Nature Physics on quantum foundations

[martinbn]

I guess I understand what is meant by "relativistic microcausal QFT is local". I don't get what is meant by "it allows correlations between far-distant parts of a quantum system (described by entanglement)", and how it would give me a notion as intuitive and unproblematic as "nonlocal randomness". So I get a feeling like "maybe those are nice words, but what do you want to tell me with those words". Somehow it feels "too abstract" to me.

I don't want to be the translator from @vanhees71 to English, but I think that non-local randomness is something he would agree is part of QFT.

 

[Demystifier]

Are you saying that in principle Bob could figure out whether Alice has done or not something with her particle?

Yes, according to some interpretations/theories. Example is Bohmian mechanics out of quantum equilibrium.

 

[Demystifier]

I don't want to be the translator from @vanhees71 to English, but I think that non-local randomness is something he would agree is part of QFT.

I don't think that he would agree that randomness is non-local.

 

[vanhees71]

I guess I understand what is meant by "relativistic microcausal QFT is local". I don't get what is meant by "it allows correlations between far-distant parts of a quantum system (described by entanglement)", and how it would give me a notion as intuitive and unproblematic as "nonlocal randomness". So I get a feeling like "maybe those are nice words, but what do you want to tell me with those words". Somehow it feels "too abstract" to me.

The correlations between far-distant parts of a quantum system (e.g., an entangled two-photon state with the photons measured at far-distant places) described by entangled states of course a consistent with relativistic microcausal QFT, and for me this implies that these long-ranged correlations are described by a local theory (QFT) and thus that what's violated in Bell's local realistic HV theories is "realism" (i.e., the assumption that all observables take determined values).

 

[vanhees71]

I don't want to be the translator from @vanhees71 to English, but I think that non-local randomness is something he would agree is part of QFT.

I don't know, what "non-local randomness" means though.

 

[martinbn]

Yes, according to some interpretations/theories. Example is Bohmian mechanics out of quantum equilibrium.

But according to QM (and current experiments) it is not possible. Of course you can find theories that have all kinds of interactions. Newtonian gravity has action at a distance.

 

[martinbn]

I don't know, what "non-local randomness" means though.

It just means randomness.

 

[martinbn]

I don't think that he would agree that randomness is non-local.

Non-local here is superfluous.

 

[PeterDonis]

The momentum's direction cannot be inferred from position of the flash, if the two branches of the wave function are made parallel after their split by the magnet

But they aren't in the standard SG experiment, which is what we're discussing. Nothing is done to the wave function between exiting the SG magnet and hitting the detector.

What the position of the flash is really correlated with is the position of the particle at the time of detection, not its momentum.

If we want to get really, strictly technical, yes, the position of the flash is a measurement of the position of the particle, which gets entangled with its momentum (which output beam it is in) by the detector, and the momentum gets entangled with the spin by the SG magnet. So strictly speaking there are two stages of deduction required to get from the observed position of the flash on the detector to the "measured" value of spin.

 

[gentzen]

I don't know, what "non-local randomness" means though.

"Not predetermined" random outcomes at spacelike separated (spacetime) points/events which are not independent.

The "not predetermined" is the important part of that notion, and also the part where some vagueness enters. Just because the outcome was not yet "fully" predetermined at any point in the intersection of the past lightcones of the points/events doesn't mean that it got determined exactly at the moment where the random outcomes became known and recorded.

But in the simplest form of the notion, one could imagine it indeed as if the random outcomes only got determined at the moment where they got recorded. The notion is unproblematic even in this "simple but unrealistic" form.

 

[haushofer]

Wave-particle duality is no phenomenon but a theoretical concept that's outdated for about 100 years.

I see this claim more often on PF, but why exactly? To me, the duality states that quantum objects show both particle and wave behaviour, which is captured in a new ontological category for the quantum object we call "quantum particle". This is how I've understood this concept for a long time. What's outdated about that? I'd say the concept has evolved, not that it's outdated.

Didn't read all the replies, sorry if this has been asked before.

 

[Demystifier]

But according to QM (and current experiments) it is not possible. Of course you can find theories that have all kinds of interactions. Newtonian gravity has action at a distance.

Of course. Let me remind you that we discuss examples that illustrate the difference between superluminal signalling and superluminal action. Standard QM is not a good example, which is why we discuss other examples.

