B How does Bell's inequality rule out realism?

[PhysicsEntanglment]

I understand Bell's inequality, and I can see how removing locality can produce the observed statistical correlations. However something that I often read is that eradicating realism can also generate the correlation observed in entanglement. I don't see how a particle not having definite properties prior to measurement will produce the correlations. Can I have someone clear this up for me?

 

[jfizzix]

When a Bell inequality is violated, one rules out that the correlations between the particles can be explained by information in their shared past (i.e., locally).

This implies either that the correlations are explained non-locally (so that there is some manner of instantaneous information transfer), or that the correlations are simply not explainable, which would deny that the measurement results can be predicted with enough information about it.

 

[DrChinese]

I understand Bell's inequality, and I can see how removing locality can produce the observed statistical correlations. However something that I often read is that eradicating realism can also generate the correlation observed in entanglement. I don't see how a particle not having definite properties prior to measurement will produce the correlations. Can I have someone clear this up for me?

Removing realism does not "generate" anything any more than removing locality does.

On the other hand: there are interpretations of QM in which things like causality are not present. Such an interpretation does not inherently violate the Bell limits. Further, they don't violate locality, which is their "selling" point.

 

[gva]

Removing realism does not "generate" anything any more than removing locality does.

On the other hand: there are interpretations of QM in which things like causality are not present. Such an interpretation does not inherently violate the Bell limits. Further, they don't violate locality, which is their "selling" point.

Isn't removing realism like saying it is the equations that are non-local and they connect the entire universe into one instantaneous whole?

 

[PhysicsEntanglement1]

Thanks for the response. I continually heard that bell's inequalities proved that either locality or realism (also referred to as counterfactual definiteness) was wrong. From my understanding now, removing realism is just another way of saying that we don't know what's happening?
In other words, probability doesn't work in the same manner in the quantum level as it does in the classical way?

Also, when you said,
"removing realism doesn't generate anything more than removing locality does" this confused me.
Removing locality, at least in the way that I've been told, permits for one entangled particle to "tell" the other entangled particle what state to be in

 

[entropy1]

(Quantum) realism, as far as I understand it, signifies the existence of quantum properties independent of whether we examine them.

How this fits into the entanglement picture remains a mystery to me...

 

[Demystifier]

I understand Bell's inequality, and I can see how removing locality can produce the observed statistical correlations. However something that I often read is that eradicating realism can also generate the correlation observed in entanglement. I don't see how a particle not having definite properties prior to measurement will produce the correlations. Can I have someone clear this up for me?

See https://arxiv.org/abs/1112.2034

 

[PhysicsEntanglement1]

See https://arxiv.org/abs/1112.2034

Thanks for the paper. I don't have time right now to read the entire thing but I did read the first page. It says, "...[some say] that nature is local but objective reality does not exist." I'm merely asking for the explanation of this. How can removing objective reality retain locality and also not violate Bell's inequality? Thank you.

 

[Simon Phoenix]

I don't see how a particle not having definite properties prior to measurement will produce the correlations.

I'm guessing here that what you're after is some kind of intuitive view for why the failure of realism helps us understand things. After all, as you said, if we have a hidden (or extra) variable model then intuitively we can see that one way to get a violation of the inequality is to have some kind of 'information' transferred between the two measurement devices; put those devices far enough apart and we would have to suppose that this 'information' gets from A to B faster than a light signal can propagate between the two devices.

So I think what you're asking is whether there is a similar kind of explanation for why the failure of realism allows the inequality to be violated (if we assume everything is hunky-dory in the locality department).

So what does the assumption of 'realism' buy us? What do we get from it? Well if you've read some of the literature you'll see that this issue generates a whole slew of metaphysical musings, often expressed in some quite sophisticated terminology. For me, the essential point is that in classical physics it is meaningful to talk of properties we might have measured - "we measured the position of the particle, but IF we'd measured the momentum we would have got this result" - kind of thinking. Just writing down the state of a classical particle with its 3 position variables and 3 momentum variables is making this assumption. This gets called 'counterfactual definiteness' but in my view this is just a shorthand for saying 'classical physics' because it's implicit in all of classical physics.

In the usual 'textbook' derivations of the BI where this assumption gets used is in the manipulation of the conditional probabilities - essentially we replace a probability representing the result of something that WAS measured with a result probability of something that MIGHT have been measured. It's a completely legitimate operation in a classical world - but somewhat less legitimate in a quantum world. In a hand-waving way we can think of the uncertainty principle preventing us from doing this in any meaningful way within a quantum perspective.

But I agree with you that it's more difficult to get a picture of why this messes things up. I go back to Feynman's path-integral paper where in the introduction he makes a very powerful point on this issue (I urge you to read it - there is no way I could match Feynman's clarity and insight). Here's his argument sort of simplified :

Let's suppose we have some object prepared in some state A. We're going to measure some property and suppose we obtain the result C. We're going to further suppose that we can 'get' to C from A via 2 intermediate states B1 and B2. Classically then we would write the conditional probability P(C|A) as
P(C|A) = P(C|B1)P(B1|A) + P(C|B2)P(B2|A)
So the classical view would tell us we can get to C either by going along B1 or B2 - we just don't know which of these paths, but we assume classically that in getting from A to C the object did indeed go via one of these paths.

But the quantum rules are different - to get the conditional probability P(C|A) if we don't know which path we must work with the amplitudes and the probability is then
P(C|A) = | a(C|B1)a(B1|A) + a(C|B2)a(B2|A) |²
where here the a are the amplitudes.

So in a nutshell if we make the assumption that the object really did go on one of these paths (we just don't know which) then we just use the classical conditional probability formula - but this gives us the wrong prediction according to QM. In other words assuming some 'real' property gets us into trouble. It's really just the 2 slit issue (and we note that the issue of commutativity vs. non-commutativity that is central here - as it is in the derivation of the BI where we effectively assume commutativity in making our 'counterfactual' argument)

I've tried here to present a bit more 'intuition' about what might be different in QM in hopefully plain terms and undoubtedly I've glossed over a lot of technical things, but I hope it's given a bit more of a feel for what the issues might be. I'm sure others will help flesh out the technicalities better than I can.

