How does QFT handle non-locality?

[vanhees71]

No, obviously not.

 

[Demystifier]

No, obviously not.

Then let me explain signal locality (and some of the other types of locality) in a few short steps.

1. In the realm of quantum foundations and interpretations, there are several different notions of locality/non-locality. Signal locality/non-locality is only one of them.

2. As you know, different interpretations claim that QM is local or non-local in one way or another. But signal locality, as one specific notion of locality, has a special status. It is special because all interpretations agree that QM has the property of signal locality.

3. So what is signal locality? Unlike other notions of locality, signal locality is a very antropomorphic concept. Signal locality means that you cannot send signal faster than light. Here "signal" means information that can be manipulated, controlled and measured by humans in practice.

4. What is signal locality not? For example, if there is a wf collapse, you cannot use it to send a signal faster than light. That's because collapse is random, so you cannot choose to which final state the wf will collapse. Since you cannot choose it, you cannot manipulate and control the collapse. Thus, even though in collapse there is some kind of information transfer faster than light, in collapse there is no signal faster than light. Therefore collapse is compatible with signal locality.

5. Similarly, non-local hidden variables such as Bohmian theory are also compatible with signal locality. For a simple explanation see
https://www.physicsforums.com/threa...ctual-definiteness.847628/page-2#post-5319182

6. Is QFT local? It depends on what exactly one mans by "local". It certainly has property of signal locality. It also has some other types of locality. However, it does not necessarily has all possible types of locality. Depending on interpretation, it may or may not be non-local due to collapse or due to hidden variables. From the known facts about QFT we cannot exclude such non-local features.

7. Is QFT non-local in some interpretation-independent sense? Yes! QFT violates Bell inequalities, and violation of Bell inequalities is also one (of many) notion of non-locality. This non-controversial type of non-locality can be reduced to the fact that QFT contains not only local operators $\phi_1(x)$, $\phi_2(x)$, ... but also "non-local" (more precisely, multi-local) operators such as $O(x,y)=\phi_1(x)\phi_1(y)+\phi_2(x)\phi_2(y)$. Clearly, this fact does not depend on interpretation.

I hope it helps.

 

[atyy]

Sure, but still the state change assumed by the collapse is instaneously acting over a long distance. Your argument of unobservability of the collapse is a perfect argument to just abandon the postulate of collapse.

It's clear that Alice get's the correct result about what Bob will find, assuming that after her measurement the state is $|uu \rangle$, but it doesn't apply that anything happens instantaneously to B's particle due to A's local spin measurement. So what you call a "collapse" here is just the adaption of A's description of the system after she made her measurement, it's not a statement of some physical process acting instantaneously on B's particle.

The collapse postulate cannot be abandoned even if it one is agnostic about its reality. This because it is very difficult to argue that it is "just the adaption of A's description of the system after she made her measurement, it's not a statement of some physical process acting instantaneously on B's particle". If it were true, then that would follow from the laws of probability, but I am not aware of any successful derivation of collapse as simply an updating of knowledge without any physical process.

This I don't understand. If the collapse is taken as a real physical phenomenon then it violates relativistic causality. If it's taken as something non-real, you can just forget about it. I don't know of any example of the application of quantum theory where you need to assume the collapse as a real physical process, and that's why I don't understand, why it is still used today (or after 1935, when EPR pointed out that it's contradicting relativistic causality).

Even if you treat collapse as non-real, you cannot save relativistic causality unless you assume something like many worlds, retrocausation etc. That is the content of the Bell theorem: relativistic causality is dead or empty.

One can be agnostic about the reality of collapse. However, it is wrong to reject on the basis of superluminal communication, since collapse does not allow superluminal communication. It is also wrong to reject collapse in order to save relativistic causality, unless one adopts many worlds, retrocausation etc, since apart from those ways of avoiding the Bell theorem, quantum mechanics implies that relativistic causality is dead or empty.

 

[vanhees71]

Then let me explain signal locality (and some of the other types of locality) in a few short steps.

1. In the realm of quantum foundations and interpretations, there are several different notions of locality/non-locality. Signal locality/non-locality is only one of them.

2. As you know, different interpretations claim that QM is local or non-local in one way or another. But signal locality, as one specific notion of locality, has a special status. It is special because all interpretations agree that QM has the property of signal locality.

