Long-distance correlation, information, and faster-than-light interaction

[edguy99]

[Mentor's note: Split off from https://www.physicsforums.com/threads/an-abstract-long-distance-correlation-experiment.852684/page-6 because this subthread was discussing possible mechanisms behind the phenomenon]

Four steps on the road from not at all weird to completely weird:
1.) Classical communication
2.) Quantum communication.
3.) PRBox Communication.
4.)FTL communication.

My take on these:

1) - not weird at all - you send a photon in one direction and the other guy gets it
2) - not weird at all - you send two photons with the same polarization (defined by its Jones Vector) in two directions and if one guy gets it, he knows what the other guy will see for polarization subject to the inherent randomness in the Jones Vector definition of photon polarization.
3) - PRBox - not sure what this is, a google search produces a wide variety of topics ?
4) - not weird at all - it doesn't happen

What I think is weird:

1) Ignoring the fact that photon polarization has inherent randomness and designing FTL or multi-world theories based on this assumption
2) Not using the "weirdness" of the experiment in question here to develop theories of what the inherent randomness of the photon is all about and how it works (the hidden variables behind its behavior).

What I look forward to:

A data set of matrices based on this experiment that someone feels cannot be explained. We could then take a careful look at the definition of photon polarization (without resorting to FTL or multi-world theories) to see how to reproduce the experimental results in the animation leading to a better understanding of "weirdness".

 

[maline]

Ignoring the fact that photon polarization has inherent randomness and designing FTL or multi-world theories based on this assumption

Noone serious does this. Please read up on Bell's theorem.

to develop theories of what the inherent randomness of the photon is all about and how it works (the hidden variables behind its behavior).

This is exactly what Bell proved impossible, assuming no FTL transmission and one real world.

 

[edguy99]

Noone serious does this. Please read up on Bell's theorem.

This is exactly what Bell proved impossible, assuming no FTL transmission and one real world.

I have read a lot on Bell's theorem, but I feel this is a change in topic. The original post describes a specific Bell type of experiment and the weirdness that this experiment is providing. If the FTL explanation is all that works to explain this experiment, then that is pretty weird.

Bell's theorem is true for the models it covers, check the definitions. It does not cover all possible models of the photon. In fact, Bell's theorem proves that his type of model is not a proper model of a photon since it doesn't work.

 

[Ken G]

By the way, there is nothing in this experiment that requires anything FTL, so we should not regard the need for FTL as the source of the weirdness. FTL only appears if you regard information as having a location other than in the intelligent brain that is analyzing that information, and I see no evidence that information does exist anywhere else. There certainly is no physics theory that requires information to exist any place else. So if we are to say the results of the experiment are "weird", we must be able to say why it is weird, without invoking anything FTL, or else it is our own belief system that is imposing the weirdness and the solution is to simply stop thinking about information as if it had a location other than in our heads.

 

[ddd123]

I don't follow, what's special about the brain anyway? The measurement results are correlated depending on the settings in a FTL fashion.

 

[Ken G]

No experiment ever requires that we assign a location to information, other than where we see it operating-- in the brain. No one would say "knowledge" has a location, so why should we imagine that "information" is any different from "knowledge"? It doesn't show up any differently. It's true that once a hand of cards is dealt, we can imagine that the "information" of what the cards are exists locally in that hand, but we never use that claim-- ever. The only information we ever use is what we know, because what could we use but what we know?

 

[ddd123]

Why does it have to be information? And even then, everything would be that way, all experimental results, all physical settings, including slower-than-light interactions. Why would you conclude that FTL should be ignored then?

 

[ddd123]

Ah sorry, there was no quote in your post. Though I don't think this reasoning is sound, because referring to gathering information in the abstracted fashion you're pointing to and claiming it eludes the FTL problem is a category mistake: "speed" is unapplicable here, either way, and your information just refers to what happened where. So the phenomena you refer to and which you have information about still have a FTL property. And I'm assuming what you're saying about information is consistent, which I doubt.