 

[Demystifier]

I see this claim more often on PF, but why exactly? To me, the duality states that quantum objects show both particle and wave behaviour, which is captured in a new ontological category for the quantum object we call "quantum particle". This is how I've understood this concept for a long time. What's outdated about that? I'd say the concept has evolved, not that it's outdated.

Nothing is outdated about notion of quantum particle. What is outdated is that sometimes it behaves like a classical wave and sometimes like a classical particle.

 

[haushofer]

Nothing is outdated about notion of quantum particle. What is outdated is that sometimes it behaves like a classical wave and sometimes like a classical particle.

But what's wrong with that? In certain cases the wavefunction is sharply peaked, which means that the quantum particle exhibits "classical behaviour". But that doesn't make it a "classic particle".

Just because a sheep can be fluffy it doesn't mean it's a pillow; we just perceive that in that case (unshaved) it shows "pillow-like behaviour".

Maybe it's nomenclature, but the wave-particle duality is not a statement about the ontology of quantum particles, afaik. That's why I'm surprised by VanHees' adament statement.

But maybe this is off-topic.

 

[Lynch101]

The correlations between far-distant parts of a quantum system (e.g., an entangled two-photon state with the photons measured at far-distant places) described by entangled states of course a consistent with relativistic microcausal QFT, and for me this implies that these long-ranged correlations are described by a local theory (QFT) and thus that what's violated in Bell's local realistic HV theories is "realism" (i.e., the assumption that all observables take determined values).

RQ: Measurements result in definite, singular values, don't they?

Q1: What does this interpretation say happens, such that the measurements made on far-distant parts of a quantum system result in definite, singular values?

Q1A: Do the [far-distant] parts have definite values from the moment they leave the preparation device and along their travel towards the measurement devices of Alice and Bob, with the correlations being explained by virtue of their shared preparation?

Q1B: Or, is each part in a superposition i.e. don't have definite values as they leave the preparation device and travel towards the measurement devices?

If the answer to Q1B is yes, then Q1 remains to be answered.

 

[martinbn]

Of course. Let me remind you that we discuss examples that illustrate the difference between superluminal signalling and superluminal action. Standard QM is not a good example, which is why we discuss other examples.

No, we are discussing whether a statement was conraversial or not. If it is a consequence of QM it shouldn't be.

 

[Morbert]

So strictly speaking there are two stages of deduction required to get from the observed position of the flash on the detector to the "measured" value of spin.

I don't think you will ever be able to eliminate such deductions entirely, as we will always need to construct a logical implication between the macroscopic measurement outcome read by the experimenter, and the measured variable of the quantum system. I.e. There will always be an intermediate dynamical model of both the measured system and relevant degrees of freedom of the measurement apparatus that justifies any inferences we make from an experiment.

E.g. forget the SGE experiment and just consider a plate detector measuring the position of a photon that strikes it. Is the flash on this plate directly measuring the photon position, or is it indirectly measuring the photon position, since we might say it is really recording an electron cascade in one of its microchannels induced by the photon?

 

[PeterDonis]

we will always need to construct a logical implication between the macroscopic measurement outcome read by the experimenter, and the measured variable of the quantum system.

Yes, this is obvious since by hypothesis the measuring device and the measured system are different degrees of freedom. But one can still look at how many stages of deduction are required to get from the measuring device to the observable of the measured system that we are interested in.

 

[Demystifier]

Maybe it's nomenclature

Yes, and I'd say this nomenclature is somewhat outdated.

 

[Demystifier]

How can there be a non-local action at a distance and no faster than light signaling at the same time?

Let me remind you that we discuss examples that illustrate the difference between superluminal signalling and superluminal action.

No, we are discussing whether a statement was conraversial or not.

Do I miss something? Did you perhaps got a satisfying answer to your question above and then switched to another topic? Discussions with you would be much easier if you could occasionally say something like: "OK, now I understand this, thank you for explaining it to me, now I have another question.". Or if that would be too much, a simple "like" would be enough to signal that I answered your question at least partially satisfying. Or if a "like" would be too much as well, you could still indicate somehow that you are no longer interested in the initial question. Otherwise, I have an impression that all your later questions are motivated by the initial one.

So, what's your question, exactly?

