I A new realistic stochastic interpretation of Quantum Mechanics

[DrChinese]

1. The final correlated 1&4 pairs are not even defined until Chris has tagged the matches and communicated to Alice and Bob so they can "filter" the random pairs.

The fact that 1&4 has never been in contact, does not matter because it's not how the remotely entangled systems are constructed.

They are constructed from two independently entangled pairs, and whose mutual "correlation" is CREATED from "filtering" based on the tag info from Chris. This is pure information processing, this is why there is no non-nocal action needed. This filtering can be done at any point in time, which is why order does not matter. But what does matter is that that info from swap tagging at Chris, must be available, otherwise one can never identify the entangled remote pairs [as being ψ+ or ψ-].

Sorry, the "filtering hypothesis" is theoretically and factually incorrect, and this has even been experimentally demonstrated as such.

First, all 2 & 3 pairs that fit within the time window of the BSM lead to entanglement of 1 & 4 into one of the 4 Bell states. There is no filtering happening; if there are 2 near-simultaneous clicks at the BSM, there is a swap.

There cannot be a hidden correlation between the 1 & 4 pairs waiting to be revealed, since they are independently prepared. It should be obvious that the 1 & 4 photons - from fully independent sources - yield random outcomes which cannot be independently sorted into buckets using a coding mechanism that does not exactly reveal the same information as Alice and Bob obtain in the first place. In other words: Chris would need to perform the same experiment on 2 & 3 as Alice and Bob do on 1 & 4 to obtain the information you imagine. That isn't being done by Chris. (And in fact there is some polarization information being obtained by the BSM; but since it is always either |HV> or |VH> it is itself useless.) And it wouldn't lead to entanglement of 1 & 4 if that's what Chris did. It would be classical (and couldn't even violate a Bell inequality, since there would be Product State statistics).

Besides, in many of the swapping experiments: the orientations that Alice and Bob are measuring are selected mid-flight. It is done too late for there to be any light speed communication between the various observation stations. See this important swapping experiment in which independent random number generators are used for the Alice and Bob observations.


Second: Were what you said true, then why does indistinguishability even matter? By your concept, all that Chris does with the Bell State Measurement (BSM) is reveal whether we have ψ+ or ψ-. You even say it is pure information processing. Well guess what, you can obtain that exact same information - i.e. the markers indicating ψ+ or ψ- - even without indistinguishability. But... no indistinguishability, no entanglement! The marker for ψ+ is simultaneous clicks on the |H> and |V> detectors by the same output port of the beam splitter portion of the BSM. The marker for ψ- is simultaneous clicks on the |H> and |V> detectors by different output ports of the beam splitter portion of that BSM.

This particular point was analyzed and tested in this swapping experiment. From the paper (and note that this particular experiment looks at the 2 φ Bell states instead of the 2 ψ Bell states):

"One can also choose to introduce distinguishability between the two projected photons. In this case, the phase between the two terms of the |φ> projected state is undefined, resulting in a mixture of |φ+> and |φ−> in the projected state, and the first and last photons do not become quantum entangled but classically correlated. We observed this when we introduced a sufficient temporal delay between the two projected photons (see Fig. 3c). It is also evidence that the first and last photons did not somehow share any entanglement before the projection of the middle photons."

If the BSM is just filtering - and there is no remote physical projection occurring - then this result should not occur.


Third: the "filtering hypothesis" implies that any 2 entangled photons arriving within the coincidence window of the BSM would have their entangled partners containing "hidden" correlations waiting to be revealed. There cannot be sunc hidden maximum entanglement - pre-existing and waiting to be revealed - between 3 photons without violating Monogamy rules. The BSM must be successful as a physical process to cause the entanglement swap.

---

To quote the earlier reference from the team of Kaltenbaek et al:

"A successful entanglement swapping procedure will result in photons 1 and 4 being entangled, although they never interacted with each other. ... We confirm successful entanglement swapping by testing the entanglement of the previously uncorrelated photons 1 and 4."

 

[lodbrok]

Sorry, the "filtering hypothesis" is theoretically and factually incorrect, and this has even been experimentally demonstrated as such.

Not sure where you get that idea from. The experimenters themselves state that filtering is happening.

in https://arxiv.org/ftp/arxiv/papers/1203/1203.4834.pdf
... What, however, is important is to relate the lists of Alice, Bob and Victor’s measurement results. On the basis of Victor’s measurement settings and results, Alice and Bob can group their earlier and locally totally random results into subsets which each have a different meaning and interpretation. This formation of subsets is independent of the temporal order of the measurements. ...

https://journals.aps.org/prl/pdf/10.1103/PhysRevLett.80.3891
Experimental Entanglement Swapping: Entangling Photons That Never Interacted

To verify that this entangled state is obtained, we have to analyze the polarization correlations between photons 1 and 4 conditioned on coincidences between the detectors of the Bell-state analyzer.
...
If entanglement swapping happens, then the twofold coincidences between D1+ and D4, and between D1- and D4, conditioned on the |ψ⟩₂₃ detection, should show two sine curves as a function of Q which are 90± out of phase
...
In that case, one could consider Alice performing the Bell-state measurement on photons 2 and 3, telling Bob, who is in possession of photon 4, the result of the Bell-state measurement.

Experimental delayed-choice entanglement swapping
https://www.nature.com/articles/nphys2294

In our experiment, the primary events are the polarization measurements of photons 1 and 4 by Alice and Bob. They keep their data sets for future evaluation. Each of these data sets by itself and their correlations are completely random and show no structure whatsoever. The other two photons (photons 2 and 3) are delayed until after Alice’s and Bob’s measurements, and sent to Victor for measurement. His measurement then decides the context and determines the interpretation of Alice’s and Bob’s data.
...
According to Victor’s choice of measurement (that is, entangled or separable state) and his results (that is, |φ+〉23, |φ−〉23 or |H H〉23, |V V 〉23), Alice and Bob can sort their already recorded data into 4 subsets. They can now verify that when Victor projected his photons onto an entangled state (|φ+〉23 or |φ^− 〉23), each of their joint subsets behaves as if it consisted of entangled pairs of distant photons. When Victor projected his photons onto a separable state (|H H〉23 or |VV〉23), Alice’s and Bob’s joint subsets behave as if they consisted of separable pairs of photons. In neither case Alice’s and Bob’s photons have communicated or interacted in the past.

