot then define locality by satisfaction of the inequality. However, one can use other definitions like the existence of a classical M4. So the solution is one agent, one classical M4, one quantum wavefunction for everything else except the agent. I think this is consistent.

 

[RUTA]

But then can you consistently say why the nonlocality in GR is not an ontological problem while that in quantum mechanics or de Broglie-Bohm theory is?

I assume you mean the nonlocality of Newtonian gravity. But, yes, the nonlocality of dBB is a theoretical problem, not an ontological problem, exactly in the same fashion. If I said otherwise, let me correct it now. The instrumentalists who aren't concerned with theoretical integrity would only care about dBB predictions that vary from standard QM. Again, their leit motif is "If the theory predicts what we observe, then I'm happy. Just tell me where it's applicable and how to use it."

 

[RUTA]

Yes. I believe that QBism can in some sense consistently assert locality, as long as it expunges statements like "We are agents". In the technical sense, Bell nonlocality has to do with P(a,b|λ), where a and b are classical outcomes of measurements. If one denies that b "exists" far away, then one cannot form the Bell inequality. Now one also cannot then define locality by satisfaction of the inequality. However, one can use other definitions like the existence of a classical M4. So the solution is one agent, one classical M4, one quantum wavefunction for everything else except the agent. I think this is consistent.

Dude, I think you nailed it. What does marcus have to say?

 

[atyy]

I assume you mean the nonlocality of Newtonian gravity. But, yes, the nonlocality of dBB is a theoretical problem, not an ontological problem, exactly in the same fashion. If I said otherwise, let me correct it now. The instrumentalists who aren't concerned with theoretical integrity would only care about dBB predictions that vary from standard QM. Again, their leit motif is "If the theory predicts what we observe, then I'm happy. Just tell me where it's applicable and how to use it."

Ah ok, that makes sense, I was confused about what you thought about the ontological status of nonlocality in dBB.

I guess the part where we might still disagree is that I consider the achievement of dBB to be the solution of the measurement problem. It is true that Bell nonlocality in the orthodox interpretation of QM is operationally defined, but it inherits that from the classical/quantum split which is within QM logically prior to the Bell nonlocality of QM. I don't think the violation of the Bell inequality requires instrumentalism, since dBB also violates Bell inequalities.

 

[atyy]

Yes. I believe that QBism can in some sense consistently assert locality, as long as it expunges statements like "We are agents". In the technical sense, Bell nonlocality has to do with P(a,b|λ), where a and b are classical outcomes of measurements. If one denies that b "exists" far away, then one cannot form the Bell inequality. Now one also cannot then define locality by satisfaction of the inequality. However, one can use other definitions like the existence of a classical M4. So the solution is one agent, one classical M4, one quantum wavefunction for everything else except the agent. I think this is consistent.

Dude, I think you nailed it. What does marcus have to say?

Well, I'm glad you agree for the moment. I should point out this was all said by Einstein many years ago. http://arxiv.org/abs/quant-ph/0509061 (section 6 on p9)

I should also say that it has been argued that Einstein preferred an epistemic interpretation of the wave function. Of course, an epistemic interpretation does not have to deny the existence of hidden variables. In fact one class of epistemic interpretations is defined using hidden variables.
http://arxiv.org/abs/0706.2661

 

[RUTA]

I guess the part where we might still disagree is that I consider the achievement of dBB to be the solution of the measurement problem.

I think dBB provides a solution to the MP. There are others, e.g., Many Worlds. There's just no consensus among foundationalists as to which is best.

I don't think the violation of the Bell inequality requires instrumentalism, since dBB also violates Bell inequalities.

I agree. If I led you to believe otherwise, let me correct that now. In general, foundationalists abhor instrumentalism. In fact, I frequently have to defend my own interpretation against the claim that it's "merely instrumentalism."

 

[RUTA]

Well, I'm glad you agree for the moment. I should point out this was all said by Einstein many years ago. http://arxiv.org/abs/quant-ph/0509061 (section 6 on p9)

I hadn't seen this view put quite this way. The way I've always seen it stated is a form of "non-separability." That is, the pair of detectors with their space-like separated outcomes are both real, but not truly ... distinct from one another. The correlated, space-like separated outcomes evidence the manner by which the two detectors are "not separate" or "connected" by the source. But, both are equally real. It's interesting to consider Einstein as a QBist

 

[marcus]

Hi RUTA, Atyy,
I've just been rehearsing a great pecs of choral music (Mendelsohn "Elijah") that took much of the afternoon.

Yes. I believe that QBism can in some sense consistently assert locality, as long as it expunges statements like "We are agents"...

Just got back. I think Atyy you can take a statement like "We are all agents" (i.e. in the same application of QM, the app that is being interpreted) as either EXPUNGED or as never having been made.

Talking informally in ordinary English you could say something like "Each of us fills the role of agent in our separate individual application of QM. But that is not reflected in the formal structure of the model. In the formal structure, which is what we are talking about and interpreting, there is one agent.