 

[CoolMint]

I see this claim more often on PF, but why exactly? To me, the duality states that quantum objects show both particle and wave behaviour, which is captured in a new ontological category for the quantum object we call "quantum particle". This is how I've understood this concept for a long time. What's outdated about that? I'd say the concept has evolved, not that it's outdated.

Didn't read all the replies, sorry if this has been asked before.

Because the classical scale of everyday objects is just too imposing.
The world is best described with the use of the 4 fundamental forces of nature - the Gravitational force, the Weak Nuclear force, the Electromagnetic force and the Strong Nuclear force.
We get 'particles'("matter") when these forces interact and produce observable and measurable effects. The wavefunction is almost certainly a mathematical abstract and nothing else and what it describes is the unobservable lower layer that underlies the observations.
The Wave/particle duality is an inferior concept to the new notion of how the world works:

Quantum fields(mathematical entities describing the underlying reality/mechanisms of nature) -- The four fundamental forces(which produce measurable outcomes upon a measurement context) -- Measurement results(observable macro reality of tables and chairs)

The wave/particle notion seems to suggest that there are two entities at play(waves and particles). The new notion explicitly states that the field is fundamental and the 'particles' are emergent momentary excitations.

 

[vanhees71]

"Not predetermined" random outcomes at spacelike separated (spacetime) points/events which are not independent.

In other words that's what I call "far-distant correlations".

The "not predetermined" is the important part of that notion, and also the part where some vagueness enters. Just because the outcome was not yet "fully" predetermined at any point in the intersection of the past lightcones of the points/events doesn't mean that it got determined exactly at the moment where the random outcomes became known and recorded.

That's true for "von Neumann filter measurements". Usually in experiments with photons the photons get absorbed by the detector. Then there are no photons with any properties prepared. In such a case we simply get a measurement result with probabilities according to the quantum state (prediction), which we can test with these experiments on an ensemble of equally prepared systems.

But in the simplest form of the notion, one could imagine it indeed as if the random outcomes only got determined at the moment where they got recorded. The notion is unproblematic even in this "simple but unrealistic" form.

That's my point! It's "unproblematic", but particularly this is what's denied by all those who think that there's a "measurement problem", because it's very hard to accept that completely undetermined properties can be strongly correlated, even if the measured properties are measured at parts of the system at very far distant places. This inseparability feature of entangled state was what Einstein really bothered (according to his Dialectica article of 1948 which is much clearer than the (in)famous EPR article).

 

[Demystifier]

That's true for "von Neumann filter measurements".

I would like to stress that all POVM measurements, not only projective measurements, are described by a collapse postulate in standard QM. The QM postulates are summarized nicely in the following excerpt from the book "Quantum Computations and Quantum Information" by Nielsen and Chuang. Eq. (2.160) is the collapse postulate for general (not necessarily projective) measurements.

 

[Lord Jestocost]

This inseparability feature of entangled state was what Einstein really bothered ...

To my mind, Einstein’s illusion was that the experiential reality could be represented by a physical theory that could be based on the assumption that the entities and their characters the theory is about might exist objectively in an ontological sense and thus literally experiencer-independently (“scientific realism”).

 

[gentzen]

I don't want to be the translator from @vanhees71 to English, but I think that non-local randomness is something he would agree is part of QFT.

[...]

In other words that's what I call "far-distant correlations".

Wow, martinbn was indeed right!

That's my point! It's "unproblematic", but particularly this is what's denied by all those who think that there's a "measurement problem", because it's very hard to accept that completely undetermined properties can be strongly correlated, even if the measured properties are measured at parts of the system at very far distant places.

I had a conversation with Ruth E. Kastner (a long time ago) on Mateus Araújo's blog, where she agreed that

Of course there is nothing wrong with instrumentalism as a ‘shut up and calculate’ tactic for evading the conceptual and physical puzzles presented by QM. What is wrong is elevating that evasion to a dogmatic prescription about how to ‘interpret’ QM–which is what we see from many instrumentalists. That’s the only thing I’ve been contesting.

Except for using ‘shut up and calculate’ as a bad name for instrumentalism (I implicitly proposed 'let me calculate and explain' instead in that exchange), I had no problems (and still don't have) with that reaction. At about the same time also Steven Weinberg's attack on quantum interpretations and specifically instrumentalism appeared, which I found totally unacceptable back then. At some later point I read something which temporarily made me understand his attack (it had something to do with being able to explain such stuff in a textbook in a satisfying way), but I forgot it again in the meantime.