3. So what is signal locality? Unlike other notions of locality, signal locality is a very antropomorphic concept. Signal locality means that you cannot send signal faster than light. Here "signal" means information that can be manipulated, controlled and measured by humans in practice.

4. What is signal locality not? For example, if there is a wf collapse, you cannot use it to send a signal faster than light. That's because collapse is random, so you cannot choose to which final state the wf will collapse. Since you cannot choose it, you cannot manipulate and control the collapse. Thus, even though in collapse there is some kind of information transfer faster than light, in collapse there is no signal faster than light. Therefore collapse is compatible with signal locality.

5. Similarly, non-local hidden variables such as Bohmian theory are also compatible with signal locality. For a simple explanation see
https://www.physicsforums.com/threa...ctual-definiteness.847628/page-2#post-5319182

6. Is QFT local? It depends on what exactly one mans by "local". It certainly has property of signal locality. It also has some other types of locality. However, it does not necessarily has all possible types of locality. Depending on interpretation, it may or may not be non-local due to collapse or due to hidden variables. From the known facts about QFT we cannot exclude such non-local features.

7. Is QFT non-local in some interpretation-independent sense? Yes! QFT violates Bell inequalities, and violation of Bell inequalities is also one (of many) notion of non-locality. This non-controversial type of non-locality can be reduced to the fact that QFT contains not only local operators $\phi_1(x)$, $\phi_2(x)$, ... but also "non-local" (more precisely, multi-local) operators such as $O(x,y)=\phi_1(x)\phi_1(y)+\phi_2(x)\phi_2(y)$. Clearly, this fact does not depend on interpretation.

I hope it helps.

Yes, that helps a lot, and it underlines that the assumption of a collapse as a physical objective process is empty and unnecessary, because you can never test it against the minimal (ensemble) interpretation.

 

[vanhees71]

The collapse postulate cannot be abandoned even if it one is agnostic about its reality. This because it is very difficult to argue that it is "just the adaption of A's description of the system after she made her measurement, it's not a statement of some physical process acting instantaneously on B's particle". If it were true, then that would follow from the laws of probability, but I am not aware of any successful derivation of collapse as simply an updating of knowledge without any physical process.
Even if you treat collapse as non-real, you cannot save relativistic causality unless you assume something like many worlds, retrocausation etc. That is the content of the Bell theorem: relativistic causality is dead or empty.

One can be agnostic about the reality of collapse. However, it is wrong to reject on the basis of superluminal communication, since collapse does not allow superluminal communication. It is also wrong to reject collapse in order to save relativistic causality, unless one adopts many worlds, retrocausation etc, since apart from those ways of avoiding the Bell theorem, quantum mechanics implies that relativistic causality is dead or empty.

I think we discuss in circles again, but for me the very successful application of local microcausal QFT to the real world proves this statement wrong. It explains perfectly the violation of Bell's inequality in accordance with very accurate observations thereof without killing relativistic causality. To the contrary: Relatistic causality is used in the very construction of this class of QT models. As I said before, locality and microcausality is sufficient but AFAIK not necessary for relativistic causality.

 

[Demystifier]

Yes, that helps a lot, and it underlines that the assumption of a collapse as a physical objective process is empty and unnecessary, because you can never test it against the minimal (ensemble) interpretation.

I agree with you that the minimal ensemble interpretation is in many respects better than the physical collapse interpretation. Yet, I don't think that the idea of a physical collapse is completely useless, at least for some physicists. For psychological reasons, many physicists can more easily think about physics if they have a visual picture in their mind of the physical processes involved. The minimal ensemble interpretation, unfortunately, does not provide such a picture. After all, that's why it is called minimal. Therefore some physicists look for alternative interpretations which do provide some picture. And among many pictures provided by many non-minimal interpretations, the physical collapse collapse interpretation is in some sense "minimal" itself. Namely, such a picture does not require any other object except the wave function, and, at the same time, does not require any other world except the world that we see. That's why the physical collapse picture is still popular among some physicists. And if that picture helps them to make calculations, as long as the results of their calculations do not differ from results of calculations done by physicists using other pictures or using no pictures at all, I do not see a reason to judge them for using a picture that works for them.