 

[DrChinese]

1. The experimenters themselves state that filtering is happening.

1. There are 2 subsets of 4 fold entangled coincidences that are relevant (all within the appropriate time window): ψ+ and ψ-. The only "filtering" is the requirement of 4 fold coincidences. Note that there are 4 Bell states, but only 2 of the 4 can be identified in most experiments. The other 2, that can't, don't result in 4 fold coincidences. If there is 3 fold coincidences (i.e. clicks at detectors 1 and 4, and just *one* at the Bell State Measurement/BSM): there is entanglement and there is a swap, but which of the other 2 Bell states occurred is unknown.

So the only "filtering" is the sorting into the ψ+ bucket and the ψ- bucket. It is true that most detections at the BSM are single detections (i.e. one click within the time window). I guess you could call that "filtering", but it is not as if those meet the BSM criteria.

---

2. Your reference for Zeilinger's group (Ma et al) said:
Experimental delayed-choice entanglement swapping
https://www.nature.com/articles/nphys2294 [This is the arxiv version, no paywall]

"In our experiment, the primary events are the polarization measurements of photons 1 and 4 by Alice and Bob. They keep their data sets for future evaluation. Each of these data sets by itself and their correlations are completely random and show no structure whatsoever. The other two photons (photons 2 and 3) are delayed until after Alice’s and Bob’s measurements, and sent to Victor for measurement. His measurement then decides the context and determines the interpretation of Alice’s and Bob’s data.
...
"According to Victor’s choice of measurement (that is, entangled or separable state) and his results (that is, |φ+〉23, |φ−〉23 or |H H〉23, |V V 〉23), Alice and Bob can sort their already recorded data into 4 subsets. They can now verify that when Victor projected his photons onto an entangled state (|φ+〉23 or |φ^− 〉23), each of their joint subsets behaves as if it consisted of entangled pairs of distant photons. When Victor projected his photons onto a separable state (|H H〉23 or |VV〉23), Alice’s and Bob’s joint subsets behave as if they consisted of separable pairs of photons. In neither case Alice’s and Bob’s photons have communicated or interacted in the past.


2. You are mixing analogy/wording here. This is the delayed choice version, just want to be clear . So in this version, we have 4 fold coincidences exactly as I described in 1. above. Alice and Bob receive their respective photons 1 & 4 before the BSM occurs (this doesn't matter for our purposes).

Yes, it is a true statement - as you highlight - that Alice and Bob's photons - when there is 4 fold coincidence - will form 2x2 subsets for each of 3 possible basis choices that are performed by Alice and Bob: |H>/|V>, |R>/|L>, |+>/|->. They match (correlate) or don't match (anti-correlate) based on whether the entangled state is φ+ or φ-. Or alternately fail to correlate on some bases when there is no swap (per Victor's choices).

These are then related to the following at the BSM, based on the choices made by Victor (see figure 3 at bottom of paper):

Victor allows entanglement swap (to Bell state φ+ or φ-):
H/V basis: Correlation indicating either entanglement or product (separable) state
R/L basis: Expected correlation indicating entanglement
+/- basis: Expected correlation indicating entanglement

Victor prevents entanglement swap (Product state):
H/V basis: Correlation indicating either entanglement or product (separable) state
R/L basis: No correlation, product state
+/- basis: No correlation, product state

In each scenario, since there are 4 fold coincidences/clicks (the stated criteria), the experimenters match the Alice/Bob data with the appropriate result at Victor's BSM and summarize their results. Which are: when swaps succeed, there is the expected correlation; when the swaps are prevented from occurring, the hallmarks of entanglement are not present.

---

Nothing tricky about any of this, normal stuff for any experiment. And none of this is important for explaining how local causal interpretations can explain the basic results. Which are: perfect correlations (as adjusted for Bell state φ+ or φ-). We know the perfectly correlated entangled state results for photons 1 and 4 would be exactly the same if:

a) The sources of the 2 photon pairs are far distant, so that photons 1 and 4 share no backward light cone;
b) The settings for Alice and Bob's measurements are changed mid-flight;
c) The choice of whether to entangle photons 1 and 4 is made after they are detected and recorded; and/or
d) The detection loophole is eliminated.

"Filtering" has nothing to do with any of this, and explains exactly: nothing. And none of this involves Bell's Theorem in any way, so any "hidden" assumptions in Bell are moot. I am asking: How do perfect Entangled State correlations appear for initially uncorrelated photons that have never shared a common past, based on a decision to execute a swap by a distant experimenter?

 

[lodbrok]

Filtering" has nothing to do with any of this, and explains exactly: nothing. And none of this involves Bell's Theorem in any way, so any "hidden" assumptions in Bell are moot. I am asking: How do perfect Entangled State correlations appear for initially uncorrelated photons that have never shared a common past, based on a decision to execute a swap by a distant experimenter?

Filtering plays a huge part as already explained. Until the BSM results are used to condition the 14 analysis, no 14 entanglement can be demonstrated. This is the absolutely necessary filtration step that answers all your questions. Your question at the end has been asked by you and answered many times by different participants here in multiple threads, sometimes in painful detail. Please could we not turn every thread into an argument about an aspect of entanglement swapping that's not even controversial in the literature?

The new interpretation tries to derive the QM equations starting from simpler axioms. That is, an indivisible non-markovian stochastic process. From this it derives all the components. In this sense it presents a different view about what the mathematical entities mean, that's why it is a new interpretation. It does not propose an ontology of what is really happening other than to clarify that the derived mathematical entities are not physically existing objects.

Therefore this discussion about entanglement swapping adds nothing to this thread in my humble opinion.

 

[DrChinese]

1. Filtering plays a huge part as already explained. Until the BSM results are used to condition the 14 analysis, no 14 entanglement can be demonstrated. This is the absolutely necessary filtration step that answers all your questions.

2. The new interpretation tries to derive the QM equations starting from simpler axioms. That is, an indivisible non-markovian stochastic process. From this it derives all the components. In this sense it presents a different view about what the mathematical entities mean, that's why it is a new interpretation. It does not propose an ontology of what is really happening other than to clarify that the derived mathematical entities are not physically existing objects.