 

[Markus Hanke]

@Simon Phoenix : I like your post, very well explained :)

 

[gva]

I'm guessing here that what you're after is some kind of intuitive view for why the failure of realism helps us understand things. After all, as you said, if we have a hidden (or extra) variable model then intuitively we can see that one way to get a violation of the inequality is to have some kind of 'information' transferred between the two measurement devices; put those devices far enough apart and we would have to suppose that this 'information' gets from A to B faster than a light signal can propagate between the two devices.

So I think what you're asking is whether there is a similar kind of explanation for why the failure of realism allows the inequality to be violated (if we assume everything is hunky-dory in the locality department).

So what does the assumption of 'realism' buy us? What do we get from it? Well if you've read some of the literature you'll see that this issue generates a whole slew of metaphysical musings, often expressed in some quite sophisticated terminology. For me, the essential point is that in classical physics it is meaningful to talk of properties we might have measured - "we measured the position of the particle, but IF we'd measured the momentum we would have got this result" - kind of thinking. Just writing down the state of a classical particle with its 3 position variables and 3 momentum variables is making this assumption. This gets called 'counterfactual definiteness' but in my view this is just a shorthand for saying 'classical physics' because it's implicit in all of classical physics.

In the usual 'textbook' derivations of the BI where this assumption gets used is in the manipulation of the conditional probabilities - essentially we replace a probability representing the result of something that WAS measured with a result probability of something that MIGHT have been measured. It's a completely legitimate operation in a classical world - but somewhat less legitimate in a quantum world. In a hand-waving way we can think of the uncertainty principle preventing us from doing this in any meaningful way within a quantum perspective.

But I agree with you that it's more difficult to get a picture of why this messes things up. I go back to Feynman's path-integral paper where in the introduction he makes a very powerful point on this issue (I urge you to read it - there is no way I could match Feynman's clarity and insight). Here's his argument sort of simplified :

Let's suppose we have some object prepared in some state A. We're going to measure some property and suppose we obtain the result C. We're going to further suppose that we can 'get' to C from A via 2 intermediate states B1 and B2. Classically then we would write the conditional probability P(C|A) as
P(C|A) = P(C|B1)P(B1|A) + P(C|B2)P(B2|A)
So the classical view would tell us we can get to C either by going along B1 or B2 - we just don't know which of these paths, but we assume classically that in getting from A to C the object did indeed go via one of these paths.

But the quantum rules are different - to get the conditional probability P(C|A) if we don't know which path we must work with the amplitudes and the probability is then
P(C|A) = | a(C|B1)a(B1|A) + a(C|B2)a(B2|A) |²
where here the a are the amplitudes.

So in a nutshell if we make the assumption that the object really did go on one of these paths (we just don't know which) then we just use the classical conditional probability formula - but this gives us the wrong prediction according to QM. In other words assuming some 'real' property gets us into trouble. It's really just the 2 slit issue (and we note that the issue of commutativity vs. non-commutativity that is central here - as it is in the derivation of the BI where we effectively assume commutativity in making our 'counterfactual' argument)

I've tried here to present a bit more 'intuition' about what might be different in QM in hopefully plain terms and undoubtedly I've glossed over a lot of technical things, but I hope it's given a bit more of a feel for what the issues might be. I'm sure others will help flesh out the technicalities better than I can.

Reading this makes me realized that the realism is about particles only.. there is still the field even if no particles.. so in the double slit experiments.. even if there are no solid particles.. there is still detection.. sometimes anti-realism feels like there would be no detection and just nothingness which is not the case at all.

 

[gva]

Reading this makes me realized that the realism is about particles only.. there is still the field even if no particles.. so in the double slit experiments.. even if there are no solid particles.. there is still detection.. sometimes anti-realism feels like there would be no detection and just nothingness which is not the case at all.

But for Aspect experiment, the field has to be instantaneous billions of light years away... hope Simon can explain what occurs in this scenario.

 

[PhysicsEntanglement1]

@Simon Phoenix
Thanks for the post. That was really insightful. If I understand you correctly, if we remove realism, then we don't have to go through some state to reach a final outcome. In your example you used B1 and B2 as the states between A and C. This produces a separate way of approaching probability in quantum mechanics than in the classical manner. If we approach it in this way, does it result in the probability observed through entanglement? Also, in the real world, what is the manifestation of states A and C?

 

[Simon Phoenix]

Reading this makes me realized that the realism is about particles only..

I'm not sure I would agree with that gva. If you look at one of the clearest papers ever written on the Bell Inequalities by the master himself (John Bell)

https://cds.cern.ch/record/142461/files/198009299.pdf

Then you'll see mid-way through that paper there is a very general model of the experimental situation that is used to derive the inequality. We have 2 measurement devices that measure 'something' and the results of the measurement are yes/no (or 1/0 or up/down - whichever takes your fancy). We can also adjust the setting on the measurement devices. If we let A and B stand for the measurement results and a and b stand for the measurement settings of the respective devices then many runs of the experiment will allow us to measure P(A,B | a,b).

The idea is to explain any correlation between the results by some extra variables - something is assumed to be happening behind the scenes, so to speak, that gives rise to the correlations. So, it is assumed, what we actually have is P(A,B | a,b, {hv}) where hv is this collection of 'hidden' variables.- which if we only knew them we could use to explain the measured statistics. Bell takes very great care to stress that no assumption is made about the nature of these hidden variables - or even the thing we're measuring - so we could have particles, fields or something hybrid entity. The hidden variables could be a collection of discrete things or functions - or even wavefunctions! No model of physics is assumed - in particular no quantum assumptions are used.

If we then make some (on the surface) very reasonable assumptions about the properties of those variables then it can be shown that there is a constraint on the correlations. Since these are things we can directly measure we can test this if we can find a physical system that gives us these yes/no answers that might have correlations.

These 'reasonable' assumptions are the realism and locality conditions which might be loosely expressed as :
- things have properties independent of measurement
- statistics 'here' are not influenced by settings 'there'

Using these assumptions we get the constraint on the correlations that have to be satisfied by probabilities of the form P(A,B | a,b, {hv}) - and this is the celebrated Bell inequality.