 

[vanhees71]

Hm, but also for the minimal interpretation, I can stick with the position representation and wavemechanics as long as we restrict ourselves to non-relativistic systems of constant particle number. I don't know, what the collapse can provide in addition to the ensemble representation in the sense of heuristic pictures. State preparation in the sense of von Neumann filter measurements are much more natural than when the collapse hypothesis is applied. I just don't bother about the formalism but filter out "partial beams" from the ensemble that don't have the properties I like to prepare, e.g., a certain spin state using a Stern Gerlach apparatus. I just block the unwanted beams and get a practically well-determined spin component in direction of the magnetic field. That's it. No complicated thinking in terms of fictitious collapses needed :-).

 

[atyy]

I think we discuss in circles again, but for me the very successful application of local microcausal QFT to the real world proves this statement wrong. It explains perfectly the violation of Bell's inequality in accordance with very accurate observations thereof without killing relativistic causality. To the contrary: Relatistic causality is used in the very construction of this class of QT models. As I said before, locality and microcausality is sufficient but AFAIK not necessary for relativistic causality.

But that is simply wrong. "Microcausality" is not what you believe it to be. "Microcausality" is not relativistic causality. "Microcausality" means no superluminal signalling. QFT in the minimal interpretation is not consistent with relativistic causality - this conclusion can only be evaded by eg. many-worlds or retrocausation.

 

[vanhees71]

Well, we seem to have different language. Microcausality+locality of the interactions indeed excludes superluminal signalling. Together with the dynamics of QT that implies relativistic causality, or what else do you need to establish it?

 

[Demystifier]

I just block the unwanted beams

This sentence is very problematic in minimal ensemble interpretation (MEI). Namely, this sentence sounds as if the "beam" is a physical object existing even without our observations. On the other hand, using only MEI, I think you cannot answer whether the beam physically exists without our observations. Thus, the language you use does not seem compatible with MEI. So either
i) you really use something more than MEI (even if you fail to recognize it), or
ii) within MEI you have to answer whether the beam exists without our observations, or
iii) stay agnostic about this question and adopt your language accordingly, to prevent false impression of believing in beams existing without our observations.

So what is your choice, i), ii), or iii)?

 

[atyy]

Well, we seem to have different language. Microcausality+locality of the interactions indeed excludes superluminal signalling. Together with the dynamics of QT that implies relativistic causality, or what else do you need to establish it?

In order to have relativistic causality, the Bell inequalities cannot be violated. So "no superluminal signalling" is a weaker constraint than "relativistic causality".

One way to see that although technically, the conditions on the quantum Hamiltonian seem to have the same conditions we impose on a classical relativistic theory, it is not the same because in QFT the Hamiltonian is not real. In the Heisenberg picture, the Hamiltonian governs the time evolution of all observables, including non-commuting observables. But non-commuting observables cannot have simultaneous reality. So in the Heisenberg picture, the Hamiltonian is not real. In the Schroedinger, picture the Hamiltonian governs the evolution of the wave function, which is also not real (or at least not necessarily real).

In general, in the minimal interpretation, QFT and QM are not theories of reality. This is why relativistic causality is empty in the minimal interpretation. If QFT and QM are taken to be theories of reality, then the Bell theorem forces (except for the usual exceptions like MWI) QFT and QM to violate relativistic causality.

Whatever language one uses, there is the idea that the constraints in order of strength from weak to strong are:
-no superluminal signalling
-quantum causality or correlations
-relativistic causality

 

[vanhees71]

This sentence is very problematic in minimal ensemble interpretation (MEI). Namely, this sentence sounds as if the "beam" is a physical object existing even without our observations. On the other hand, using only MEI, I think you cannot answer whether the beam physically exists without our observations. Thus, the language you use does not seem compatible with MEI. So either
i) you really use something more than MEI (even if you fail to recognize it), or
ii) within MEI you have to answer whether the beam exists without our observations, or
iii) stay agnostic about this question and adopt your language accordingly, to prevent false impression of believing in beams existing without our observations.

So what is your choice, i), ii), or iii)?

Of course I have in mind the most simple example for a von Neumann filter measurement like a Stern-Gerlach experiment. Say, I have silver atoms as in the original experiment and want to prepare a pure state with spin up. I send the silver atoms (originally in a thermal state from the oven) through an appropriate magnetic field. This leads to a state, where the position of the atoms is entangled with its spin-z component. In other words the many particles of the ensemble of silver atoms are sorted into two well-separated locations, and at each location they have an almost perfectly prepared spin-z up or down. Now I put something in the beam at the location where the particles have spin down. What's then left are particles with spin up, i.e., I have prepared a pure state with determined spin-z component out of a thermal ensemble (or any other initial state you have in the beginning). I don't see, where I need more than just the postulates of quantum theory to come to this setup of a von Neumann filter preparation.