 

[Ken G]

I'm not sure what you are arguing, I agree that information only refers to what happened where. There is never any need for anything FTL in this experiment, that all comes from the idea that a system must in some sense itself "store" information locally, which is the idea we need to let go of. If you just put information processors everywhere, and "store" the information in them, rather than in the system, you never need the processors to communicate FTL-- their information is local, it is not FTL, and they can predict any correlations in any experiment they have access to. Go back to the example of the highly cooled white dwarf, an information processor anywhere in that white dwarf will discover it cannot alter the momentum state of the electrons it encounters. It will not require anything FTL to happen to reach that conclusion, it is a simple local experiment, but the processor will be able to conclude that it is in a Fermi sea all the same.

 

[ddd123]

There is never any need for anything FTL in this experiment, that all comes from the idea that a system must in some sense itself "store" information locally, which is the idea we need to let go of.

We can let go of it, but it doesn't change the fact that a polarizer angle choice on Alice's side can influence the measurement outcome on Bob's side FTL.

 

[Ken G]

I agree we don't want to get caught up in this, but there is simply no reason to say that anything that happens at Alice "influences" what happens at Bob. The experiment shows there is a correlation, not an influence. Indeed, one cannot even say which event influenced the other, it just isn't the way "influences" normally work. I think the whole concept of "influence" is misplaced in entanglement experiments, but all that can be demonstrated is that the concept is never necessary to make the correct predictions. Similarly, when we find we cannot put an electron into a momentum state already occupied by another electron in a white dwarf, even though we have no idea where that other electron is (and indeed, there is no such "other electron", electrons are indistinguishable), we have no need to say that other electron "influenced" the one we are experimenting on. We simply find that the state of the system does not permit the momentum change-- that's all we actually observe, and no other concept is necessary.

 

[ddd123]

The shortcoming of language doesn't call for dropping the issue. The issue of FTL is still there. Sure there's a time ordering problem, but if Alice had chosen a different angle Bob's result would've been FTL different (to not say instantaneously).

 

[Ken G]

All we can say is the correlations observed depend on the choice of the angles, that's it. Interpreting that as an "influence" on the events is never necessary, it is unnecessarily limiting the language. The correlations depend on the state of the system, and the measurements done on the system, that's what we observe.

 

[ddd123]

Maybe it has to do with SR's framework where FTL is just not possible, I guess that's why Bohmian mechanics needs Lorentz's ether. This is where my little knowledge hits the wall. I hope for other users to comment further.

 

[georgir]

... a "classical" model where each photon has a polarization angle λ... This illustrated statistical model of the photon behaves exactly as the quantum mechanical model behaves.

Maybe "exactly like QM" for a single photon, but not for multiple entangled particles. It does not reproduce the Bell violations. But I am sure you already know this. Also you probably know the trivial "non-classical" hidden FTL mechanism that needs to be added to this model in order to make it reproduce QM entanglement. If I am wrong and you don't, maybe we should discuss this in a different thread, to not sidetrack this one.

TBH I am not exactly sure what your purpose of bringing this up here is...
I also did not quite understand your relationship to the simulation website you posted earlier, that used this model:
http://www.animatedphysics.com/games/photon_longdistance_nonlocality.htm
If you are the author of that simulation and you want to make it useful for the current topic, maybe you can add the FTL mechanism to it (or make it optional for bonus points?) so it matches QM and reality, and also add a way to more quickly simulate a large number of trials so the end data can finally get some statistical significance.
And even then, I'm not sure now useful it would be for the current topic. It can certainly help people new to the subject in understanding the Bell violation that's happening in reality, and how a FTL effect helps in simulating/explaining that, but I think this thread is already targeted at people understanding this well enough.

 

[georgir]

All we can say is the correlations observed depend on the choice of the angles, that's it. Interpreting that as an "influence" on the events is never necessary, it is unnecessarily limiting the language. The correlations depend on the state of the system, and the measurements done on the system, that's what we observe.

But exactly there is the tricky part. The correlations depend on the state of the system as a whole. You can not break it apart and say this measurement depends only on the state in this part of the system. You either have to stick to always looking at the whole thing only, or you can break it up into local parts but then have to admit a FTL influence between them.