It is important to note that absolutely no assumption is made about particles or fields or the nature of the extra variables (other than these 2 reasonable assumptions). I have heard it argued that Bell makes the implicit assumption that we have particles - but I don't understand this myself since all we're connecting is measurement results using pretty straightforward probability models. The detectors just go 'ping' or 'ding' - and we just model the statistics of the pings and dings using good old probability functions. What causes the pings or dings is of earth-shattering irrelevancy.

There is an implicit assumption - and that is that it is possible to make truly random choices of the measurement device settings.

 

[gva]

I'm not sure I would agree with that gva. If you look at one of the clearest papers ever written on the Bell Inequalities by the master himself (John Bell)

https://cds.cern.ch/record/142461/files/198009299.pdf

Then you'll see mid-way through that paper there is a very general model of the experimental situation that is used to derive the inequality. We have 2 measurement devices that measure 'something' and the results of the measurement are yes/no (or 1/0 or up/down - whichever takes your fancy). We can also adjust the setting on the measurement devices. If we let A and B stand for the measurement results and a and b stand for the measurement settings of the respective devices then many runs of the experiment will allow us to measure P(A,B | a,b).

The idea is to explain any correlation between the results by some extra variables - something is assumed to be happening behind the scenes, so to speak, that gives rise to the correlations. So, it is assumed, what we actually have is P(A,B | a,b, {hv}) where hv is this collection of 'hidden' variables.- which if we only knew them we could use to explain the measured statistics. Bell takes very great care to stress that no assumption is made about the nature of these hidden variables - or even the thing we're measuring - so we could have particles, fields or something hybrid entity. The hidden variables could be a collection of discrete things or functions - or even wavefunctions! No model of physics is assumed - in particular no quantum assumptions are used.

If we then make some (on the surface) very reasonable assumptions about the properties of those variables then it can be shown that there is a constraint on the correlations. Since these are things we can directly measure we can test this if we can find a physical system that gives us these yes/no answers that might have correlations.

These 'reasonable' assumptions are the realism and locality conditions which might be loosely expressed as :
- things have properties independent of measurement
- statistics 'here' are not influenced by settings 'there'

Using these assumptions we get the constraint on the correlations that have to be satisfied by probabilities of the form P(A,B | a,b, {hv}) - and this is the celebrated Bell inequality.

It is important to note that absolutely no assumption is made about particles or fields or the nature of the extra variables (other than these 2 reasonable assumptions). I have heard it argued that Bell makes the implicit assumption that we have particles - but I don't understand this myself since all we're connecting is measurement results using pretty straightforward probability models. The detectors just go 'ping' or 'ding' - and we just model the statistics of the pings and dings using good old probability functions. What causes the pings or dings is of earth-shattering irrelevancy.

There is an implicit assumption - and that is that it is possible to make truly random choices of the measurement device settings.

Thanks for this very clear presentation.. the violations of Bell's Inequality means either realism or locality is wrong. If we assumed realism was wrong, then
- things have no properties independent of measurement

but the following can't be right

- statistics 'here' are not influenced by settings 'there'

I think the OP was saying both had to be wrong.. that is if realism was wrong, it didn't mean locality was right.. it's because the correlations can only occur if there are "non-local" influence. Or was it about semantics where locality is defined as the particles having objective properties and "non-locality" means they are communicating by signals? But non-locality can also mean the equations are non-local and causing the correlations. Are we being constraint to use the precise meaning of locality and limit the possibilities (trying to push us in one thought only)?

 

[Simon Phoenix]

that is if realism was wrong, it didn't mean locality was right.. it's because the correlations can only occur if there are "non-local" influence

Well if we derive something using assumption X and assumption Y - and then we do an experiment and find our prediction doesn't match, then it means that at least one (and maybe both) those assumptions can't be correct.

In the case of QM it is possible to set up a 'realistic' theory that reproduces the predictions of QM - but this theory (de Broglie/Bohm theory) is non-local in that the results of experiments in your lab depend on the configuration of everything in the universe, at least as I understand it. So dBB theory keeps 'realism' at the cost of discarding locality. In dBB theory we have a real particle with real properties (as I understand it) that is 'guided' by some mysterious complex-valued potential associated with it. The potential is constructed so that it agrees with the results obtained from the Schrodinger equation. I'm not a big fan of the dBB approach myself (although John Bell did seem to be) but it does give the same answers as standard QM as far as I can tell.

Locality is one of those words that get used in different ways. In the context of Bell's inequality I like to think of it in the following way.

Suppose we set up an experiment in lab A and another experiment in lab B. If the guy in lab B can set his device to measure x or y, then the results obtained in lab A should not depend on the choice of the setting x or y. This is how I define locality in the context of the BI.

I think we would live in a very strange world indeed if that were not true - but some people do prefer some kind of non-local influence as an explanation of nature, rather than dispense with 'realism'. Dispensing with 'realism' is admittedly a bit odd too :-)

If we define 'locality' in the above fashion then (standard) QM is a fully local theory. Results of measurements in (standard) QM do not depend of the settings of remote devices.

 

[Nugatory]

I think the OP was saying both had to be wrong.. that is if realism was wrong, it didn't mean locality was right.. it's because the correlations can only occur if there are "non-local" influence. Or was it about semantics where locality is defined as the particles having objective properties and "non-locality" means they are communicating by signals? But non-locality can also mean the equations are non-local and causing the correlations. Are we being constraint to use the precise meaning of locality and limit the possibilities (trying to push us in one thought only)?

You are trying to find more in Bell's theorem than is there. Bell starts with two assumptions (which are sometimes called "locality" and "realism", but are more precisely and mathematically stated in the proof) and derive an inequality. Experments show that this inequality is violated, so we can conclude that one or both of the assumptions is invalid.

But that's all that we can ever learn from the theorem - it cannot tell us which of the assumptions are wrong. To do that we'd need to propose and experimentally confirm a theory of what is going on, and all that we'll ever get from Bell's theorem is a constraint on what properties that theory might have.

 

[gva]

Well if we derive something using assumption X and assumption Y - and then we do an experiment and find our prediction doesn't match, then it means that at least one (and maybe both) those assumptions can't be correct.

