Has Zeilinger disproven DeBroglie-Bohm?

[dipstik]

I read (in Scientific American) that Zeilinger had disproven some non-local hidden varible theories. If he claims to have done this, does anyone know the article?

I'm also having trouble with determinism vs. realisim.

thanks

 

[cesiumfrog]

I vaguely remember hearing Zeilinger speak about doing a Bell type experiment over a sufficiently large length scale as to close one of those "loopholes" (guaranteeing that the measurements are space-like separated). If you want his articles, just search on google-scholar! But you shouldn't be interested in hidden variable theories: all the current evidence suggests strongly that such theories are impossible, and even if they aren't impossible they certainly aren't viable yet whilst mainstream quantum mechanics works.

Realism? If you doubt this, it will be difficult to formulate any physics.

Determinism? Philosophically, do you want it or not? Whether QM is deterministic is basically just a question of your interpretation of the formalism (eg. Copenhagen vs Everett). Historically, the deterministic nature of classical physics has caused enough headaches itself.

 

[JesseM]

But you shouldn't be interested in hidden variable theories: all the current evidence suggests strongly that such theories are impossible, and even if they aren't impossible they certainly aren't viable yet whilst mainstream quantum mechanics works.

What evidence suggests nonlocal hidden variable theories are impossible? Bell's theorem only rules out local hidden variables. As I understand it, it's already been proven that Bohm's interpretation makes identical predictions as orthodox nonrelativistic quantum mechanics, although I don't think it's known whether one can create a Bohmian version of quantum field theory.

 

[JesseM]

Also, if you're looking for the Zeilinger result, it's probably the one discussed on this thread. However, this only applies to nonlocal hidden variables theories which respect a condition known as "outcome independence" (meaning that the outcome of one experimenter's measurement is independent of the outcome of the other experimenter's measurement), but Bohm's interpretation does not respect this condition so it isn't ruled out by Zeilinger's result--see here for some more info.

 

[Demystifier]

although I don't think it's known whether one can create a Bohmian version of quantum field theory.

One can do that, but not in such a simple and elegant way as for nonrelativistic particles. There is some evidence that the Bohmian interpretation of strings may make the whole picture much simpler. As a bonus, the unification of two "not-even-wrong" theories (Bohmian mechanics and string theory) provides testable predictions, suggesting that the theory could be "even-right".

 

[DrChinese]

Realism? If you doubt this, it will be difficult to formulate any physics.

That's not a fair statement by any means. Everything in both QM is fully consistent with a non-realistic explanation, although that is not the only possible explanation. So apparently it IS possible to formulate physics if you doubt realism.

The realism in question is a very specific kind of realism: that all possible observables have definite values at all times. If you accept the non-realism possible under QM, then the limits of realism are defined by the Heisenberg Uncertainty Principle.

 

[Crosson]

My thoughts are the same as Dr Chinese, but I would like to point out that the "very specific kind of realism" he discusses is determinism, not in the sense of "given the initial state the future state is determined" but rather in the sense of having definite values.

 

[cesiumfrog]

the "very specific kind of realism" [Dr Chinese] discusses is [..] having definite values.

Isn't that just mainstream QM?

 

[JesseM]

The realism in question is a very specific kind of realism: that all possible observables have definite values at all times. If you accept the non-realism possible under QM, then the limits of realism are defined by the Heisenberg Uncertainty Principle.

I'm not sure this is a perfect definition--would you consider Bohmian mechanics to be a (nonlocal) "realistic" theory? In Bohmian mechanics my understanding is that only position has a well-defined value, although the results of any other types of measurements follow in a deterministic way from the particle's position and the state of the pilot wave. See the section on quantum observables from the Stanford Encyclopedia of Philosophy article on Bohmian mechanics.

 

[DrChinese]

I'm not sure this is a perfect definition--would you consider Bohmian mechanics to be a (nonlocal) "realistic" theory? In Bohmian mechanics my understanding is that only position has a well-defined value, although the results of any other types of measurements follow in a deterministic way from the particle's position and the state of the pilot wave. See the section on quantum observables from the Stanford Encyclopedia of Philosophy article on Bohmian mechanics.