 

[maline]

Dehlinger and Mitchell use a "classical" model where each photon has a polarization angle λ.

In their words "Our Hidden Variable Theory is very simple, and yet it agrees pretty well with quantum mechanics. We might hope that some slight modification would bring it into perfect agreement."

You are ignoring the very next sentence in their text: "In 1964 Bell showed that this is impossible." Their model is intended as an example of what doesn't work.

 

[ddd123]

But exactly there is the tricky part. The correlations depend on the state of the system as a whole. You can not break it apart and say this measurement depends only on the state in this part of the system. You either have to stick to always looking at the whole thing only, or you can break it up into local parts but then have to admit a FTL influence between them.

In the latter case, how do you consider it with respect to different inertial frames? In a frame where Bob measures first, it's Bob influencing FTL and Alice's side is being influenced, and vice-versa?

 

[Ken G]

In the latter case, how do you consider it with respect to different inertial frames? In a frame where Bob measures first, it's Bob influencing FTL and Alice's side is being influenced, and vice-versa?

Exactly, if you regard it as an "influence", it's more than just FTL, it requires that the influence be able to go back in time. That defeats the whole purpose of the "influence" concept, because you have lost the concept of one-way causation. You can still have causation emerge as some kind of macroscopic limit that is one-way even though it is comprised of two-way "influences", but then we've lost the "influence" concept as it is normally formulated. If we're going to lose that concept anyway, I say just dump it from the start. Electrons in white dwarf stars do not "influence" each other, they are simply indistinguishable. Parts of an EPR experiment do not "influence" each other, they are simply entangled. A system is what it is, if we wish to imagine it is broken into pieces that "influence" each other, that's on us-- and when it doesn't work, it's on us to recognize it isn't working, and use a different picture.

 

[Ken G]

You either have to stick to always looking at the whole thing only, or you can break it up into local parts but then have to admit a FTL influence between them.

Yes-- so why not look at the whole thing? You never see anything FTL if you look at only part of a system, you only see it when you look at the whole thing (the correlations), but also insist on imagining it is made of pieces that "influence" each other. Why does a system have to be made of pieces that influence each other, when that is simply not what we are observing to be the case? You can observe many types of influences by intercepting the influencing signal, so when you cannot do that, as here, it is a good indication this is the wrong way to think about it.

 

[ddd123]

It's not catastrophic since there are "locally retrocausal" hidden variable models. Also the FTL scenario in SR already means "backward causation", so you could say this is a peculiar case of this being possible since there are no inconsistencies in spacetime involved (like with closed timelike loop solutions in GR). Though it's more nuanced, because was it Alice "retro-influencing" Bob or just Bob influencing Alice? Of course you can't say that the local information is enough, the correlation is explained only by looking at the whole system nonlocally if you want to avoid talking about FTL (which, to me, isn't that different other than a rephrasing).

I would really like georgir's take if he has something to add.

 

[ddd123]

Yes-- so why not look at the whole thing? You never see anything FTL if you look at only part of a system, you only see it when you look at the whole thing (the correlations), but also insist on imagining it is made of pieces that "influence" each other. Why does a system have to be made of pieces that influence each other, when that is simply not what we are observing to be the case? You can observe many types of influences by intercepting the influencing signal, so when you cannot do that, as here, it is a good indication this is the wrong way to think about it.

Well, if you just look at the whole thing in one go, it's still self-influencing (FTL!) because of different choices that could've been made :P

I know I sound stubborn, it's just that dismissing the problem in that way just seems to me to be sweeping it under the carpet.

 

[Nugatory]

I know I sound stubborn, it's just that dismissing the problem in that way just seems to me to be sweeping it under the carpet.

Well, this thing that you're calling "sweeping it under the carpet" is something that Neils Bohr would call laudable intellectual rigor and hard-headed rejection of unwarranted assumptions. You don't like it, other people do, and both positions are basically a matter of personal taste... If this were broccoli and cabbage instead of physics we'd know better than to argue about it.