So I think (ii) is the right answer. The quantum dynamics, including several conservation laws, ensure that the silver atoms are where they should be in the spin-position entangled state after they've run through the Stern-Gerlach apparatus, and I can use the remaining beam to check, whether it is really a pure spin-up state by measuring the spin-z component again using a second Stern-Gerlach apparatus. I know that the particles are in a pure spin-up state after the preparation procedure due to the very natural laws, I've established by observation before. Any experiment in physics rests on the physical laws we use to construct the apparatus to perform it, and the very purpose of experiments is to check whether this works really out. If you find a discrepancy between your expectations and the outcome of the experiment, you have to figure out whether it's due to inaccuracies in your experiment or whether you found a violation to the physical laws known so far. Then you have a discovery ;-)).

 

[vanhees71]

In order to have relativistic causality, the Bell inequalities cannot be violated. So "no superluminal signalling" is a weaker constraint than "relativistic causality".

I think I give up. Obviously I cannot make this very simple argument clear. Just once more very brief: Relativistic local microcausal QFT by construction obeys relativistic causality, including no superluminal signalling, and it precisely describes all violations of the Bell inequality so far. This is no contradiction to Bell's theoryem, because QFT is not a local deterministic HV model of the world but a special realization of QT tailored to be consistent with relativistic causality. So relativistic causality does not exclude the violation of Bell's inequality but only local deterministic theories do so, and the observed violations in my opinion rule out any local deterministic theory. It seems that even the most sceptical physicists nowadays believe that loophole-free Bell tests have been performed (at least according to several recent publications titled as loophole-free Bell tests, but it's a Nature paper ;-))).

 

[atyy]

I think I give up. Obviously I cannot make this very simple argument clear. Just once more very brief: Relativistic local microcausal QFT by construction obeys relativistic causality, including no superluminal signalling, and it precisely describes all violations of the Bell inequality so far. This is no contradiction to Bell's theoryem, because QFT is not a local deterministic HV model of the world but a special realization of QT tailored to be consistent with relativistic causality. So relativistic causality does not exclude the violation of Bell's inequality but only local deterministic theories do so, and the observed violations in my opinion rule out any local deterministic theory. It seems that even the most sceptical physicists nowadays believe that loophole-free Bell tests have been performed (at least according to several recent publications titled as loophole-free Bell tests, but it's a Nature paper ;-))).

Sure, but then you don't mean anything by relativistic causality. You just mean whatever QFT is, in a way which is not microcausality. At best your definition is tautological. What is worse, is that it is very misleading.

 

[vanhees71]

Why is it misleading? Indeed, it's a tautology (although it's not trivial to prove that it is one). Again, what do you need in addition to the impossibility of faster-than-light propagation and a causal dynamical law that describes the time evolution of observables to call a theory "relativistically causal"?

Also "reality" is a pretty empty idea. After all the discussions here in the forum and also reading some papers, I could not make out what's the clear definition of what makes a theory "realistic". For me QFT in the minimal interpretation is very "realistic", because it describes all known phenomena concerning the behavior of elementary particles.

 

[atyy]

Why is it misleading? Indeed, it's a tautology (although it's not trivial to prove that it is one). Again, what do you need in addition to the impossibility of faster-than-light propagation and a causal dynamical law that describes the time evolution of observables to call a theory "relativistically causal"?

Also "reality" is a pretty empty idea. After all the discussions here in the forum and also reading some papers, I could not make out what's the clear definition of what makes a theory "realistic". For me QFT in the minimal interpretation is very "realistic", because it describes all known phenomena concerning the behavior of elementary particles.

Well, there are two concepts - no superluminal signalling and no local hidden variables. You can pick whichever one you wish to be "relativistic causality". But the fact remains that the minimal interpretation has collapse, and that violates "no local hidden variables". Also, collapse does not violate "no superluminal signalling". So whichever interpretation you choose for "relativistic causality", you cannot reject collapse on the your grounds. If you reject collapse because it violates "no local hidden variables", that is wrong since there is no way to save "no local hidden variables". If you reject collapse because it violates "no superluminal signalling", that is wrong because collapse is consistent with "no superluminal signalling".