What I am referring to is effectively Bell realism, as opposed to Bell locality. The issue being whether, as Einstein assumed, "the moon was there when you weren't looking at it." In this case, the question is not so much whether the particle exists, but whether all possible observables have definite values independent of the act of measurement.

Recall that with our entangled photons, they can be measured at identical settings and they will always give predictable results. This implies realism, at first glance. But when you add Bell's Theorem, you encounter the effects of the act of measurement. And that implies that either our realism - or our locality - assumption must fall. Zeilinger's experiment tends to point to realism as the assumption to drop. But as mentioned earlier in the thread, it is not clear that this is absolute.

 

[JesseM]

What I am referring to is effectively Bell realism, as opposed to Bell locality. The issue being whether, as Einstein assumed, "the moon was there when you weren't looking at it." In this case, the question is not so much whether the particle exists, but whether all possible observables have definite values independent of the act of measurement.

Recall that with our entangled photons, they can be measured at identical settings and they will always give predictable results. This implies realism, at first glance. But when you add Bell's Theorem, you encounter the effects of the act of measurement. And that implies that either our realism - or our locality - assumption must fall. Zeilinger's experiment tends to point to realism as the assumption to drop. But as mentioned earlier in the thread, it is not clear that this is absolute.

Bell's specific notion of "local realism" seems clear enough, but what I'm wondering is whether "realism" on its own is sufficiently clearly-defined that one could clearly say whether a given nonlocal interpretation or theory is realistic or not. For example, as I asked above, is Bohmian mechanics a realist interpretation? How about the MWI? (and in the latter case, can the MWI be called a local realist interpretation even if it doesn't really match Bell's notion of local realism?) I'd be inclined to say yes in both cases since both interpretations give you an objective picture of the universe as a self-contained system where measurements are treated using the same laws that govern quantum systems when they're not being measured, but I don't know if there's any official consensus on how realism is supposed to be defined.

 

[JesseM]

Isn't that just mainstream QM?

No, in mainstream QM all observables don't have definite values at all times (i.e. even when they're not being measured)--for example, a particle doesn't have a definite value for position and momentum simultaneously.

 

[xantox]

Bell's specific notion of "local realism" seems clear enough, but what I'm wondering is whether "realism" on its own is sufficiently clearly-defined

Einstein defined it as follows, "there exists a physical reality independent of substantiation and perception". Rejecting realism means that reality is different for different observers. Bohmian mechanics retains realism and rejects locality. MWI retains locality and rejects realism.

 

[cesiumfrog]

what I'm wondering is whether "realism" on its own is sufficiently clearly-defined

It seems like much disagreement stems from that word being interpreted here with a number of conflicting meanings.

So originally, I was commenting on xantox's notion ("metaphysical realism"?) that there exists something independent of, and therefore agreed upon by, all observers (though I wouldn't say MWI rejects this).

 

[metacristi]

I read (in Scientific American) that Zeilinger had disproven some non-local hidden varible theories. If he claims to have done this, does anyone know the article?

I'm also having trouble with determinism vs. realisim.

thanks

The weaker sense of determinism implies that all events have causes (Bohm's interpretation implies primarily this, not necessarily predetermination) whilst the stronger sense implies predetermination of all events (aka strong determinism, superdeterminism etc). Realism is a broad philosophical position ranging from the mere assumption that observational statements (about macro objects) are about an external reality independent of mind till strong versions of scientific realism which state that there is an external reality independent of mind which we can perceive and understand at least partially (via direct observations and scientific theories - all unobservables in our best existing theories being real).*

As for Zeilinger's experiment (apart from failing to discard Bohmian mechanics, pilot-wave solutions more generally) it has to be observed that the definition of 'realism' (used in the premises) is way too strong (basically noncontextual) or it is much more reasonable to adopt a contextual approach (where the definition of 'realism' is relaxed to take in account the influence of the measurement device; valid for the macro level too). I'm afraid it will be extremely difficult to eliminate contextual non-local hidden variables theories...for example even an experimental test of Kochen-Specker theorem will not be enough. Finally Zeilinger's experiment (as Aspect's by the way) fails to discard superdeterminism (which rejects counterfactual definiteness) and other 'conspiracy theories' where there is no freedom of choice for the experimenters so even local deterministic hidden variables theories are still alive (of course physicists reject superdeterminism because it lacks, to paraphrase Duhem, 'le bon sens' but this is far from counting as a rejection). The only rational conclusion which one can draw from the long history of (failed) attempts to discard hidden variables (enough no-go theorems here) is that we should refrain from claims of 'refuting hidden variables' before having very very strong evidence (it should not be forget that for 20 years almost all physicists believed that von Neumann's fifth postulate discarded all hidden variables theories...until 1952 when, in Bell's words, 'the impossible happened').