3. Therefore this discussion about entanglement swapping adds nothing to this thread in my humble opinion.

1. Yes, it's true that the results of 3 distant observers must be brought together to demonstrate nonlocality. That would always be true in demonstrating quantum nonlocality*. So if that is your "out", it's circular reasoning. There is no meaningful filtering occurring, in the sense that some 4 fold events are ignored. And there is no reference to Bell's Theorem in these experiments to need to get around.


2. The entire point is that the perfect correlations between distant experiments outside of backward light cones cannot be explained by the interpretation which is the subject of this thread. "Not physically existing objects" (your words) as an out? That's it?? They are physically demonstrated effects. As with many interpretations which claim to explain Bell inequalities by redefining language, challenging Bell assumptions, etc., this one appears to claim that a series of local actions can lead to a nonlocal result. Well, how specifically? Big claims are cheap (quote below from his abstract), especially when non-Bell experiments (featuring perfect non-local correlations, as I presented) say otherwise. And it sure sound like new ontology to me, in fact a hidden variables approach (I deduced that from his claim about "hidden-variables").

"...this paper introduces a new principle of causal locality that is intended to improve on Bell's criteria, and shows directly that systems that remain at spacelike separation cannot exert causal influences on each other, according to that new principle. These results therefore lead to a general hidden-variables interpretation of quantum theory that is arguably compatible with causal locality."


3. I agree that I have had my say. Apparently you (as best I can tell) and some others dismiss the many important post-Bell developments in entanglement theory and experiment which invalidate virtually any interpretation holding on to local causality** (or Barandes' "causal locality"***, or almost any "local anything"). I have cited plenty of relevant research, and explained its relevance as a yardstick for new interpretations of QM. I doubt I can anything useful.


*Unless you could perform FTL signaling (which you can't of course)
**As for example described by Bell's 1975 paper. If only he were alive today to see how far things have progressed. "Zeilinger recalls when he met Bell at a conference in Amherst, in 1990, how enthusiastic Bell was about the [GHZ] result." from section 8.3 here. GHZ being another result demonstrating nonlocal "elements of reality" based on the actions of mutually distant experimenters.
***His changing the wording order from "local causality" to "causal locality" makes me . Like that changes everything...

 

[WernerQH]

It is unfortunate that Barandes' "interpretation" does not make explicit what quantum theory is about. A stochastic theory could be a significant improvement over a statistical theory purportedly describing the properties of quantum "particles". It is misleading to think of photons as objects with "properties" (like vertical, diagonal, or circular polarization), because according to the theory those may be completely undefined, and are just as well properties of the detectors, which can detect only particular kinds of polarization. In a recent thread there wasn't even consensus on the conditions under which photons can be called "entangled".

 

[Fra]

I planned to comment more detail but lodbrok responded well on the filtering part so no need to rephrase that. So I will shorten my comments, and focus on what I think the key points and the focus of the thread?

Sorry, the "filtering hypothesis" is theoretically and factually incorrect, and this has even been experimentally demonstrated as such.

First, all 2 & 3 pairs that fit within the time window of the BSM lead to entanglement of 1 & 4 into one of the 4 Bell states. There is no filtering happening; if there are 2 near-simultaneous clicks at the BSM, there is a swap.

As already said in the thread, the full unfiltered data at Alice/1 & Bob/4 is not correlated - unless gated/filtered using mandatory information classically sent from the BSM results. This is the "filtering" I refer to and that is in place.

But I think this is not really what you object to! You just seem to object due to tweaking of wordings, because you seek to UNDERSTAND HOW the correlation is possible (*), if there are no "hidden keys" that is a local causal explanation, that are created during pair production and swapped during swapping?

Do you find that the filtering explanation sort of hints that some kind of hidden key is there but at the same time, you conclude that (in line with assumptions in bells theorem) this can not be right! Thus the explanation must be something else - some nonlocal influence? And this is why you keep objecting to explanations that you feel is in contradiction to bells theorem?

Before I go on to associate to Barande: is this a fair assessment of your perspective or did I miss something else?

(*) And Barandes paper does not explain this in detail.

/Fredrik

 

[Fra]

And there is no reference to Bell's Theorem in these experiments to need to get around.

This is the point.

My view, and I think that is one Barandes keys as well is that the ansatz of Bell simply does not apply to the general case. Barandes puts is so that markov divisibility is wrong in the general case - and them specificially in the case of quantum interactions.

The difficulty is to conceptually understand or motivate this. Barande claims that the assumption does not follow from the general inference rules, and I think this is correct. Bells ansatz is indeed following from classical intuition of "ignorance of the observer". But just becase no such hidden variable model is allowed, does not mean that more general hidden variable models are not, right?

As with many interpretations which claim to explain Bell inequalities by redefining language, challenging Bell assumptions, etc., this one appears to claim that a series of local actions can lead to a nonlocal result. Well, how specifically? Big claims are cheap (quote below from his abstract), especially when non-Bell experiments (featuring perfect non-local correlations, as I presented) say otherwise. And it sure sound like new ontology to me, in fact a hidden variables approach (I deduced that from his claim about "hidden-variables").

I agree this criticism is fair as coming from an opposing perspective. This is exactly what I meant in my first post in the thread.

/Fredrik

 

[Fra]

3. I agree that I have had my say. Apparently you (as best I can tell) and some others dismiss the many important post-Bell developments in entanglement theory and experiment which invalidate virtually any interpretation holding on to local causality** (or Barandes' "causal locality"***, or almost any "local anything"). I have cited plenty of relevant research, and explained its relevance as a yardstick for new interpretations of QM. I doubt I can anything useful.

Given that Barande rejects the markov divisibility (which is the KEY problem in Bell ansatz that is essentially the same as the "physicists ignorance" assumption) then we do NOT have kind of causality that bells theorem applies to, so it's fair to give it a new name. I think it's not just a renaming of the same thing.

So I think noone dismiss bell theorem, nor claim bell was wrong. It's just that not ALL theories that entertain some sort of HV or stochastics to them are of the bell type, and then the theorem does not apply. What is wrong about that?

A problem of the meaning of locality and causation if one is thinking in terms of 3D space + time, or wether than is thinking in terms of an information processing scheme, where 3D is likely ultimately emergent anyway.