Arguments about interpretations cause some of our members to break out in hives, so if this thread deteriorates into a one of those arguments we may have to close it as a public health measure.

 

[ddd123]

You don't like it.

Not necessarily. I don't know about that, I said "it seems to me" with the limited judgement I have available. My purpose is to have the other outlook explained to me, since the FTL one has its own problems. So once I understand the other position I can agree to disagree or embrace it.

 

[Ken G]

Well, this thing that you're calling "sweeping it under the carpet" is something that Neils Bohr would call laudable intellectual rigor and hard-headed rejection of unwarranted assumptions. You don't like it, other people do, and both positions are basically a matter of personal taste... If this were broccoli and cabbage instead of physics we'd know better than to argue about it.

Yes, I think that sums it up. There doesn't have to be anything FTL or retrocausal in entanglement, but we get those if we insist on other assumptions. It's too high a price for my taste, so I avoid assuming that correlations that respond to our choices of what to measure must be explained either by "information carried with the parts" or by "influences between the parts." The problem is treating a system like a sum of independent parts, that's what doesn't seem to work. Entanglement is our reason to drop that approach, just as we did with the indistinguishable electrons in a white dwarf.

 

[Mentz114]

Not necessarily. I don't know about that, I said "it seems to me" with the limited judgement I have available. My purpose is to have the other outlook explained to me, since the FTL one has its own problems. So once I understand the other position I can agree to disagree or embrace it.

One 'other position' is the field approach. We have a quantum field which is in a two particle state ( it has 2 child particles ). Immediately after the field is prepared it has angular momentum $J_0$. When the first measurement is made an amount $\pm j$ of angular momentum is assigned to one particle, leaving the field (immediately) with angular momentum $J_0\mp j$. By the time the second measurement is made this has been corrected by assigning $\mp j$ to the second particle. Nothing strange has happened. It is accepted that the energy and momenta of a field (quantum or classical ) can do this. Unless a fundamental conservation law is to be broken ( even for an infinitesimal time ) this is the only thing that can happen.

Maybe it's naive but it works for me.

 

[edguy99]

Maybe "exactly like QM" for a single photon, but not for multiple entangled particles. It does not reproduce the Bell violations. But I am sure you already know this. Also you probably know the trivial "non-classical" hidden FTL mechanism that needs to be added to this model in order to make it reproduce QM entanglement. If I am wrong and you don't, maybe we should discuss this in a different thread, to not sidetrack this one.

TBH I am not exactly sure what your purpose of bringing this up here is...
I also did not quite understand your relationship to the simulation website you posted earlier, that used this model:
http://www.animatedphysics.com/games/photon_longdistance_nonlocality.htm
If you are the author of that simulation and you want to make it useful for the current topic, maybe you can add the FTL mechanism to it (or make it optional for bonus points?) so it matches QM and reality, and also add a way to more quickly simulate a large number of trials so the end data can finally get some statistical significance.
And even then, I'm not sure now useful it would be for the current topic. It can certainly help people new to the subject in understanding the Bell violation that's happening in reality, and how a FTL effect helps in simulating/explaining that, but I think this thread is already targeted at people understanding this well enough.

There is no FTL mechanism used in this animation, it is structured in a step by step method so its path though time can be tracked, give it a try as it can generate larger photon runs now. As to why post it here, the original poster of this thread laid out some parameters to generate the matrices used in analyzing photon polarization, the animation creates these matrices and saves a lot of calculation time - try it out if you get a chance.

The animation uses 2 different methods of calculating what happens when a photon interacts with a polarizer. Both the classical and the statistical/quantum method use only the angle of their own polarizer and the photon that hits it to determine the outcome, hence the calculation does not rely on the other photons orientation or the setting of the other polarizer to determine the outcome - hence no FTL communication for either animation method.

The "statistical/quantum" imagines spin with precession.

Imagine a spinning object with precession coming straight up at you from below. You will see the object's axis either spinning left or right. If you imagine the object coming directly at you from the side, you will see the object's axis wobbling left and right with an average direction and no overall spin. This is the classic definition of a spin 1 object with 3 states: 1/ spin left, 2/ spin right or 3/ no spin at all. A photon is a spin 1 object.