*Bohm's interpretation is definitely realistic in nature although it departs to some extent from classical physics and the program is far from being finished (the proposed, more detailed, ontology - the 'holographic hypothesis' - is far from being testable at this time)

 

[IMP]

Superdeterminism: in a millisecond after the bigbang, all of the particles positions, location, temp (etc) determined that I would be wearing a blue shirt today? The particles positions, location, and temp (etc) determined I would post this statement in this thread on physicsforums.com on 11-19-07?

 

[cesiumfrog]

Superdeterminism: in a millisecond after the bigbang, all of the particles positions, location, temp (etc) determined that I would be wearing a blue shirt today?

No, that's just ordinary determinism.

Super determinism is something that often appears in hidden variable theories, where you insist that every possible measurement has a definite result AND that there are no superluminal effects, therefore to reproduce the results of quantum experiments you have to explain the apparent conspiracy by which you always chose to measure exactly those particles whose properties (contrary to usual probability) already happened to be just so as to give the results predicted by mainstream QM. Such examples are impossible to explain if they involved (for all practical purposes) "uncorrelated" sets of data, so super determinism is an appeal to determinism that no two things can ever truly be treated as uncorrelated (thereby undermining mainstream analysis).

 

[Count Iblis]

Yes, superdetermism is the loophole that 't Hooft has argued saves his deterministic theory. If the universe is deterministic, then you cannot consider a counterfactual situation where the experimentor made a different choice while keeping everything else the same. That is impossible because if you evolve the counterfactual state back in time, the history of the universe leading to that state will start to diverge quite rapidly from the known history of the universe.

In fact, to get the history even remotely correct you would need to make sure that the entropy of the universe will decrease if you evolve the countrfactual state back in time. That can only be achieved by making changes in the entire universe to put it in a specially prepared state.

 

[Schneibster]

Hi Dr. Chinese, Jesse and I have discussed some of these matters before. Hi Jesse.

I think you're speaking, Dr., of the two principles, one of which must be violated for the quantum statistics based on uncertainty to be correct:
1. Bell realism, the principle that all variables have real values whether they can be measured or not under Uncertainty;
2. Locality, the principle that local effects stem from local causes.
It's my understanding that violation of Bell realism means that (for example- and it's the most-used method in Bell tests, I believe) when spin is measured on one axis of a particle, spin does not have a definite value on any other axis at that time. This accounts for the measurements of entangled particles on the same axis being correlated, by conservation of angular momentum, since they are not conjugate, and also for the measurements of two entangled particles on different axes not being correlated according to the applicable Bell inequality (depending on the number of states measured). On the other hand, violation of locality says that there is never a correlation between the values except at the moment when the values are measured, at which point a non-local "channel" (I believe that's the current terminology) of unknown type "collapses the wave function" (or whatever terminology fits your favorite interpretation- I'm sticking with Copenhagen for clarity, but I like TI this week; Jesse, IIRC, and if his opinion has not changed, likes Everett) and by some means causes the conservation law to be observed, if the measurements are in the same plane, or the quantum distribution if they are not (I'm not a fan of non-locality, thus my tendency to typecast it as a quantum conspiracy theory). In either case, of course, Bell realism fails; the only question is, does locality fail, or not?

Cramer has proposed an experiment to attempt to differentiate between the two basically by sending the idler photons from a DCQE through light pipes in order to get enough time delay to determine whether the signal photons stop showing interference before the idler photons get their path "flopped over" to be directed to the welcher weg detectors instead of the QE. This would appear to violate causality, though there is an argument that says that what's really happening is that the coincidence counting merely allows you to detect what's already there- that is, to differentiate between two interference patterns whose combination yields the appearance of non-interference when the QE is in use, or not when the welcher weg detector is.