Anyone that wants to defend the markov divisibility assumption? It surely applies to a ignorant physicists, but when else? In game theoretic analogy, the ignorant physicists would be like an ignorant "judge", but the alternative is there even the players are ignorant; then their information becomes part of the game, and such simple partitions into hidden keys does not work.

/Fredrik

 

[DrChinese]

1. As already said in the thread, the full unfiltered data at Alice/1 & Bob/4 is not correlated - unless gated/filtered using mandatory information classically sent from the BSM results. This is the "filtering" I refer to and that is in place.

2. ...you seek to UNDERSTAND HOW the correlation is possible (*)... Thus the explanation must be something else - some nonlocal influence? And this is why you keep objecting to explanations that you feel is in contradiction to bells theorem?

3. Given that Barande rejects the markov divisibility (which is the KEY problem in Bell ansatz that is essentially the same as the "physicists ignorance" assumption) then we do NOT have kind of causality that bells theorem applies to

1. Agree, that's true in all swapping setups. The experiment consists of having another distant scientist who makes a decision to entangle - or not - the 1 & 4 photons. Of course, she must his report her results just as Alice and Bob do. I wouldn't call requiring the results of distant observers being brought together for analysis "filtering", when the purpose is to demonstrate quantum nonlocality in the first place.


2. It (the entanglement/correlation) is through a quantum nonlocal mechanism, the nature of which I have no clue except what is learned from experiments and standard theory. What I am asking is for those who come up with new interpretations to explain these experiments. Yes, I of course agree with the usual results of Bell's Theorem - and I disagree with attempts by some interpretations to evade Bell.

This is why I am reverting back to experiments that do not require a Bell-like approach - because those can't be evaded by the same hand-waving. Entanglement swapping and GHZ being 2 alternate approaches that demonstrate nonlocal "elements of reality" (by Einstein's definition). No probability involved, a certain prediction of distant outcomes based on the decisions of remote observers. (Einstein would have been surprised at these experiments, as they directly demonstrate the remote steering of reality using the same basic quantum mechanical theory he was familiar with - which he thought must be incomplete.)


3. I have never (in hundreds of papers on the subject) seen "Markov divisibility" seen as relating in any manner whatsoever to Bell. Where did Bell touch on anything close to that in his 1964 paper?

If someone has a different-than-Bell usage of "locality" and/or "causality", considers that an "out" from the Bell conclusion, and then decide that explains QM via local hidden variables... well, I'm not trying to debate that either way here.

 

[Fra]

3. I have never (in hundreds of papers on the subject) seen "Markov divisibility" seen as relating in any manner whatsoever to Bell. Where did Bell touch on anything close to that in his 1964 paper?

That term, is from Barandes... You are right of course Bell did not use that term, neither do I, but however we label it, it is IMO one of two key assumptions going into Bell's ansatz. I mean this ansatz in Bells's paper

-- https://cds.cern.ch/record/111654/files/vol1p195-200_001.pdf

IMO, if you elavorate Bells expression involving conditional probabilities it contains TWO assumptions,
1) statistical independence of A and B (this is fine)
2) What I've called the equipartition assumption in terms of the hidden variable, in the lack of better world, but Baranders calls it "Markov divisibility". (this is not so fine)

-- Jacob Barandes - "A New Formulation of Quantum Theory"

I think both these assumptions are FINE in the original context however, and in the context of looking for an actual classical "ignorance interpretation" which would suggest QM beeing incomplete. But from my and apparently others, interpretational stance, this is far from obvious. In particular, in the agent interpretations, I personally find this assumption is not even plausible. (This is my personal opinon, and why i highlight this, as I am happy to see others point the finger on the exact same spot, but perhaps from a different angle because I admit I do not see Barandes "big plans" if they exist).

/Fredrik

 

[iste]

From this it derives all the components. In this sense it presents a different view about what the mathematical entities mean, that's why it is a new interpretation. It does not propose an ontology of what is really happening other than to clarify that the derived mathematical entities are not physically existing objects.

It is unfortunate that Barandes' "interpretation" does not make explicit what quantum theory is about.

...

Well apart from suggesting that much of the quantum objects are not physical objects, it also gives actual, definite outcomes at every moment in time. It does this without needing to add anything new since the core of the formulation is just to show that quantum systems are equivalent to certain kinds of stochastic ones. But yes, apart from that, it doesn't actually say exactly how or why the stochastic system produces quantum phenomena beyond his assumptions about the structure of a generalized stochastic system characterized by a "linear marginalization condition" and non-markovianity. But from an online presentation he gave on it, he views this formulation as giving a kind of "physical picture" of quantum mechanics.

Looking at the last Barandes paper specifically on locality a little more closely, I think maybe his criteria for locality here is maybe a bit weak imo; and at face value, actually I am not entirely sure entanglement swapping would strictly meet the criteria set for locality at the end of the paper so I guess Mr. Chinese has a point there.

That said, I still like the view that these quantum non-local correlations don't necessarily need to be more than a very unintuitive statistical effect. I mean, it's interesting that you can have what seem like relatively minimal assumptions of a non-Markovian stochastic system and get non-local correlations without needing to specify further underlying mechanisms. The non-markovianity seems to be the big part of what matters; and even though I imagine what Mr. Barandes has presented so far probably doesn't have perfect non-local correlations like Mr. Chinese talks about, it still seems pretty explanatorily significant that you can get these illicit non-local correlations in a manner that looks just like a generic quantum entanglement from a kind of basic stochastic system.

As I have said before, the non-markovianity condition looks just like the kind of violation of total probability which you get for incompatible observables and at the center of contextual phenomena and Bell/Boole violations. It relates to interference terms in same way as total probability violations do in standard quantum mechanics. Specifically the condition would correspond to total probability violations for the trajectories in the path integral formulation (square of sums is not sum of squares); the definite outcomes in this formulation then correspond to actual realizations of the paths which naturally should then sample all possible trajectories if you keep repeating an experimental scenario indefinitely. Could you also get the perfect Bell correlations of quantum mechanics through this formulation? Presumably, if Mr. Barandes' formulation is sound then you should be able to just express those Bell scenarios, or at least maybe ones that are sufficiently analogous, in terms of generalized stochastic systems by using Mr. Barandes' dictionary; he does say in one paper you can build spin into the model. If strange interference as a consequence of phase shows up then presumably the perfect correlations should also show up too under some particular construction of a stochastic system... if the quantum constructs required to produce the Bell correlations can be, effectively, translated into stochastic form? Obviously we haven't seen this yet - I guess its an open question, the state of this formulation further on down the road. But since this formulation is just about the bi-directional "dictionary" translation between quantum and stochastic systems then what I have said should be the case if Barandes' formulation is sound... right?