The "classical" method modeled that is often used, assumes the photon has no precession and gets through a polarizer if the 2 angles are within 45 degrees and not otherwise.

• if |γ − λ| ≤ π/4 then vertical
• if |γ − λ| > 3π/4 then vertical
• horizontal otherwise.

Resulting in this picture of the probability of a photon getting though a polarizer.

For the "statistical/quantum animation", whether or not this object gets through a polarizer depends on these equations, again, with no reference to the other photon orientation or the other polarizer angle.

• Chance of vertical measurement = (cos((γ − λ)*2)+1)/2
• Chance of horizontal measurement = (cos((γ − λ + π/2)*2)+1)/2

Resulting in this picture of the probability of a photon getting though a polarizer.

Again, the important feature of the animation is that no FTL calculation or principle is used.

 

[Ken G]

Not necessarily. I don't know about that, I said "it seems to me" with the limited judgement I have available. My purpose is to have the other outlook explained to me, since the FTL one has its own problems. So once I understand the other position I can agree to disagree or embrace it.

Here's how I would describe my way of thinking about it. It involves treating the system holistically, so is like Mentz114's field approach, but it's less realist in quality, in the sense that it does not treat the reality of the system as something separate from the questions that are put to the system. Or is that more realistic-- to count the full reality? You see, in my view, a question never asked to a system, by an actual apparatus capable of answering that question, is a question that is not answered by that system. So I don't think of a system as a kind of "answer man", just waiting to be asked any question (that's standard realism, local or otherwise), I view the answers established by the apparatus as part of the system. In other words, the system is not just holistic with itself, it is also holistic with the apparatuses it is subjected to-- the separation between the system, and the apparatus that establishes the "facts" of the system, is artificial and undesirable in some cases.

In short, the reason the system violates Bell's inequality is that if different questions are put to the system, it's a different system, because the system is part of the full reality there. So all you do is relax not only the local realism we often try to attach to the parts of the system, but also the local realism we attach to the system as independent from the environment that establishes the facts of the system. If you do this, you never need anything FTL, because Bell violations are no particular problem.

 

[zonde]

Yes, I think that sums it up. There doesn't have to be anything FTL or retrocausal in entanglement, but we get those if we insist on other assumptions. It's too high a price for my taste, so I avoid assuming that correlations that respond to our choices of what to measure must be explained either by "information carried with the parts" or by "influences between the parts." The problem is treating a system like a sum of independent parts, that's what doesn't seem to work. Entanglement is our reason to drop that approach, just as we did with the indistinguishable electrons in a white dwarf.

Possible reasons to drop some approach is inconsistency or contradiction with observations.
Treating entangled particles as parts gives consistent picture. You of course can use alternative approach if you like. But make sure you can build consistent model that agrees with observations.

 

[ddd123]

In other words, the system is not just holistic with itself, it is also holistic with the apparatuses it is subjected to-- the separation between the system, and the apparatus that establishes the "facts" of the system, is artificial and undesirable in some cases.

In short, the reason the system violates Bell's inequality is that if different questions are put to the system, it's a different system, because the system is part of the full reality there. So all you do is relax not only the local realism we often try to attach to the parts of the system, but also the local realism we attach to the system as independent from the environment that establishes the facts of the system. If you do this, you never need anything FTL, because Bell violations are no particular problem.

Ok that makes sense. I wonder if this causes problems of circularity in the notion of measurement though.

 

[kith]

You see, in my view, a question never asked to a system, by an actual apparatus capable of answering that question, is a question that is not answered by that system. So I don't think of a system as a kind of "answer man", just waiting to be asked any question (that's standard realism, local or otherwise), I view the answers established by the apparatus as part of the system.

From your point of view, what is the significance of the fact that in classical mechanics, we can view the system as the "answer man"?

 

[zonde]

In short, the reason the system violates Bell's inequality is that if different questions are put to the system, it's a different system, because the system is part of the full reality there. So all you do is relax not only the local realism we often try to attach to the parts of the system, but also the local realism we attach to the system as independent from the environment that establishes the facts of the system. If you do this, you never need anything FTL, because Bell violations are no particular problem.