This happens because the QE beam splitter receives idler photons from both slits, whereas the welcher weg beam splitter receives them from only one. The two beams received by the QE beam splitter interfere, expressing the quantum probability set by the slits and sorting the photons by phase, allowing detection of the interference pattern; there is no opportunity for this to occur at the welcher weg beam splitter, because it is only struck by one beam. Note that the QE beam splitter must always receive photons from both welcher weg beams; otherwise, the QE detector(s) are welcher weg detector(s), because all the photons only came from one path. The welcher weg splitter must always receive photons from only one slit; otherwise, it's not welcher weg information. Twist it how you might, these two things must always be true for the experiment to work. This is also the explanation for Malus' Law: phase, and interference, and how phase changes under decoherence. (I am into strong decoherence: that is, interactions are measurements, in the HUP sense, and measurement results in uncertainty of the conjugate parameters.)

Later, after doing a bit of research so I wouldn't look so stupid: I went back and looked, and found this thread. It looks like you guys have found Kramer's daughter's "Retrodiction" post; however, some people seem to think that it will be possible to see interference; I claim it will not, because without the coincidence detector, the two interference patterns cannot be disentangled and without differentiation between them it will be impossible to see anything but non-interference. And I don't see the coincidence detector in the Cramer experiment slide, and notes on the Dopfer slide make it apparent that Cramer intends to eliminate it. One cannot detect a single photon interfering. It's pretty much the sound of one hand clapping.

The idea, of course, is to differentiate between simple Bell non-realism, and more complicated non-locality (which includes Bell non-realism). The problem is, it may be impossible to differentiate them (depending on how baroque you want to get about non-locality) in the case of a negative result, i.e. no interference; the only definitive result is a positive one. And that means seeing interference without the coincidence detector, and that's not possible. All that's accomplished is to force the non-local explanation to be more baroque; it's not eliminated. (Although the idea of making the non-local crowd look more like conspiracy theorists is not unattractive. :D )

 

[JesseM]

Hey Schneibster. The OP of this thread suggests that Cramer might acknowledge that if his experiment succeeded it would require us to modify the theory of QM somewhat by adding nonlinear effects (meaning that the retrocausal effects would be impossible in 'standard' QM), although the link is to a powerpoint presentation which I don't have the software to read. I think there exist rigorous proofs that Bohmian mechanics must always give the same predictions as standard QM, and it's hard to see how retrocausal effects could be possible in Bohmian mechanics, a totally deterministic theory in which later states can be derived from earlier ones (although I suppose it's conceivable that, being deterministic, earlier states could in some sense 'anticipate' what measurement would be performed later).

 

[Schneibster]

The experiment PDF is gone; I'm looking to see if I saved a copy. I know it wound up being different from the Dopfer experiment; but I can't remember how. The description in the thread is almost good enough for me to imagine it. Oh well, hope I have a copy.

 

[cesiumfrog]

huh? PDF link still works for me. I'm just disappointed his website has nothing *new* about that experiment. Maybe I should email his co-authors next time (wonder what that undergrad has to say..)?

 

[Schneibster]

He had a document that showed how he intended to run the experiment; it wasn't a slide, looked more like a proposal. The proposal was rather different from the Dopfer experiment. I'm still looking. It was on the UW's server; linked from his page there, I thought. Cramer_2007.pdf or something very similar.

 

[cesiumfrog]

Yes. It's still there. Still works. Are you having trouble following the links?

 

[metacristi]

Superdeterminism: in a millisecond after the bigbang, all of the particles positions, location, temp (etc) determined that I would be wearing a blue shirt today? The particles positions, location, and temp (etc) determined I would post this statement in this thread on physicsforums.com on 11-19-07?

Superdeterminism is basically just another label for strong determinism (no matter the ultimate nature of consciousness), that is if we could 're-run' somehow the Universe from the beginning exactly the same things would happen* (the additional assumption that there are no non-local connections is logically distinct, as Bell put it we do not need non-locality to explain the result of Aspect's experiment if everything is predetermined). See here some of Bell's thoughts on the subject.


*but not all 'conspiracy theories' - which deny the free will of the experimenters - require it with necessity

 

[vanesch]

Superdeterminism is basically just another label for strong determinism

Well, my understanding is that superdeterminism makes the assumption that there are strong correlations between events which have a "long" causal link ; in other words, that we cannot assume "for all practical purposes" that events that "seem unrelated" (but which are, of course, in a deterministic universe, related somehow by a common cause in the past), are statistically independent, but rather have *very peculiar* correlations.