In the case of QM it is possible to set up a 'realistic' theory that reproduces the predictions of QM - but this theory (de Broglie/Bohm theory) is non-local in that the results of experiments in your lab depend on the configuration of everything in the universe, at least as I understand it. So dBB theory keeps 'realism' at the cost of discarding locality. In dBB theory we have a real particle with real properties (as I understand it) that is 'guided' by some mysterious complex-valued potential associated with it. The potential is constructed so that it agrees with the results obtained from the Schrodinger equation. I'm not a big fan of the dBB approach myself (although John Bell did seem to be) but it does give the same answers as standard QM as far as I can tell.

Locality is one of those words that get used in different ways. In the context of Bell's inequality I like to think of it in the following way.

Suppose we set up an experiment in lab A and another experiment in lab B. If the guy in lab B can set his device to measure x or y, then the results obtained in lab A should not depend on the choice of the setting x or y. This is how I define locality in the context of the BI.

I think we would live in a very strange world indeed if that were not true - but some people do prefer some kind of non-local influence as an explanation of nature, rather than dispense with 'realism'. Dispensing with 'realism' is admittedly a bit odd too :-)

If we define 'locality' in the above fashion then (standard) QM is a fully local theory. Results of measurements in (standard) QM do not depend of the settings of remote devices.

Results of measurements in (standard) QM do not depend of the settings of remote devices? But since Bell's Inequality is violated, then shouldn't it mean "Results of measurements in (standard) QM DEPEND of the settings of remote devices"? Or does anti-realism mean there is no remote devices in the first place? But standard QM is about anti-realism (no counterfactual definiteness). What is your idea about dispensing with realism? Hope someone can standardize what is the case about settings of remote devices and locality, realism label because different people have different semantics and some of us may be quite confused. Thank you.

 

[DrChinese]

Thanks for the paper. I don't have time right now to read the entire thing but I did read the first page. It says, "...[some say] that nature is local but objective reality does not exist." I'm merely asking for the explanation of this. How can removing objective reality retain locality and also not violate Bell's inequality? Thank you.

I can give you some examples of non-realistic interpretations. There are several classes of such. First, keep in mind that the opposite of objective realism is subjective realism. This is realism which is observer dependent. In other words, the observer is a part of the overall measurement context. In the classical world, there is objective realism. But not so in the quantum world. The observer's choice of measurement basis is relevant to the correlated outcomes.

A. Many Worlds Interpretation (MWI) is non-realistic. Any measurement causes a "splitting" of worlds. This is a fairly popular interpretation.
B. Time symmetric / retrocausal / acasual interpretations are non-realistic. There is not strict causality. In other words, the future and the past are part of an overall context that describes outcome probabilities. While this seems strange and counterintuitive, there are a number of experiments for which this is actually a very apt descriptive interpretation.

Don't ask me HOW any of the above work mechanically; I have no clue and I doubt anyone else does either.

 

[DrChinese]

Results of measurements in (standard) QM do not depend of the settings of remote devices? But since Bell's Inequality is violated, then shouldn't it mean "Results of measurements in (standard) QM DEPEND of the settings of remote devices"?

There is a lot of semantics involved. In entangled state statistics, generally: Predictions for correlated results of measurements in (standard) QM DEPEND of the settings of remote devices (in the sense that the measurement context is not limited to a local region).

 

[Demystifier]

Thanks for the paper. I don't have time right now to read the entire thing but I did read the first page. It says, "...[some say] that nature is local but objective reality does not exist." I'm merely asking for the explanation of this. How can removing objective reality retain locality and also not violate Bell's inequality? Thank you.

Suppose two measurements A and B, which are spatially separated, are correlated. The correlation at spatial separation implies nonlocality. However, you can never observe the nonlocal correlation. One observer cannot see both measurement results at once, as long as they are spatially separated. The best you can is to remember the result obtained at A, then travel to B to see the result there, and then compare the results. In this way you see a correlation, but not a nonlocal correlation. The correlation is nonlocal only if you assume that measurement outcomes existed even before you observed and compared them. This assumption is the assumption of reality. If this assumption is not fulfilled, then nonlocality is avoided.

 

[entropy1]

I agree with you @Demystifier. However, if we assume Alice and Bob measuring at A and B, would in case of non-realism Alice not have existed for Bob, and Bob not for Alice? This would be asymmetrical as well as a little absurd in my eyes.

The situation would become symmetrical if Alice would simultaneously exist and not exist (in a certain state), and similarly for Bob. In my eyes this suggests superposition of Alice and likewise for Bob. One would arbitrarily put collapse at the end, the act of observing what you want to observe, i.e. the correlation calculation?

 

[gva]

Suppose two measurements A and B, which are spatially separated, are correlated. The correlation at spatial separation implies nonlocality. However, you can never observe the nonlocal correlation. One observer cannot see both measurement results at once, as long as they are spatially separated. The best you can is to remember the result obtained at A, then travel to B to see the result there, and then compare the results. In this way you see a correlation, but not a nonlocal correlation. The correlation is nonlocal only if you assume that measurement outcomes existed even before you observed and compared them. This assumption is the assumption of reality. If this assumption is not fulfilled, then nonlocality is avoided.

Police or FBI Forensics are so successful because they deal with logic. It seems Bell's Theorem tried to make one bypass logic altogether. In Forensics. If there are bullets. Guns are fired. But in quantum entanglement, correlations are observed but doubts were cast whether there were non-local correlations. Well even if the particles don't exist before measurements. Their connections (whatever it is before particles were precipitated) is non-local. Isn't it the correlations can be compared in paper and there are indeed effects.. so of course there should be non-local connections even if it's not the particles communicating at each end. Why do people even doubt this?

 

[entropy1]

so of course there should be non-local connections even if it's not the particles communicating at each end. Why do people even doubt this?

There is no way to prove that there is 'communication', as you call it. The correlation is quantized, and Alice nor Bob can influence any instance of a quantum property measured. However, it seems appealing to believe in some sort of 'communication', for else we have to give up realism. And that, seems to me, remains one of the main paradoxes in QM, in my view.