I think it is not about getting rid of non-local correlations, but making them non-spooky. Unintuitive consequences of non-commutativity, implemented through local entangling interactions, local measurement interactions. The attraction of this kind of formulation maybe is that, just like some might give the challenge of "How do you produce this strange correlation in this scenario?", I can see someone giving the challenge of asking how/why would you inject strange ontologies into it? Because my impressions is it naturally gives weird quantum behaviors from a very sparse, unextravagant kind of foundation (there is no physical collapse either - it appears in the formulation purely as statistical conditioning). It makes it look like strange quantum phenomena don't need to be explained by some novel metaphysics (apart from natural stochastic behavior) but as very unintuitive consequences that just seem to emerge from certain kinds of statistical structures because the kind of constraints you would normally expect have been lifted. I think maybe I can see why Barandes' is pushing for this to be seen as a "causally locally" hidden variable model; it looks nominally local because the components are so unextravagant until you inspect the behavior in certain situations. Maybe quantum mechanics is generally like that but the quantum formalism makes it seem much more mysterious and open to interpretation compared to formulating it as just stochastic processes. The need for excessively novel metaphysics then seems deflated.

 

[WernerQH]

Well apart from suggesting that much of the quantum objects are not physical objects, it also gives actual, definite outcomes at every moment in time.

Does it? This needs to be substantiated. What "definite outcomes" are you talking about?

It does this without needing to add anything new since the core of the formulation is just to show that quantum systems are equivalent to certain kinds of stochastic ones.

I like the view that quantum theory is basically a stochastic theory, but would like to hear more details about those "certain kinds of stochastic ones".

Barandes wrote:

Seen from another point of view, this stochastic-quantum correspondence yields an alternative way to formulate quantum theory, in the language of trajectories unfolding stochastically in configuration spaces.

In my opinion already the concept of a trajectory is deeply flawed.

 

[DrChinese]

Mr. Chinese

It's Mr. DrChinese. Note that I am not a doctor of any kind that involves a degree. My musician friends simply call me Doc, so that's good as well. We're all friends here.

Oh, and I am not Chinese (according to 23 and me).

 

[Fra]

As I have said before, the non-markovianity condition looks just like the kind of violation of total probability which you get for incompatible observables

Yes I think it's related...

1) Different observers generally have different incompatible information, this may sometimes be cured by transformations that defines the communication, and this is then related to interaction terms. When no transformations exists, perhaps one can consider the symmetries to be emergent in some way.

2) Any one observer can choose an observational basis or also choose partial information in non-commutative observables; this generally leads to a balance or uncertainty relation between them for exampele conjugate variables.

if the quantum constructs required to produce the Bell correlations can be, effectively, translated into stochastic form? Obviously we haven't seen this yet - I guess its an open question, the state of this formulation further on down the road. But since this formulation is just about the bi-directional "dictionary" translation between quantum and stochastic systems then what I have said should be the case if Barandes' formulation is sound... right?

Yes I think these are some hard details missing in the puzzle.

Details aside, the vision I have that may be in line with some of the details in barandes ideas, that relate to stochastics is essential stochastic decision making - based on non-commutative and contextual information/observables. This has the potential to be extremely beautiful but to do from the concept to explicit mathematical models that are computable and not just a symbolic or formal expressions is I think extremely difficult. In essence my expectartion is that it would reduce ANY hamiltonian to "stochastics" of all interactions, and in this sense also add explanatory power to the nature of causation. Something that we do not have today! The is the main different from "rational action" in old game theory, because the problem is to DEFINE what is "rational". But what if its simply stochastic, and the rest is self organisation?

Markov processes in statitic state spaces here are clearly too simple.

/Fredrik

 

[iste]

Does it? This needs to be substantiated. What "definite outcomes" are you talking about?

Sorry, late and long reply.

Well if we look at a stochastic process, we may have basically a sequence of random variables across time defined on probability spaces. These random variables are not physical events but just provide predictive probabilities. We can then sample the variables, where they will only realize one outcome at a time. For instance, our random variable could give probabilities for a dice roll, and then we can sample the variable by rolling the dice which gives a single outcome at any time - a single physical event.

As another example, the variables could be referring to particle positions (like a dust particle floating in a glass of water) and if we take a single sample from a sequence of variables over time we get a single trajectory of where the dust particle moves over time in the glass of water, or something comparable to that. If we keep on sampling the sequence of variables indefinitely (i.e. repeating the experiment), we get many many trajectories whose frequencies should approach the variable probabilities as a consequence of the law of large numbers.

Now, at the center of Mr. Barandes' formulation is a dictionary which effectively translates from a generalized stochastic system' random variables into a unitary quantum system. It goes both ways so we can start from a quantum system and translate it to variables of a stochastic process. From this latter perspective, we note that there is nothing about the initial quantum system that necessarily leads to the notion of having these definite outcomes - on the contrary, most think this difficult to imagine, as you suggest further down in your comment. But then it is translated to the stochastic system which can be sampled for definite outcomes like you would for rolling a dice.

Why this is possible is that the sampled outcomes are something in addition to the objects of quantum mechanics so they do not contradict or change standard quantum mechanics in anyway. Without the assumption that quantum objects correspond to information about random variables, there is no reason for anyone to assume these sampled outcomes actually can exist like they commonsensically do for stochastic processes. But if quantum systems are equivalent to stochastic systems, then they are formally equivalent to systems which can produce those definite outcomes at every point in time. In this formulation, the quantum objects only correspond to (only translate to) information about the random variables in a stochastic process; they don't strictly correspond to any kind of physical object, event or outcome. They only carry information about statistics and so, according to Mr. Barandes' formulation, what has been interpreted in quantum mechanics as describing definite collapsed eigenstates or indefinite superpositions was never really about physical events or outcomes at all. Definite outcomes at every point in time can be sampled from coherent superpositions and decoherent mixtures; wave-function collapse corresponds purely to statistical conditioning without any physical consequence. Its worth noting that while I have said that there is no necessary reason to believe these stochastic outcomes exist from just looking at regular quantum formalism, they are arguably hinted at in the path integral formulation. Mr. Barandes' formulation is set up similarly to the path integral view where we are talking about how a particle's configuration evolves between two time points. The sampled trajectories between the two points in this formulation then correspond to the paths whose amplitudes are summed over in the path integral formulation. Naturally you can imagine that with its random behavior, repeating an experimental scenario indefinitely would sample all possible trajectories.

but would like to hear more details about those "certain kinds of stochastic ones".