Indeed, the point is not so much that "unrelated" events (say, the number of leafs from your neighbour's tree that fell in your garden and the number you threw using 3 dice ; which have of course a common origin, namely the state of the universe 5 billion years ago) are not 100% strictly statistically independent, no, they have to be correlated in very peculiar ways (that reproduce, for instance, EPR experiment results). *This* is superdeterminism to me.

This is in fact the theoretical basis of astrology

 

[Schneibster]

Yes. It's still there. Still works. Are you having trouble following the links?

You sure it's not cached in your browser? Try Cntl-reload. IE claims it cannot find the document. It's "Nonlocal_2007.pdf." I've seen it before now, but I can't seem to put my hands on a copy.

 

[ueit]

Well, my understanding is that superdeterminism makes the assumption that there are strong correlations between events which have a "long" causal link ;

No such assumption is being made. Superdeterminism assumes that any given state is uniquely determined by the previous state. In some circumstances this may imply "strong correlations between events which have a "long" causal link" in others not. Using Occham's Razor one can stop sooner or later with the exploration of the causal chain.

in other words, that we cannot assume "for all practical purposes" that events that "seem unrelated" (but which are, of course, in a deterministic universe, related somehow by a common cause in the past), are statistically independent, but rather have *very peculiar* correlations.

Indeed.

Indeed, the point is not so much that "unrelated" events (say, the number of leafs from your neighbour's tree that fell in your garden and the number you threw using 3 dice ; which have of course a common origin, namely the state of the universe 5 billion years ago) are not 100% strictly statistically independent, no, they have to be correlated in very peculiar ways (that reproduce, for instance, EPR experiment results). *This* is superdeterminism to me.

On what basis do you find EPR correlations "peculiar"? Do you find, say, energy conervation a "peculiar" fact? I can, for example construct a very complicated experiment in which energy is transformed in different ways, using all kinds of devices including free-willed experimenters. Should I expect a violation of energy conservation because all those devices "seem unrelated" "for all practical purposes"? I believe not.
As for EPR, if a particle is only emitted when a signal from a available absorber arrives at the source, the "freedom assumption" falls. It is simply irrelevant how the absorber looks like or how it moves. The only thing that matters is that the absorber's motion is predictable so that it can be "calculated" at the source.

This is in fact the theoretical basis of astrology

Astrology is cut by Occham's Razor.

 

[JesseM]

From reading the wikipedia article I get the impression that superdeterminism is basically the same as the notion of a "conspiracy" in the initial conditions of the universe, which ensures that the hidden-variables state in which two particles are created will always be correlated with the "choice" of measurements that the experiments decide to make on them. So, for example, in any trial where the experimenters were predetermined to measure the same spin axis, the particles would always be created with opposite spin states on that axis, but in trials where the experimenters were not predetermined to measure the same spin axis, the hidden spin states of the two particles on any given axis would not necessarily be opposite.

Since in a deterministic universe the state of an experimenter's brain which determines his "choice" of what to measure on a given trial can be influenced by a host of factors in his past which have nothing to do with the creation of the particle (what he had for lunch that day, for example), the only way for such correlations to exist would be to pick very special initial conditions of the universe--the correlations would not be explained by the laws of physics alone (unless this constraint on the initial conditions is itself somehow demanded by the laws of physics).

 

[cesiumfrog]

You sure it's not cached in your browser? Try Cntl-reload. IE claims it cannot find the document.

It's currently the first link on http://faculty.washington.edu/jcramer/NLS/NL_signal.htm" . I'm not caching it locally, I even checked in IE (though I don't recommend you continue with that browser). Are you behind some kind of filter?

 

[JesseM]

On what basis do you find EPR correlations "peculiar"? Do you find, say, energy conervation a "peculiar" fact? I can, for example construct a very complicated experiment in which energy is transformed in different ways, using all kinds of devices including free-willed experimenters. Should I expect a violation of energy conservation because all those devices "seem unrelated" "for all practical purposes"? I believe not.