 

[PhysicsEntanglement1]

Suppose two measurements A and B, which are spatially separated, are correlated. The correlation at spatial separation implies nonlocality. However, you can never observe the nonlocal correlation. One observer cannot see both measurement results at once, as long as they are spatially separated. The best you can is to remember the result obtained at A, then travel to B to see the result there, and then compare the results. In this way you see a correlation, but not a nonlocal correlation. The correlation is nonlocal only if you assume that measurement outcomes existed even before you observed and compared them. This assumption is the assumption of reality. If this assumption is not fulfilled, then nonlocality is avoided.

Thank you. This was the explanation I was looking for.
I do have a follow up question, however. It seems that removing objective reality to not violate bell's inequality would not produce the correlation that is consistently viewed in each and every experiment. Also, the way in which we conduct the experiment also changes the results. It is the square of the cosine of the angle between the 2 polarizer settings. This would seem to indicate that there is at least some communication.

 

[entropy1]

It is the square of the cosine of the angle between the 2 polarizer settings.

However, this is symmetric; If we take the basis for one of the polarizers to be 0°, the other becomes cos²(α). There is no directionality.

 

[gva]

There is no way to prove that there is 'communication', as you call it. The correlation is quantized, and Alice nor Bob can influence any instance of a quantum property measured. However, it seems appealing to believe in some sort of 'communication', for else we have to give up realism. And that, seems to me, remains one of the main paradoxes in QM, in my view.

Why, is there no communication if we give up realism? But the spacetime is communicating to produce the correlations, even if there are no particles at either side before observation. Is this not called communication? What else is it called (when spacetime is communicating)? Any mentors give us the authoritative statements?

 

[Simon Phoenix]

"Results of measurements in (standard) QM DEPEND of the settings of remote devices"?

Well let's think of the classic example of a pair of spin-1/2 particles that are maximally entangled. So up to a normalization constant we've got a state that is the form |0>|1> + |1>|0>.

Now let's imagine the following scenario. Bob is going to prepare 2 different ensembles - so basically 2 boxes.
box 1 : contains N spin-1/2 particles each taken from a maximally correlated pair as above
box 2 : contains N spin-1/2 particles where for each particle a coin toss has determined whether it's a spin up or spin down (let's say in the z-basis)

Bob is going to randomly choose which of the 2 boxes to give to Alice. Alice is then allowed to perform ANY experiment whatsoever on the particles in this box she's been given. Her job is to figure out whether she's been given box 1 or box 2 with a probability that's better than guessing.

There's no experiment she can perform, on the particles in the box alone, that will enable her to distinguish which box she's been given any better than guessing.

So if from her particles alone she cannot tell whether she has one of a pair of entangled spins, or simply a randomly chosen up or down particle - how then can we say that in this example the results of (local)* experiments depend on the settings of remote devices in QM?

*by 'local' here I mean in Alice's lab - that is, 'local' in the same kind of sense that I would say "my local supermarket", for example.

 

[gva]

Well let's think of the classic example of a pair of spin-1/2 particles that are maximally entangled. So up to a normalization constant we've got a state that is the form |0>|1> + |1>|0>.

Now let's imagine the following scenario. Bob is going to prepare 2 different ensembles - so basically 2 boxes.
box 1 : contains N spin-1/2 particles each taken from a maximally correlated pair as above
box 2 : contains N spin-1/2 particles where for each particle a coin toss has determined whether it's a spin up or spin down (let's say in the z-basis)

Bob is going to randomly choose which of the 2 boxes to give to Alice. Alice is then allowed to perform ANY experiment whatsoever on the particles in this box she's been given. Her job is to figure out whether she's been given box 1 or box 2 with a probability that's better than guessing.

There's no experiment she can perform, on the particles in the box alone, that will enable her to distinguish which box she's been given any better than guessing.

So if from her particles alone she cannot tell whether she has one of a pair of entangled spins, or simply a randomly chosen up or down particle - how then can we say that in this example the results of (local)* experiments depend on the settings of remote devices in QM?

*by 'local' here I mean in Alice's lab - that is, 'local' in the same kind of sense that I would say "my local supermarket", for example.

Simple. There are correlations in the results later when compared. It's like the airplane flying into the World Trade Center.. the passengers have no way to distinguish if the actions depend on the settings of remote devices (such as radio) in Saudi Arabia or Libya.. but later the correlations can be found out. So even if there are no particles before measurements.. spacetime is communicating the correlations. So there is always communications whether there is realism or not.. unless one defines communication as only occurring between particles.. what term do you use then for spacetime that is "communicating"?

 

[Simon Phoenix]

Simple. There are correlations in the results later when compared.

But in this case we're not having Alice determine things from her experiments alone. This is equivalent to giving Alice 2 boxes - say box pair A and box pair B.

pair A : box 1 contains spin-1/2 particles and box 2 contains their corresponding entangled partners
pair B : box 1 contains spin-1/2 particles prepared in their randomly chosen states, box 2 contains spin-1/2 particles prepared in their randomly chosen states (no entanglement at all).

Now, of course, by measuring the properties of BOTH boxes in a pair Alice can determine whether she has pair A or pair B.

But this is not the same as the situation I outlined in my first post where Alice is assumed to have access to only one box.

Here's another one to think about :

Bob is going to try to deceive Alice again.
He takes spin-1/2 particles and randomly prepares them up or down states (randomly chosen) in the spin bases spin-0, spin-60, and spin-120 (angles in degrees relative to some chosen direction). He now gives the particles to Alice and tells her that what he has given her is a collection of particles for which he has the entangled partners. Can Bob deceive her? Can she tell Bob is lying?

No she can't - even if Bob gives her the list of states and tells Alice these were his measurement results - Alice can't tell the difference between what Bob has actually done and whether Bob's (and her) results have actually come from true entanglement.

 

[morrobay]

Thank you. This was the explanation I was looking for.
I do have a follow up question, however. It seems that removing objective reality to not violate bell's inequality would not produce the correlation that is consistently viewed in each and every experiment. Also, the way in which we conduct the experiment also changes the results. It is the square of the cosine of the angle between the 2 polarizer settings. This would seem to indicate that there is at least some communication.