The two papers in the beginning of the thread go into detail. But basically, we have this linear marginalization condition which allows us to calculate probabilities for particles being in configurations at one time point using an initial reference time point and transition probabilities between the two points.

What marks out the certain kind of stochastic process is that when we come to the issue of characterizing joint probabilities for specific trajectories of intermediate configurations between the two time points (dividing a trajectory into sub-trajectories e.g. like A-D into A-B-C-D with distinct transition probabilities), we just cannot do this - the system is indivisible. That is probably the biggest component that gives these stochastic systems quantum behavior. This is exactly the way in which you yourself said that the concept of the trajectory was flawed. These systems cannot give well-defined joint probabilities for the intermediate trajectory between two points even though, with some initial reference time point, we can define probabilities for any arbitrary time point. So we do have random variables with outcomes that can be sampled at every time point when starting from an initial time point, and form a trajectory of actual definite outcomes when you run the model or simulation or whatever. What is not well-defined is the (Markovian) joint probabilities when asked to specify intermediate parts to (the trajectory inbetween) the initial and final time points. So even though there are definite outcomes, it still has the same flawed trajectory issue as in quantum mechanics. The translation from stochastic to quantum is actually helpful here because you can effectively translate transition probabilities to what would more or less correspond to the transition amplitudes of the path integral formulation, and these are divisible (and therefore paths can be summed over to find them).

 

[iste]

It's Mr. DrChinese. Note that I am not a doctor of any kind that involves a degree. My musician friends simply call me Doc, so that's good as well. We're all friends here.

Oh, and I am not Chinese (according to 23 and me).

Oh I beg your pardon! I have been reading your name and then mentally referring to it as Mr. Chinese all this time, ha.

1) ... 2)

You might find this interesting:

https://royalsocietypublishing.org/doi/full/10.1098/rsta.2019.0036

Relating to your thoughts on observers, incompatible information, contextual.

You may also find this next paper interesting:

https://scholar.google.co.uk/scholar?cluster=10954599080507512058&hl=en&as_sdt=0,5&as_vis=1

(and maybe this https://www.sciencedirect.com/science/article/pii/S037015732300203X )

Its not quantum mechanics but from computational neuroscience but chimes with your ideas maybe. The paper essentially is generalizing a theory the "free energy" principle from biology to all things. The free energy principle is about biological life in terms of bayesian inference which can be framed as observers with beliefs and is a very big theory in neuroscience. The generalization to all things seems to basically originate from this paper as far as I see:

https://scholar.google.co.uk/scholar?cluster=17970774975628711245&hl=en&as_sdt=0,5&as_vis=1

So stochastic dynamics can be linked to the variational free energy of the theory. The author of the free energy paper is therefore in some ways linking stochastic dynamics together with ideas about inference and "beliefs" in a very general way. In some sections they go through formulations of different areas of physics (because they can be related to stochastic dynamics) including a section where they derive bits of quantum mechanics based on stochastic dynamics and complex roots which isn't the same at all but not exactly a trillion miles off conceptually from the Barandes papers. So I just thought you might find that interesting if you haven't already read about it, seems in the kind of direction of your ideas.

 

[WernerQH]

[...] the sampled outcomes are something in addition to the objects of quantum mechanics so they do not contradict or change standard quantum mechanics in anyway.

But if quantum systems are equivalent to stochastic systems, then they are formally equivalent to systems which can produce those definite outcomes at every point in time. In this formulation, the quantum objects only correspond to (only translate to) information about the random variables in a stochastic process; they don't strictly correspond to any kind of physical object, event or outcome.

This looks like the opposite of an interpretation. Barandes is hinting at a larger mathematical apparatus behind quantum theory, instead of mapping the elements of the theory to the real world. As for the stochastic variables purportedly having definite values at all times, it would help to see an application to a concrete physical example. Is "sampling" a mathematical or a physical concept? If it refers to "measurement", does the new formulation shed any light on which interactions count as measurements?

 

[Fra]

You might find this interesting:

https://royalsocietypublishing.org/doi/full/10.1098/rsta.2019.0036

Relating to your thoughts on observers, incompatible information, contextual.

You may also find this next paper interesting:

https://scholar.google.co.uk/scholar?cluster=10954599080507512058&hl=en&as_sdt=0,5&as_vis=1

(and maybe this https://www.sciencedirect.com/science/article/pii/S037015732300203X )

Its not quantum mechanics but from computational neuroscience but chimes with your ideas maybe. The paper essentially is generalizing a theory the "free energy" principle from biology to all things. The free energy principle is about biological life in terms of bayesian inference which can be framed as observers with beliefs and is a very big theory in neuroscience. The generalization to all things seems to basically originate from this paper as far as I see:

https://scholar.google.co.uk/scholar?cluster=17970774975628711245&hl=en&as_sdt=0,5&as_vis=1

So stochastic dynamics can be linked to the variational free energy of the theory. The author of the free energy paper is therefore in some ways linking stochastic dynamics together with ideas about inference and "beliefs" in a very general way. In some sections they go through formulations of different areas of physics (because they can be related to stochastic dynamics) including a section where they derive bits of quantum mechanics based on stochastic dynamics and complex roots which isn't the same at all but not exactly a trillion miles off conceptually from the Barandes papers. So I just thought you might find that interesting if you haven't already read about it, seems in the kind of direction of your ideas.

The conceptual idea in several your links are indeed just in line with what I mention! although there are many deep and difficult open problems with the approaches, so tricky that I think many are rejected by the ideas.

Some quotes from the references illustrating the conceptual hooks...

"these peculiar features of quantum theory are mathematically equivalent to a general notion of disagreement between information sources"

"tenet of the FEP is that everything must provide an accurate account of things that is as simple as possible—including itself"

"These mechanics have the same starting point as quantum, statistical and classical mechanics. The only difference is that careful attention is paid to the way that the internal states of something couple to its external states."