Relaxing local "realism" (local predeterminism) allows violation of Bell's inequalities only as much as it allows non-locality (FTL).
So what you are saying does not really make any sense.

 

[ddd123]

Relaxing local "realism" (local predeterminism) allows violation of Bell's inequalities only as much as it allows non-locality (FTL).

I took it to mean: since even the measuring instruments are part of this whole, the nonlocal changes are of the whole too; so in this sense there isn't some "moving" (which the notion of speed implies). As I said above, this to me seems like a rephrasing and it's not substantially different from just saying it's FTL. But at least it made sense to me with respect to the problem of action.

 

[DrChinese]

There is no FTL mechanism used in this animation, it is structured in a step by step method so its path though time can be tracked, give it a try as it can generate larger photon runs now. As to why post it here, the original poster of this thread laid out some parameters to generate the matrices used in analyzing photon polarization, the animation creates these matrices and saves a lot of calculation time - try it out if you get a chance.

The animation uses 2 different methods of calculating what happens when a photon interacts with a polarizer. Both the classical and the statistical/quantum method use only the angle of their own polarizer and the photon that hits it to determine the outcome, hence the calculation does not rely on the other photons orientation or the setting of the other polarizer to determine the outcome - hence no FTL communication for either animation method.

The "statistical/quantum" imagines spin with precession. ...

For the "statistical/quantum animation", whether or not this object gets through a polarizer depends on these equations, again, with no reference to the other photon orientation or the other polarizer angle.

• Chance of vertical measurement = (cos((γ − λ)*2)+1)/2
• Chance of horizontal measurement = (cos((γ − λ + π/2)*2)+1)/2

... Again, the important feature of the animation is that no FTL calculation or principle is used.

I am going to absolutely challenge everything you are saying about recreating the quantum statistics in a computer program using independent calculations. That is impossible.

I checked the stats in the referenced simulation for the quantum case by performing a run of 100. There were 28 cases where Alice and Bob's angles were selected the same, and all 28 (100%) yielded matches as expected. There were 72 cases where the selected angles were different, and 22 of those (30.55%) were matches. The local realistic max is 25% and the QM prediction is 33.33%. You can run this yourself, and I am sure as long as you do it you will see the following: 100% matches when the angle is the same, and something close to 33.33% matched when the angles for Alice and Bob are 120 degrees different.

But guess what? When the angles are the same, that is NOT the statistics you get when there is evaluation of the formulae you supplied. Surely you can see that when Alice and Bob measure at the same angle, there is an identical difference of γ − λ degrees for both Alice and Bob. That evaluates to something different than 1 or 0 - the values which would be required for a certain matching. So there would be some cases where a measurement at the same angle would yield a mismatch instead of a match. That doesn't ever happen in the simulation, not even once, for the quantum statistics.

I ran another 100 and the results were 33 at the same angles which were 100% matched; and 67 at a difference of 120 degrees which were 37.31% matched. This model produces the QM statistics and has both Alice and Bob's setting as part of the algorithm; there is NOT independent calculation for Alice and Bob's outcomes. In the rules of simulations, there is non-local influence.

Please, you are wildly off to offer a model that is so obviously refuted by Bell. This is far outside generally accepted physics, and is personal speculation that should not be presented here. You couldn't even hand pick values and make it work out. That is the "DrChinese" challenge, in fact.

 

[zonde]

I took it to mean: since even the measuring instruments are part of this whole, the nonlocal changes are of the whole too; so in this sense there isn't some "moving" (which the notion of speed implies). As I said above, this to me seems like a rephrasing and it's not substantially different from just saying it's FTL. But at least it made sense to me with respect to the problem of action.

I agree that we have to consider measuring instruments too. But if we talk about violation of Bell inequalities and non-locality it does not change anything.
If distant measurement results are determined locally then they can't violate Bell inequality.

Not sure what do you mean by "nonlocal changes" here and what is moving. Maybe you mean measurement settings as rotation of polarizers?