Regarding the first part of the follow up question above, with slight modification :
"It seems that removing objective reality [to account for Bell inequality violations]
would not produce the correlation that is consistently viewed in each and every experiment".

 

[Demystifier]

I agree with you @Demystifier. However, if we assume Alice and Bob measuring at A and B, would in case of non-realism Alice not have existed for Bob, and Bob not for Alice? This would be asymmetrical as well as a little absurd in my eyes.

Yes, non-reality is quite absurd. Yet, it is a logical possibility. For more details see the paper I linked a few posts above.

 

[Demystifier]

Police or FBI Forensics are so successful because they deal with logic. It seems Bell's Theorem tried to make one bypass logic altogether. In Forensics. If there are bullets. Guns are fired. But in quantum entanglement, correlations are observed but doubts were cast whether there were non-local correlations. Well even if the particles don't exist before measurements. Their connections (whatever it is before particles were precipitated) is non-local. Isn't it the correlations can be compared in paper and there are indeed effects.. so of course there should be non-local connections even if it's not the particles communicating at each end. Why do people even doubt this?

Because some physicists desperately want to save locality. I don't know why.

 

[Demystifier]

Thank you. This was the explanation I was looking for.
I do have a follow up question, however. It seems that removing objective reality to not violate bell's inequality would not produce the correlation that is consistently viewed in each and every experiment. Also, the way in which we conduct the experiment also changes the results. It is the square of the cosine of the angle between the 2 polarizer settings. This would seem to indicate that there is at least some communication.

There is no communication in a human friendly sense. See the Appendix in the paper I linked a few posts above.

 

[gva]

Because some physicists desperately want to save locality. I don't know why.

If there were no particles before measurements.. and it is spacetime that is communicating.. is it not correct to call it non-local too? "locality" has more to do with "spacetime" than matter.. so if spacetime is communicating without particles.. it should be "non-local" too.. why is it not?

 

[rubi]

Spacetime isn't communicating. What is going wrong is the following:

Intuitively, we would think that objects carry around with them some numbers $\lambda_1,\lambda_2,\ldots$ that characterize all their properties. For example, their spin in $z$ direction may be calculated from these numbers by a function $S_z(\lambda_1,\lambda_2,\ldots)$. You can prove that this naive idea is incompatible with quantum mechanics. However, our intuition for statistics derives from this idea. Only objects, whose spin can be calculated in such a way are forced to obey the statistics that we would find intuitive.

As I said, the previous naive idea is incompatible with QM. However, there are ways to fix it:
1. Give up the idea that objects have spin. The notion of spin changes depending on the situation. There are many observables $S_z^{(1)}(\lambda_1,\lambda_2,\ldots), S_z^{(2)}(\lambda_1,\lambda_2,\ldots), \ldots$ and you need to use a different one depending on the context. This is what hidden variable theories do.
2. Give up the idea that there is a (potentially transfinite) list of numbers $\lambda_1, \lambda_2, \ldots$ that characterizes the situation of the physics. Particles still have spin, but the naive way to describe it mathematically fails. This may produce statistics that is incompatible with our intuition, since the statistics that we would find intuitive is derived from the idea that there are numbers $\lambda_1,\lambda_2,\ldots$ that describe the situation completely.

The first way gives up locality. The second way gives up classicality. Some people call it "realism", but I don't think this is a good name. The universe is still real, but its mathematical description is less naive.

The reason for why we find quantum statistics non-intuitive is that in our daily life, we are used to statistics that can be derived from some underlying list of numbers. However, it is only a problem with our intuition and not with physics itself.

 

[atyy]

Spacetime isn't communicating. What is going wrong is the following:

Intuitively, we would think that objects carry around with them some sombers $\lambda_1,\lambda_2,\ldots$ that characterize all their properties. For example, their spin in $z$ direction may be calculated from these numbers by a function $S_z(\lambda_1,\lambda_2,\ldots)$. You can prove that this naive idea is incompatible with quantum mechanics. However, our intuition for statistics derives from this idea. Only objects, whose spin can be calculated in such a way are forced to obey the statistics that we would find intuitive.

As I said, the previous naive idea is incompatible with QM. However, there are ways to fix it:
1. Give up the idea that objects have spin. The notion of spin changes depending on the situation. There are many observables $S_z^{(1)}(\lambda_1,\lambda_2,\ldots), S_z^{(2)}(\lambda_1,\lambda_2,\ldots), \ldots$ and you need to use a different one depending on the context. This is what hidden variable theories do.
2. Give up the idea that the the idea that there is a (potentially transfinite) list of numbers $\lambda_1, \lambda_2, \ldots$ that characterizes the situation of the physics. Particles still have spin, but the naive way to describe it mathematically fails. This may produce statistics that is incompatible with our intuition, since the statistics that we would find intuitive is derived from the idea that there are numbers $\lambda_1,\lambda_2,\ldots$ that describe the situation completely.

The first way gives up locality. The second way gives up classicality. Some people call it "realism", but I don't think this is a good name. The universe is still real, but its mathematical description is less naive.

The reason for why we find quantum statistics non-intuitive is that in our daily life, we are used to statistics that can be derived from some underlying list of numbers. However, it is only a problem with our intuition and not with physics itself.

Just to be clear, are you working within consistent histories?

 

[rubi]

Just to be clear, are you working within consistent histories?

No, this is essentially independent of the interpretation. The boundary is between hidden variable theories that describe the situation completely by a list of numbers $\lambda_1,\lambda_2,\ldots$ and theories that describe the situation using different mathematics.

 

[gva]

Spacetime isn't communicating. What is going wrong is the following:

Intuitively, we would think that objects carry around with them some numbers $\lambda_1,\lambda_2,\ldots$ that characterize all their properties. For example, their spin in $z$ direction may be calculated from these numbers by a function $S_z(\lambda_1,\lambda_2,\ldots)$. You can prove that this naive idea is incompatible with quantum mechanics. However, our intuition for statistics derives from this idea. Only objects, whose spin can be calculated in such a way are forced to obey the statistics that we would find intuitive.