This is a poor analogy--there would be nothing in any classical experiment that would require the system being investigated to behave as if it "knew" in advance what choice of measurement an experimenter would make, and alter its behavior in anticipation, as would be required to explain EPR correlations in the superdeterminism explanation. In classical experiments we would expect complete statistical independence between the state of the system at moments before a measurement and the experimenter's choice of what measurement to perform (assuming the experiments were repeated multiple times and the experimenters made their choices each time on a whim), while the superdeterminism explanation is explicitly based on rejecting this sort of assumption of statistical independence.

 

[Schneibster]

It's currently the first link on http://faculty.washington.edu/jcramer/NLS/NL_signal.htm" .

I confirm that statement, but I still can't donwload it. You must be right, it must be my browser. My box is screwed up anyway; I've been lazily not handling it; looks like it's now becoming obstructive. Perhaps this will motivate me.

I'm not caching it locally, I even checked in IE (though I don't recommend you continue with that browser). Are you behind some kind of filter?

Thanks for checking. I don't prefer Internet Exploder; Firefox is going to be nice to get back to, once I get motivated.

 

[vanesch]

From reading the wikipedia article I get the impression that superdeterminism is basically the same as the notion of a "conspiracy" in the initial conditions of the universe, which ensures that the hidden-variables state in which two particles are created will always be correlated with the "choice" of measurements that the experiments decide to make on them. So, for example, in any trial where the experimenters were predetermined to measure the same spin axis, the particles would always be created with opposite spin states on that axis, but in trials where the experimenters were not predetermined to measure the same spin axis, the hidden spin states of the two particles on any given axis would not necessarily be opposite.

Since in a deterministic universe the state of an experimenter's brain which determines his "choice" of what to measure on a given trial can be influenced by a host of factors in his past which have nothing to do with the creation of the particle (what he had for lunch that day, for example), the only way for such correlations to exist would be to pick very special initial conditions of the universe--the correlations would not be explained by the laws of physics alone (unless this constraint on the initial conditions is itself somehow demanded by the laws of physics).

Yes, that's also how I see it.

 

[vanesch]

No such assumption is being made. Superdeterminism assumes that any given state is uniquely determined by the previous state. In some circumstances this may imply "strong correlations between events which have a "long" causal link" in others not.

That's "normal" determinism.

On what basis do you find EPR correlations "peculiar"? Do you find, say, energy conervation a "peculiar" fact? I can, for example construct a very complicated experiment in which energy is transformed in different ways, using all kinds of devices including free-willed experimenters. Should I expect a violation of energy conservation because all those devices "seem unrelated" "for all practical purposes"? I believe not.

No, because in this case, there is a clear and simple correlation at each step which is evident from the laws of nature. But the obscure correlation which needs to exist between the spins of the sent couples of particles, and the *choices* that the observers are making on each side, is NOT the consequence of a straightforward and obvious chain of cause-effect relationships, but must be due to very very peculiar "initial conditions" far far behind in time, as JesseM pointed out. In other words, this correlation doesn't follow from a straightforward application of the *laws of nature* as we know them, but rather from very very "improbable" initial conditions billions of years ago.

The difference between normal determinism and superdeterminism is that in normal determinism, we assume that all events which are not obviously related by a *rather straightforward* cause-effect relationship, can for all practical purposes be assumed to be statistically independent (even though one might expect *small* deviations from strict statistical independency, depending on the "cutoff" one places on the straightforwardness of the cause-effect relationships). In superdeterminism, we assume that arbitrarily strong correlations can exist for arbitrary long "chains of cause-effect", such as "emission of a pair of photons" and "brain of Alice to decide to put the analyser to 60 degrees".

As for EPR, if a particle is only emitted when a signal from a available absorber arrives at the source, the "freedom assumption" falls. It is simply irrelevant how the absorber looks like or how it moves. The only thing that matters is that the absorber's motion is predictable so that it can be "calculated" at the source.

Right, and if that motion is determined by choices of people, for instance, then in this "calculation" must also be included the entire dynamics of the brain of that person. This is where, in normal determinism, one considers the causal chain that makes "the source calculate" and the "brain think" too long to be statistically correlated.

 

[Count Iblis]

Perhaps superdeterminism can be more natural if particles don't really exist. You can imagine that there exists some deterministic theory and to do statistical computations you need to introduce ghost fields and a ghost Lagrangian which takes the form of the Standard Model.