"“what must things do, in order to exist?” The FEP turns this question on its head and asks: “if things exist, what must they do?”

The challenge is to dress this in a computable model, map it's phenomenology to the standard model. The above quotes also explicitly relate to the fine tuning problems we are painfully aware of. The third point also emphasises that these idea are NOT "simple entropic reasoning"... because there is no fixed context, the context is simply the systems environment, this is the external/internal coupling! Basically relating internal and external complexions. This conceptually also associates to dualities.

/Fredrik

 

[iste]

This looks like the opposite of an interpretation. Barandes is hinting at a larger mathematical apparatus behind quantum theory, instead of mapping the elements of the theory to the real world. As for the stochastic variables purportedly having definite values at all times, it would help to see an application to a concrete physical example. Is "sampling" a mathematical or a physical concept? If it refers to "measurement", does the new formulation shed any light on which interactions count as measurements?

Well strictly it is a formulation of quantum mechanics, not an interpretation, based around showing that quantum mechanics is equivalent to, and so can be formulated as, generalized stochastic systems. In one, maybe more, of the papers, Barandes refers to the generalized stochastic system that quantum systems are equivalent to as a hidden-variable theory. But I think the interpretation or "physical picture" (in Barandes' words) follows because if the stochastic system generates definite outcomes, it is very hard not to interpret this just as particles always being in one definite place or configuration at a time and moving around under some random influence. Its only natural to interpret the outcomes of stochastic processes or random variables as physical events - e.g. the outcome of a dice roll is a physical event.

Obviously, this is all just a mathematical formalism, so the argument would be something like - if we can formulate quantum mechanics successfully in terms of particles being "classical-looking" then this seems like the most parsimonious way to interpret quantum mechanics because that is what we expected reality to look like before all the quantum confusion turned up.

Is "sampling" a mathematical or a physical concept?

Its a mathematical one in statistics. Like a statistician might sample from a model in the sense of looking at outcomes it generates. The outcomes we can see as representing physical events. Because the events occur randomly, the probabilities that predict the occurence of the events cannot be empirically observed unless you observe outcomes of random variables over and over again (i.e. repeat an experiment over and over again) and examine the frequencies.

I am not sure if the papers show concrete examples in the way you want though they do give quite detailed descriptions of all thr components that go into the model. I think the thing is that even though the definite outcomes are mentioned throughout the papers, they don't do any of the actual heavy lifting of describing behavior so they aren't really described all that much compared to probabilities, random variables, transition matrices and the quantum objects, etc. etc. He does refer to the double slit experiment in one of the papers (but not concretely) which I guess is a familiar example - the particle really does go through one slit at a time and cause the interference pattern through the gradual build up of individual outcomes on the screen under the stochastic picture.

If it refers to "measurement", does the new formulation shed any light on which interactions count as measurements?

Yes, measurement is in the theory and it is modelled by just explicitly including the measurement apparatus in the model as its own kind of stochastic system - it is just another physical interaction in the theory, thats it. The only special properties measurement has comes from basically how we design measurement devices. Decoherence naturally occurs from interactions with indivisible stochastic systems. Wave-function collapse just appears as statistical conditioning.

 

[iste]

The conceptual idea in several your links are indeed just in line with what I mention!

I'm glad they resonated with you!

 

[Fra]

Is "sampling" a mathematical or a physical concept? If it refers to "measurement", does the new formulation shed any light on which interactions count as measurements?

This is a good question, and while it is a mathematical concept in the abstractions, just like there are different "interpretations" of probability, that must somehow be reflected in physical view of "sampling".

In a pure minimal statistical interpretation, like the typical one for QM, the "sampling" is defined from the perspective of a measurement devices that collectes statistics by repeating experiments, defined by preparation procedures.

But such as "interpretation" of sampling makes not sense in a gaming perspective or agent interpretation, for several reasons.

I think interactions and measurements must be unified, and the difference between a measurement and an interaction is a matter of perspective. Typically the inside perspective of the "observer/agent" is measurements and actions, the perspective of a third observer observing the "first agent measuring a system" will typically describe it not as a measurement, but as an interaction.

The challenge is to unifify the two views in the sense to revealing the relation between the dual views of measurement and interactions. And this is where the conceptual view on "quantum interactions" as disagreement between information sources/agents becomes explicit. Ie the "explanation of interactions" is likely simple, in the dual view, then you can work there, and then transform back! Simple trick that is a generic one also in adS/CFT or more general dualities, gravity etc. I think they keys in barandes paper is steps in the direction, but there is alot more do be clarified.

/Fredrik

 

[WernerQH]

if we can formulate quantum mechanics successfully in terms of particles being "classical-looking" then this seems like the most parsimonious way to interpret quantum mechanics because that is what we expected reality to look like before all the quantum confusion turned up.

Bohmian philosophy. My expectations are quite different. From what we have learned in the last century it seems evident to me that quantum theory is not about "particles". The awkwardness of quanta is rooted in our desire to describe the world around us as composed of objects. As it turned out these quantum objects have properties that are blurred or even undefined (but somehow "entangled") and become manifest only upon "measurement", something the theory leaves undefined. It is ironic that our most successful tool for describing the microworld, Feynman diagrams, pictures it in terms of particles. But, as some members here on PF never tire to emphasize, one should not think of the lines in Feynman diagrams as representing real particles. It is a strange concept that these "particles" are identical. There is no fact of the matter that "this" photon interacted with "that" electron. The Feynman rules tell us to connect the vertices in all possible ways, multiply the propagators, and add up the resulting expressions. A stochastic interpretation of quantum theory should treat the interaction events as real, and not the lines connecting them. I see QFT as a machinery for calculating the correlations between events (points) in spacetime, what mathematicians would call a point process or point field. I think it's a hopeless task of attributing (hidden) stochastic variables to electrons and photons to produce a theory equivalent to QED, but still having a "classical" flavor.

it is very hard not to interpret this just as particles always being in one definite place or configuration at a time and moving around under some random influence. Its only natural to interpret the outcomes of stochastic processes or random variables as physical events - e.g. the outcome of a dice roll is a physical event.