As I said, the previous naive idea is incompatible with QM. However, there are ways to fix it:
1. Give up the idea that objects have spin. The notion of spin changes depending on the situation. There are many observables $S_z^{(1)}(\lambda_1,\lambda_2,\ldots), S_z^{(2)}(\lambda_1,\lambda_2,\ldots), \ldots$ and you need to use a different one depending on the context. This is what hidden variable theories do.
2. Give up the idea that there is a (potentially transfinite) list of numbers $\lambda_1, \lambda_2, \ldots$ that characterizes the situation of the physics. Particles still have spin, but the naive way to describe it mathematically fails. This may produce statistics that is incompatible with our intuition, since the statistics that we would find intuitive is derived from the idea that there are numbers $\lambda_1,\lambda_2,\ldots$ that describe the situation completely.

The first way gives up locality. The second way gives up classicality. Some people call it "realism", but I don't think this is a good name. The universe is still real, but its mathematical description is less naive.

The reason for why we find quantum statistics non-intuitive is that in our daily life, we are used to statistics that can be derived from some underlying list of numbers. However, it is only a problem with our intuition and not with physics itself.

We are told that spacetime and quantum is only relevant below the Planck scale, and above it we have effective field theory and QFT (or quantum mechanics) is enough. But is our failure to understand what goes on in entanglement is because we still need to understand what goes on above the Planck scale, like how exactly is matter connected to geometry? Perhaps this would give us insight into entanglement? or totally no connection and how come?

 

[Nugatory]

But is our failure to understand what goes on in entanglement is because we still need to understand ...

It's unlikely. But without a candidate theory we'd just be speculating idly (something not allowed under the PF rules).

 

[stevendaryl]

The reason for why we find quantum statistics non-intuitive is that in our daily life, we are used to statistics that can be derived from some underlying list of numbers. However, it is only a problem with our intuition and not with physics itself.

I disagree with that diagnosis. To me, what makes quantum probability seem so weird is that the probabilities are contingent on performing a particular measurement. That would make sense in terms of a local interaction. If I have a glass bottle and I'm wondering how many pieces it would split into if I smashed with a hammer, I can only find out by actually smashing it. But in an quantum entangled system, it seems that my local measurement here affects the way I describe the situation at a far distant place. In an spin-1/2 anti-correlated twin pair EPR-type experiment, Alice's description of Bob's probabilities goes from: "For any orientation, the probability of his measuring spin-up is 1/2" to "His probability of measuring spin-up is sin^2(\frac{\theta}{2}) where \theta is the angle between Bob's orientation and Alice's measured spin."

The mystery is how Alice's measurement can affect her description of Bob's probabilities, unless either (1) there is a nonlocal effect of Alice's measurement on Bob, or (2) Bob's particle's was already in the "collapsed" state, and Alice's measurement just informed her of this. Neither possibility is plausibly true, but they seem to be the only two possibilities.

 

[DrChinese]

Yes, non-reality is quite absurd.

In this case, it's all in the eye of the beholder as to what is reasonable and what is absurd...

I think in many cases, it is more useful for "non-reality" to be replaced by "contextual reality" or "subjective reality" or "acausal reality". Those are probably less of a turn-off.

 

[atyy]

No, this is essentially independent of the interpretation. The boundary is between hidden variable theories that describe the situation completely by a list of numbers $\lambda_1,\lambda_2,\ldots$ and theories that describe the situation using different mathematics.

But this definition of reality doesn't seem to have much to say about whether the moon is there when one is not looking at it.

 

[rubi]

We are told that spacetime and quantum is only relevant below the Planck scale, and above it we have effective field theory and QFT (or quantum mechanics) is enough. But is our failure to understand what goes on in entanglement is because we still need to understand what goes on above the Planck scale, like how exactly is matter connected to geometry? Perhaps this would give us insight into entanglement? or totally no connection and how come?

I don't think there is a connection. Quantum gravity is rather going to make the situation even more complicated, since the causal structure of spacetime is a consequence of gravitational physics and quantum gravity is a more complicated theory than classical gravity.

The mystery is how Alice's measurement can affect her description of Bob's probabilities, unless either (1) there is a nonlocal effect of Alice's measurement on Bob, or (2) Bob's particle's was already in the "collapsed" state, and Alice's measurement just informed her of this. Neither possibility is plausibly true, but they seem to be the only two possibilities.

I don't think these are the only possibilities. They are the only possibilities if we insist on an underlying description of the universe by a list $\lambda_1, \ldots$ of numbers, but if we give that up, I can't think of an argument that would enforce these to be the only possibilities. I believe that Bob would have gotten the same result if Alice hadn't performed her measurement. But that doesn't mean that there must be numbers $\lambda_1,\ldots$ such that Bob's measurement can be computed from them by a function $S(\lambda_1,\ldots)$. At least I don't see how this logically follows. And if this doesn't follow logically, I don't expect the statistics to be consistent with the statistics that is implied by such an assumption.

But this definition of reality doesn't seem to have much to say about whether the moon is there when one is not looking at it.

Well, quantum theory just doesn't have an opinion on whether the moon is there when nobody is looking or not. That doesn't mean that the moon isn't there. The theory is just silent on this issue. I think physical theories in general can't answer such kind of questions.

My definition of reality is "that which I experience". And quantum theory adequately describes that which I experience. Hence, it adequately describes reality. I can't think of any other useful definition of "reality".

 

[gva]

Well let's think of the classic example of a pair of spin-1/2 particles that are maximally entangled. So up to a normalization constant we've got a state that is the form |0>|1> + |1>|0>.

Now let's imagine the following scenario. Bob is going to prepare 2 different ensembles - so basically 2 boxes.
box 1 : contains N spin-1/2 particles each taken from a maximally correlated pair as above
box 2 : contains N spin-1/2 particles where for each particle a coin toss has determined whether it's a spin up or spin down (let's say in the z-basis)

Bob is going to randomly choose which of the 2 boxes to give to Alice. Alice is then allowed to perform ANY experiment whatsoever on the particles in this box she's been given. Her job is to figure out whether she's been given box 1 or box 2 with a probability that's better than guessing.

There's no experiment she can perform, on the particles in the box alone, that will enable her to distinguish which box she's been given any better than guessing.