That's true for "von Neumann filter measurements".

I would like to stress that all POVM measurements, not only projective measurements, are described by a collapse postulate in standard QM. The QM postulates are summarized nicely in the following excerpt from the book "Quantum Computations and Quantum Information" by Nielsen and Chuang. Eq. (2.160) is the collapse postulate for general (not necessarily projective) measurements.

Eq. (2.160) (which first appeared as Eq. 2.92) does not describe a POVM measurement. Section 8.2.4 explains something about the interpretational background of that formulation, but (from my POV) this doesn't change that vanhees71 is "not wrong" here.
I learned this later and also that this equation is related to Kraus operators

The axioms are the same as defining Kraus operators, or quantum channels/maps/operations, as N&C do in 8.2.4, and they're even on wikipedia

N&C never mention Kraus operators, but refer to [Kra83] (K. Kraus. States, Effects, and Operations: Fundamental Notions of Quantum Theory. Lecture Notes in Physics, Vol. 190. Springer-Verlag, Berlin, 1983) via "The theory of generalized measurements which we have employed was developed between the 1940s and 1970s. Much of the history can be distilled from the book of Kraus [Kra83]." and "We mention just a few key references, primarily the book by Kraus [Kra83] , which contains references to much earlier work on the subject."

 

[vanhees71]

[...]

Wow, martinbn was indeed right!I had a conversation with Ruth E. Kastner (a long time ago) on Mateus Araújo's blog, where she agreed that

Except for using ‘shut up and calculate’ as a bad name for instrumentalism (I implicitly proposed 'let me calculate and explain' instead in that exchange), I had no problems (and still don't have) with that reaction. At about the same time also Steven Weinberg's attack on quantum interpretations and specifically instrumentalism appeared, which I found totally unacceptable back then. At some later point I read something which temporarily made me understand his attack (it had something to do with being able to explain such stuff in a textbook in a satisfying way), but I forgot it again in the meantime.

The point is, of course, that you are free to "want more" from science than it "promises" to do, i.e., to deliver a way to find "laws of nature" that describe, how nature behaves (or more precisely how we observe it to behave). That's why there are so many "interpretations" of the QT formalism. In principle it indeed describes what we observe (except that there's this fundamental problem of "quantum gravitation", i.e., the incompatibility between GR and QT) when just accepting the probabilistic interpretation of the "quantum state", using Born's rule, as well as the quantum-mechanical time evolution. That's in a way a "shut up and calculate" paradigm, because we restrict ourselves of the purely scientific meaning of the theory and don't expect to find some "deeper truth" in it. What I never understood is the fact that such a demand nobody ever had in view of classical physics ;-)).

Eq. (2.160) (which first appeared as Eq. 2.92) does not describe a POVM measurement. Section 8.2.4 explains something about the interpretational background of that formulation, but (from my POV) this doesn't change that vanhees71 is "not wrong" here.
I learned this later and also that this equation is related to Kraus operators

I don't think that the measurement problem, whatever you think it might be, is solved by generalizing the very idealized idea of von Neumann filter measurements (POV) with POVMs, which is simply taking into account the possibility for "incomplete measurements", which is almost always what we are able to do in the lab. The probabilistic nature of the quantum state persists also in this more general formulation. I don't see that POVMs is something that goes much beyond standard QT, although of course you can take POVMs as the measurement postulate as in the book quoted above (this seems to be a very concise formulation of the postulates, which is much more careful than the standard formulations in the introductory textbooks).

N&C never mention Kraus operators, but refer to [Kra83] (K. Kraus. States, Effects, and Operations: Fundamental Notions of Quantum Theory. Lecture Notes in Physics, Vol. 190. Springer-Verlag, Berlin, 1983) via "The theory of generalized measurements which we have employed was developed between the 1940s and 1970s. Much of the history can be distilled from the book of Kraus [Kra83]." and "We mention just a few key references, primarily the book by Kraus [Kra83] , which contains references to much earlier work on the subject."

Sounds interesting. My source for this direction of more recent research is A. Peres, Quantum Theory, Concepts and Methods.

 

[Demystifier]

Eq. (2.160) (which first appeared as Eq. 2.92) does not describe a POVM measurement.