I find it hard to believe in particles, and I have a much more primitive idea of physical event. For example, the emission of a photon. Actually, the emission of a photon does not occur in an instant -- the event is itself composed of two elementary (more "primitive") events separated in time. This leaves enough room for non-Markovian indivisibility.

 

[PeterDonis]

I find it hard to believe in particles, and I have a much more primitive idea of physical event. For example, the emission of a photon.

You're contradicting yourself. A "photon" is a particle, and "emission of a photon" is a particle process.

Actually, the emission of a photon does not occur in an instant -- the event is itself composed of two elementary (more "primitive") events separated in time.

Where are you getting this from? Do you have a reference?

In quantum field theory, "emission of a photon" is a mathematical artifact and has no physical meaning.

 

[iste]

Bohmian philosophy. My expectations are quite different. From what we have learned in the last century it seems evident to me that quantum theory is not about "particles".

That's fair enough if thats your inclination. I can only say that Barandes' formulation would actually agree with the notion that quantum mechanics isn't strictly because the objects of quantum objects are translated only from the stochastic random variables and probability spaces so these are at least implicitly are about the long-run statistics of particles, not individual particles and physical events themselves. It's no contradiction that the quantum objects can then have "properties that are blurred or even undefined", because statistics is about uncertainty. Barandes also says the formulation is general enough to be applicable to quantum field theory. So I don't think there are necessarily any contradictions here either when broadening or deepening the kinds of objects or ontologies being talked about.

 

[Fra]

Bohmian philosophy. My expectations are quite different. From what we have learned in the last century it seems evident to me that quantum theory is not about "particles". The awkwardness of quanta is rooted in our desire to describe the world around us as composed of objects. As it turned out these quantum objects have properties that are blurred or even undefined (but somehow "entangled") and become manifest only upon "measurement", something the theory leaves undefined.

I don't agree the essence of the ideas are traditional Bohmian philosophy. I do however see a conceptual relation to the non-traditional "solipsist hidden variables", even thought that is only conceptually, the actual theory is still missing. But as I see it, agent-defined "solipsist" hidden variables, would be expected to violate the non-markovian divisiblity.

Why? IMO, it has not so much to do with what a particle "is", wether it's a point like bullet, a blob or entangled parts, the essential part which i think is at the heart of that assumption in bells theorem is how these system "particles, or blob or whatever" are INTERACTING.

I think the most problematic preconception from classical physics, is not that we want to imagine that something is definite, but how to describe the dynamics and interactions of interacting parts. How can we possible "understand" interactions that emerge at low energies, from the high energy picture? Without running into extreme fine tuning?

There are two commen ways to model, system dynamics (basically differential equations), and by nature they tend to work best on markovian systems, as the "differential change" typically depend on the current state only. That's how it works. This is why such "models" have great difficulty to model emergent phenomena, and thus you just get "effective models". This is partly I think emphasies by CHOOSING to focus on "system dynamics". That itself puts form-constraints on the theory.

Here ABM (agent based models) have sometimes and edge to model non-markovian phenomena, with emergence. As this more naturally can encode the causal mechanisms of the emergence.

This is now new, but it's not as common in particle physics simply because we do not have an a theory of interactions in and inside perspective. This is why examples are more abundant in comlpex systems such as social and human interactions. I think we can learn from it, and gain insight into modelling emergence also in physics.

To agent based models and system dyamics are complementary. Systems dynamics typically always yield at static state space and timeless law (which is also what Smoling object to in evolution of law, this is related to the same topic imo)

Random paper relating to this just as generic reference (nothing of this is new, but i just suggest this have the same abstractions as exists in physics, and is part of the interpretations)

Agent-based modeling: Methods and techniques for simulating human systems​

"One may want to use ABM when there is potential for emergent phenomena, i.e., when:

• Individual behavior is nonlinear and can be characterized by thresholds, if-then rules, or nonlinear coupling. Describing discontinuity in individual behavior is difficult with differential equations.

• Individual behavior exhibits memory(*), path-dependence, and hysteresis, non-markovian behavior, or temporal correlations, including learning and adaptation."

- https://www.pnas.org/doi/10.1073/pnas.082080899

These things are exactly why the key assumption in Bells theorem is not applicable imo, and this is also the key in barandes ideas. (the ontop of this there are lots more details you could argue about, but it gets opaque to discuss every thing at every level at once)

So I suggest the "stochastic modelling" should be applied NOT to system dynamics but to corresponding agent based model. That might work, but we still have to see the full theory.

Edit: (*) Before someone jumps into conclusions that particles must have brains, the idea here would be that the "memory" is not encoded in the STATE, but in the stateSPACE - which is presumable EMERGENT. So no human observers are required here. That is the persistent misinterpretation that tends to never die. And in terms of System dynamics, this is then "solved" by increasing the dimenstionality of the dependent variables, making it "higher order" etc. But all that does, is increasing complexity to the point where suddently you have a terrible fine tuning problem, and you get only an "effective theory" still, without much explanatory value. Another association to "memory" is IMO weakly related to underlying ideas in this paper from Smolin https://arxiv.org/abs/1205.3707, what he calls "principle of precedence".

/Fredrik

 

[WernerQH]

It's no contradiction that the quantum objects can then have "properties that are blurred or even undefined", because statistics is about uncertainty.

Saying that quantum (field) theory is about statistics is not completely wrong. But it leaves many questions unanswered.

 

[iste]

that quantum mechanics isn't strictly because the objects of quantum objects are translated only from the stochastic

Uhh just noticed this mistake. Should read:

Isn't strictly about particles because the objects of quantum mechanics are translated only from the stochastic

Saying that quantum (field) theory is about statistics is not completely wrong. But it leaves many questions unanswered.

Yes, fair!

 

[WernerQH]

Photons do not exist -- emission of radiation as a mathematical artifact:

In quantum field theory, "emission of a photon" is a mathematical artifact and has no physical meaning.

Obviously light quanta are still controversial.

 

[collinsmark]

Photons do not exist -- emission of radiation as a mathematical artifact:

In quantum field theory, "emission of a photon" is a mathematical artifact and has no physical meaning.

Obviously light quanta are still controversial.

Actually, I never really thought about this until now. But I think the key word here is "emission." I don't think anybody was debating the existence of photons.

Sure, it's easy enough to measure the reception of a photon (well, easier perhaps), but how do you measure/observe the emission of a photon without inferring it via mathematical artifacts?