So if from her particles alone she cannot tell whether she has one of a pair of entangled spins, or simply a randomly chosen up or down particle - how then can we say that in this example the results of (local)* experiments depend on the settings of remote devices in QM?

*by 'local' here I mean in Alice's lab - that is, 'local' in the same kind of sense that I would say "my local supermarket", for example.

What other situation in life that a sentence can mean many things and people understood it differently depending on one's experience. That following sentence is one such:

"Results of measurements in (standard) QM DEPEND of the settings of remote devices"

It can be answered Yes or No. No. Because Alice doesn't know the settings of remote devices.. Yes.. because there are correlations later on. Simon was using the context of the former.. but it also be answered in the affirmative.. correct Science Advisors?

 

[atyy]

Well, quantum theory just doesn't have an opinion on whether the moon is there when nobody is looking or not. That doesn't mean that the moon isn't there. The theory is just silent on this issue. I think physical theories in general can't answer such kind of questions.

My definition of reality is "that which I experience". And quantum theory adequately describes that which I experience. Hence, it adequately describes reality. I can't think of any other useful definition of "reality".

Sure, but then I don't think you should make out that Ilja or the rest of us realists are not understanding something profound. As he and most of us say - quantum theory in the minimal interpretation is not a theory about fundamental reality. We all understand the minimal interpretation well. In fact, we may understand it better than those who claim to use it. If it is only important that quantum theory makes accurate predictions, then what is so important about a technical point that in some obscure sense one is still able to say that quantum theory is able to "explain" the correlations and preserve "locality"?

It's pretty much the same as vanhees71's confusion over collapse. If he is truly using the minimal interpretation, then why should he care about collapse, since it isn't necessarily real?

Realists do not deny the minimal interpretation - we in fact support it. But we also say that the measurement problem points to new physics (or new conceptions of reality, eg. MWI)

 

[rubi]

Sure, but then I don't think you should make out that Ilja or the rest of us realists are not understanding something profound.

I don't remember saying this. However, it is certainly true for Ilja.

As he and most of us say - quantum theory in the minimal interpretation is not a theory about fundamental reality.

I disagree. It's a theory about fundamental reality as much as classical physics is. Classical physics also doesn't tell us whether the moon is there when nobody is looking at it. Such questions are just completely inaccessible to physics or even pure reason. It's like asking about the existence of god.

There is a difference between reality and the mathematical description of reality. Physical theories can only ever describe reality. The belief in the existence of reality is independent of physics. I am a theorist and I am only interested in the mathematical desciption. I'm interested in whether the mathematical reasoning that leads to certain conclusions about the theory is watertight. All I'm telling you is that the claims of some people (such as Ilja) that quantum theory must be non-local are false. That's just a hard mathematical fact. Nobody can prove that quantum theory is non-local. One can only get this idea by being sloppy about the mathematical arguments and this is unacceptable in science.

If it is only important that quantum theory makes accurate predictions, then what is so important about a technical point that in some obscure sense one is still able to say that quantum theory is able to "explain" the correlations and preserve "locality"?

I don't find the explanation obscure at all. If you find it obscure, it means that you cannot let go the idea that reality can be described by a list $\lambda_1,\lambda_2,\ldots$ of numbers and the consequences that follow from it. Someone who doesn't adhere to this idea should have no problem with explanations of correlations and locality. And if there is no problem, then there is also no necessity to give them up.

It's pretty much the same as vanhees71's confusion over collapse. If he is truly using the minimal interpretation, then why should he care about collapse, since it isn't necessarily real?

It's an interesting mathematical question whether the collapse postulate can be deduced from the other postulates (at least in some limit) by including more details about the evolution of the environment. This already justifies investigations. If the collapse is emergent from the other axioms, then the mathematical structure of quantum theory simplifies, which is beneficial to other parts of physics.

Realists do not deny the minimal interpretation - we in fact support it. But we also say that the measurement problem points to new physics (or new conceptions of reality, eg. MWI)

I also say that it points to new conceptions of reality. If we are interested in a local theory, we have to give up the idea that reality can be described by a set of numbers $\lambda_1,\ldots$. We are not sure how to understand this, since evolution has only equipped us with intuition for situations that can be described by such a list, but that doesn't mean we cannot understand it. I don't think the goal should be to recover a description that uses such a list $\lambda_1,\ldots$, since it seems like an unjustified ad-hoc assumption.

 

[morrobay]

If in an experiment locality is given.* And realism is rejected in accounting for Bell violations,
then how is it that this non realism is so consistent and derivable
As in this simple inequality where both entangled particles are detected at A & B
0⁰ and 45⁰
45⁰ and 90⁰.
0⁰ and 90⁰
Then the inequality for both particles detected based on 1/2 (sin θ/2)<sup>2
n(0+45+) + n(45+90+) ≥ n (0+90+)
Results in .1464 ≥ .25.
And in this example where A and B = ± 1.
(AB) + (AB') + (A'B) - (A'B') ≤ 2, in experiment approaches 2√2
* Locality
t₀ photons from source sent to detectors at A and B
t₁ photons reach detectors and measurement process starts, photons are sent from A to B and B to A
t₂ measurements completed.
t₃ photons reach detectors A and B</sup>

 

[Ilja]

So there is always communications whether there is realism or not..

If one uses "there is" that means one makes a claim about existence, thus, one has to presuppose some form of realism. Without realism, "there is" is meaningless. There would be no difference between "there is a stone" and "there is a ghost", because what makes the second invalid - the non-existence of ghosts - is a claim about reality.

 

[DrChinese]

If one uses "there is" that means one makes a claim about existence, thus, one has to presuppose some form of realism. Without realism, "there is" is meaningless. There would be no difference between "there is a stone" and "there is a ghost", because what makes the second invalid - the non-existence of ghosts - is a claim about reality.

Again, it is more useful for "non-realism" to be replaced by "contextual reality" or "subjective reality" or "acausal reality". In these scenarios, there is not an element of reality independent of the observer's choice of measurement basis.

As to whether there is communication between Alice and Bob in the realistic scenarios: who is "communicating" with whom? (This goes back to gva's statement that "there is always communications whether there is realism or not".)