Then what does Eq. (2.160) describe? Perhaps it doesn't describe the POVM measurement itself, but it certainly describes the update of information after the POVM measurement.

Of course there is nothing wrong with instrumentalism as a ‘shut up and calculate’ tactic for evading the conceptual and physical puzzles presented by QM. What is wrong is elevating that evasion to a dogmatic prescription about how to ‘interpret’ QM–which is what we see from many instrumentalists. That’s the only thing I’ve been contesting.

With that, I absolutely agree.

 

[Demystifier]

I don't see that POVMs is something that goes much beyond standard QT, although of course you can take POVMs as the measurement postulate as in the book quoted above

Indeed, today POVM is considered to be a part of standard QT. In fact, POVM measurement of a measured system can be described as a projective measurement of a larger system, which includes not only the measured system but also a part of its "ancilla" environment. From this bigger point of view, all measurement are projective.

 

[gentzen]

Eq. (2.160) (which first appeared as Eq. 2.92) does not describe a POVM measurement.

Then what does Eq. (2.160) describe?

The provided wikipedia link succinctly explains this:

A measurement upon a quantum system will generally bring about a change of the quantum state of that system. Writing a POVM does not provide the complete information necessary to describe this state-change process.  To remedy this, further information is specified by decomposing each POVM element into a product: ...

In my own words: A POVM doesn't include a collapse postulate (a change of the quantum state), but Eq. (2.160) does include one.

 

[Demystifier]

In my own words: A POVM doesn't include a collapse postulate (a change of the quantum state), but Eq. (2.160) does include one.

With that I agree. But my point was to deny a frequent misconception that "a collapse is something that is associated only with projective measurements, not with POVM measurements".

 

[gentzen]

With that I agree. But my point was to deny a frequent misconception that "a collapse is something that is associated only with projective measurements, not with POVM measurements".

Interesting. What exactly is the misconception that you have in mind? You agree that a collapse postulate (a change of the quantum state) is not included in a POVM measurement. I guess you also agree that a typical description of a projective measurement includes a collapse postulate.

Is the "misconception" the believe that there would be some deeper reason (beyond mere historical accident and convention) why descriptions of projective measurements typically include a collapse postulate, and POVM do not?

 

[vanhees71]

Indeed, today POVM is considered to be a part of standard QT. In fact, POVM measurement of a measured system can be described as a projective measurement of a larger system, which includes not only the measured system but also a part of its "ancilla" environment. From this bigger point of view, all measurement are projective.

Sure, that's why I don't think that it changes any of the "fundamental quibbles" some people have, and I'd also take only the part giving the probabilities for the "POVM measurement" result but not the "collapse postulate" since as in the case of the more restricted POV measurements it depends on the specific apparatus used to realize the POVM measurement. So whether or not the measured system is described after the POVM measurement by the assumed "collapsed state", cannot be generally stated either but maybe for some specific POVM measurement it's possible to realize such a specific "state preparation by measurement".

 

[Demystifier]

Interesting. What exactly is the misconception that you have in mind? You agree that a collapse postulate (a change of the quantum state) is not included in a POVM measurement. I guess you also agree a typical description of a projective measurement include a collapse postulate.

It's a misconception to say that one of those includes a collapse postulate and the other doesn't. In a consistent joint treatment of projective and POVM measurements, either both include a collapse postulate or neither does. Unfortunately, some "typical" presentations of projective and POVM measurements look as you described above, which is a misconception.

Is the "misconception" the believe that there would be some deeper reason (beyond mere historical accident and convention) why descriptions of projective measurements typically include a collapse postulate, and POVM do not?

Yes, exactly!

 

[Demystifier]

Sure, that's why I don't think that it changes any of the "fundamental quibbles" some people have

Exactly!

 

[gentzen]

In a consistent joint treatment of projective and POVM measurements, either both include a collapse postulate or neither does.

The only problem is that the definition of POVM explicitly doesn't include a collapse postulate. The general case presented by N&C seems to be called "quantum measurement" by them. Wikipedia claims that also the name quantum instrument would be in use, but the reference section of that article does not convince me. More convincing is Wikipedia's claim that the mathematical formalism is actually called quantum operation, and used to describe the effects of measurement and transient interactions with an environment. The reference section seems to support that claim, and it is also plausible: Why on Earth would anybody want to call every manipulation or preparation of a quantum system a measurement?