 

[marcus]

This paper came out recently and may be relevant to the QBist (quantum bayesian) discussion:

http://arxiv.org/abs/1403.1146
Why I am not a QBist
Louis Marchildon
(Submitted on 5 Mar 2014)
Quantum Bayesianism, or QBism, is a recent development of the epistemic view of quantum states, according to which the state vector represents knowledge about a quantum system, rather than the true state of the system. QBism explicitly adopts the subjective view of probability, wherein probability assignments express an agent's personal degrees of belief about an event. QBists claim that most if not all conceptual problems of quantum mechanics vanish if we simply take a proper epistemic and probabilistic perspective. Although this judgement is largely subjective and logically consistent, I explain why I do not share it...

 

[RUTA]

I have the same complaint with QBism -- how can it be used to advance theoretical physics? If you rather assume there is a physical state that needs to be found, then you're at least looking. As Weinberg said about unification -- "It may be there is no simple underlying theory, but if we don't at least assume such a theory exists, we'll never find it if it does."

 

[strangerep]

I notice that QBists / epistemists use phrases like "the wave function corresponds to (lack of) knowledge" in an agent's mind.

But... what then is the Hilbert space? Is it the set of possible configurations of one's ignorance?

If so, then why do we measure only half-integral values for the spins of elementary particles? Surely my ignorance didn't change the classical continuous spectrum into the discrete half-integral spectrum of QM.

IMHO, this fact makes QBist/epistemic interpretations untenable.

 

[atyy]

I notice that QBists / epistemists use phrases like "the wave function corresponds to (lack of) knowledge" in an agent's mind.

But... what then is the Hilbert space? Is it the set of possible configurations of one's ignorance?

If so, then why do we measure only half-integral values for the spins of elementary particles? Surely my ignorance didn't change the classical continuous spectrum into the discrete half-integral spectrum of QM.

IMHO, this fact makes QBist/epistemic interpretations untenable.

In some epistemic interpretations, the physical/epistemic distinction is made as follows.

If the wave function is such that knowing the microscopic state of hidden variables determines the the wave function uniquely, then the wave function is "physical" or "ontic".

But if the wave function is such that the microscopic state of hidden variables does not determine it uniquely, then the wave function is "epistemic".

There are tight constraints on epistemic theories, and there are no hidden variable theories such that the wave function is maximally epistemic. However strong arguments suggest that there are hidden variable theories in which the wave function is epistemic. http://arxiv.org/abs/1303.2834

Rather interestingly, Ballentine has just written a paper defending epistemic views http://arxiv.org/abs/1402.5689.

 

[RUTA]

I notice that QBists / epistemists use phrases like "the wave function corresponds to (lack of) knowledge" in an agent's mind.

But... what then is the Hilbert space? Is it the set of possible configurations of one's ignorance?

If so, then why do we measure only half-integral values for the spins of elementary particles? Surely my ignorance didn't change the classical continuous spectrum into the discrete half-integral spectrum of QM.

IMHO, this fact makes QBist/epistemic interpretations untenable.

IMO, there's nothing wrong with an epistemic view as long as it comes with an ontic view. QBism doesn't say anything about a corresponding ontic view, so I just don't find it interesting per se.

 

[strangerep]

In some epistemic interpretations, the physical/epistemic distinction is made as follows.

If the wave function is such that knowing the microscopic state of hidden variables determines the the wave function uniquely, then the wave function is "physical" or "ontic".

But if the wave function is such that the microscopic state of hidden variables does not determine it uniquely, then the wave function is "epistemic".

Hmm,... but,... I don't see how this really addresses my point: which was about the range of possible values of those physical variables which are not hidden.

Or is the spectrum of values of non-hidden variables determined by a conspiracy among the hidden variables?

There are tight constraints on epistemic theories, and there are no hidden variable theories such that the wave function is maximally epistemic. However strong arguments suggest that there are hidden variable theories in which the wave function is epistemic. http://arxiv.org/abs/1303.2834

Rather interestingly, Ballentine has just written a paper defending epistemic views http://arxiv.org/abs/1402.5689.

I read Ballentine's paper shortly after it appeared and was surprised by the initial tone, which did indeed seem to be (partially) defending epistemic views. But, as I got further into it, his tone seemed to change. I got the impression he was more interested in exploring rigorous models within which such questions can be discussed more transparently, and also to clarify distinctions between (e.g.,) "epistemic" and "subjective" -- cf. the diagram in fig 2 on p8.

Indeed, he writes:

For the record, my own writings on this subject are firmly in the classes of ensemble and objective. So far, I maintain an open mind regarding ontic versus epistemic.

I take this to mean that he has not had a fundamental change of mind, but rather is clarifying some subtleties in the terminology to make them more accessible to physicists who cannot tolerate the usual waffle of philosophy.

IMO, there's nothing wrong with an epistemic view as long as it comes with an ontic view. [...]

If Ballentine's diagram is correct, at least this means the interpretation is necessarily objective.

 

[atyy]

If Ballentine's diagram is correct, at least this means the interpretation is necessarily objective.

I don't think so. In dBB the wave function is ontic, but because dBB is probabilistic it can be subjective (Ballentine's diagram is missing a line from ontic to subjective).