 

[vanhees71]

No you confusing me even more. Standard QFT has no superluminal signalling and no local hidden variables (and also no collapse). Collapse explicitly violates "no superluminal signalling", because proponents claim that A's measurement of the polarization of the photon at her place instantaneously changes the polarization of B's photon at a far distant place. Nothing observed and also QFT doesn't necessarily justify this claim. The only thing that changes by A's measurement is her knowledge about B's measurement's outcome because of the polarization entanglement of the two photons observed. There is no action at a distance according to standard QFT because the interaction of one of the photons with A's measurement apparatus is local and cannot affect anything spacelike separated from the local detection event. So how can collapse be consistent with "no superluminal signalling"?

 

[atyy]

No you confusing me even more. Standard QFT has no superluminal signalling and no local hidden variables (and also no collapse). Collapse explicitly violates "no superluminal signalling", because proponents claim that A's measurement of the polarization of the photon at her place instantaneously changes the polarization of B's photon at a far distant place. Nothing observed and also QFT doesn't necessarily justify this claim. The only thing that changes by A's measurement is her knowledge about B's measurement's outcome because of the polarization entanglement of the two photons observed. There is no action at a distance according to standard QFT because the interaction of one of the photons with A's measurement apparatus is local and cannot affect anything spacelike separated from the local detection event. So how can collapse be consistent with "no superluminal signalling"?

Standard QFT has collapse. It just means that after Alice measures |uu>+|dd> to get the u result, the state collapses to |uu>. Of course you don't have to ascribe reality to the collapse. But even if you do, it doesn't violate no superluminal signalling, because Bob cannot tell by measuring his spin whether Alice has measured yet.

 

[Demystifier]

Collapse explicitly violates "no superluminal signalling",

Argh! Your ability of fast forgeting is amazing. Just a few posts above I explained you that it is not so, and you said that this post was helpful to you and you liked it, but now you wrongly repeat again that collapse violates no superluminal signalling.

 

[Demystifier]

So I think (ii) is the right answer.

I guess it means that you think that the beam does exist even if we don't measure it. Am I right? But then
1) You are not adherent of MEI (even if you think you are), and
2) The beam itself is a hidden variable (because it exists without measurement), so the Bell theorem implies that entangled beams involve a kind non-locality that you can't accept. But of course, you can't understand it as long as you keep repeating wrong claims such as those that "collapse contradicts signal locality".

 

[Demystifier]

Standard QFT has ... no collapse.

Yes it has. For example, a few posts above atyy quoted the precise equation in Weinberg's QFT I describing the collapse. What standard QFT does not have is an answer to the question whether the collapse is a real physical process or only a mental tool for information update. Standard QFT is agnostic about that. But it is precisely this agnosticism (namely refusing to make clear statements about certain interesting questions) that creates a lot of confusion about foundational issues among physicists who read only standard QFT/QM.

So if you want to really understand non-locality of QFT, collapse, hidden variables, etc ... forget standard books such as Weinberg. Instead, take a look at a text that more seriously deals with such questions. For example, F. Laloe, "Do We Really Understand Quantum Mechanics?" would be a good choice. After a while, you should realize that QFT is more non-local than you currently think.

 

[vanhees71]

Argh! Your ability of fast forgeting is amazing. Just a few posts above I explained you that it is not so, and you said that this post was helpful to you and you liked it, but now you wrongly repeat again that collapse violates no superluminal signalling.

No, I didn't forget that posting, but there you didn't mention the collapse but defined signal locality as being fulfilled by QT (which I agree with), but if you put the collapse hypothesis (which for me is clearly an addition to minimally interpreted QT) you explicitly assume signal nonlocality, because it implies that a quantum state instantaneously collapses after a measurement (even if this measurement involves only local interactions of (parts of) the system with the measurement appartus). It is a (for me fictitious) process outside of the quantumtheoretical dynamics.

 

[vanhees71]

Standard QFT has collapse. It just means that after Alice measures |uu>+|dd> to get the u result, the state collapses to |uu>. Of course you don't have to ascribe reality to the collapse. But even if you do, it doesn't violate no superluminal signalling, because Bob cannot tell by measuring his spin whether Alice has measured yet.

It is an assumption that after alices measurement the state collapses to $|uu \rangle$. How do you know that from quantum theory, and how can that be even independent of how Alice has measured her photon?

 

[zonde]

Relativistic local microcausal QFT by construction obeys relativistic causality, including no superluminal signalling, and it precisely describes all violations of the Bell inequality so far.

What about local microcausality of Fock states (or rather superposition of Fock states)? Fock spaces are by construction nonlocal when they incorporate Hilbertspaces of distant particles. I have asked something similar before but somehow I have not received any answer. Is there some problem with my question?

 

[atyy]

It is an assumption that after alices measurement the state collapses to $|uu \rangle$. How do you know that from quantum theory, and how can that be even independent of how Alice has measured her photon?

I know that from quantum theory, because quantum theory has exactly the same structure as quantum mechanics, which has collapse (eg. Landau & Lifshitz, Cohen-Tannoudji, Diu & Laloe, Sakurai, Nielsen & Chuang). The collapse is dependent on the result Alice gets when she measures her photon.

 

[mfb]

Sorry atyy, but repeating that claim over and over again won't make it true.
Most textbooks use the Copenhagen interpretation. So what?
Most textbooks about classical mechanics discuss its application on inclined planes. Does this mean inclined planes are a crucial part of the theory of classical mechanics? Does classical mechanics break down if you don't talk about inclined planes?

The analogy is not perfect as you can use the equations of classical mechanics to describe inclined planes, but you cannot use the equations of quantum mechanics to describe collapses.

 

[atyy]

Sorry atyy, but repeating that claim over and over again won't make it true.
Most textbooks use the Copenhagen interpretation. So what?
Most textbooks about classical mechanics discuss its application on inclined planes. Does this mean inclined planes are a crucial part of the theory of classical mechanics? Does classical mechanics break down if you don't talk about inclined planes?

The analogy is not perfect as you can use the equations of classical mechanics to describe inclined planes, but you cannot use the equations of quantum mechanics to describe collapses.

The claim is true. Are you assuming MWI in order to avoid collapse?

 

[mfb]

I'm not assuming anything, I am just aware that there are multiple interpretations with different answers to questions of locality, determinism and so on. And there is no particular reason to prefer one over the other. Copenhagen is so widespread mainly for historic reasons.

 

[atyy]

I'm not assuming anything, I am just aware that there are multiple interpretations with different answers to questions of locality, determinism and so on. And there is no particular reason to prefer one over the other. Copenhagen is so widespread mainly for historic reasons.

But are there? As far as I know, Copenhagen (in one flavour or another) is the only consensus interpretation. All other interpretations have some problem - a technical problem, so it is not a matter of taste.

 

[vanhees71]

I'm not assuming anything, I am just aware that there are multiple interpretations with different answers to questions of locality, determinism and so on. And there is no particular reason to prefer one over the other. Copenhagen is so widespread mainly for historic reasons.

There is not one "Copenhagen interpretation". I think the minimal interpretation (which doesn't use a collapse or unobservable parallel universes but just uses the quantum formalism and the probabilistic interpretation of the states a la Born) is also a flavor of the Copenhagen interpretation, but that doesn't matter too much. I don't know any example of an experiment, for which you need to invoke a collapse assumption, and since the collapse assumption is at least very problematic in the context of the EPR problem, I simply don't use it.

I think the minimal interpretation (which I consider to be a flavor of the Copenhagen interpretation) is the only one which is really consensus among physicists. All other interpretations (including some flavors of the Copenhagen interpretations) add some additional assumption, which in my opinion is just unnecessary to use
quantum theory as a physical description of the (so far known part) world.

Under collapse, I unserstand the assumption that there is a split of phenomena in a "quantum" and a "classical" part, and neither describes the dynamics of the described systems completely and the "collapse" is a process which again is inconsistent with either dynamics, because both classical and quantum dynamics in the relativistic realm by construction does not involve instantaneous interactions or signal propagation, while the collapse assumption exactly claims this: the measurement of A's photon's polarization immediately changes B's measurement of his photon at a far-distant place. This assumption is of course unnecessary, because the outcome of B's measurement is not affected by the collapse. The probabilities for finding a certain polarization state at Bob's place are given as well by the initial entangled state, in which the biphoton has been prepared, including the non-classical correlations violating Bell's inequality. This is the minimal interpretation, and the collapse even unobservable. So why should I assume it to happen?