Understanding Barandes' microscopic theory of causality

[pines-demon]

In this video, at 1:04:00, Dustin Lazarovici asks Barandes if his principle of "causal locality" corresponds to what is usually called non-signaling, and Barandes answers affirmatively, pointing out that the proof of his principle of causal locality is very similar to that of the non-communication theorem, but that the difference lies in the fact that, in his formalism/interpretation, non-communication does not apply only to measurements, but to any physical system, at any time, according to the assumed ontology.

Lucas.

Sure but we already know quantum mechanics, in any of its interpretation does not violate non-signaling so if by causal locality he means that then it is pointless definition.

 

[pines-demon]

Following on from what I mentioned in the previous post, these are two different concepts. In the case of QM, we know that it is causally local, according to Barandes' stochastic-quantum theorem, but it is not locally causal, according to Bell's theorem.

Lucas.

Again showing this is a play on words and definitions and not really explaining anything.

 

[javisot]

Again showing this is a play on words and definitions and not really explaining anything.

I don't know what to say... based on Barandés's words, I'd say his interpretation doesn't correspond to the textbook. Since its correspondence excludes any result that, in a standard way, implies non-locality.

In the terms you used, Pines Demon, local causality is not causally local.

It seems that for Barandés, the violation of Bell's inequality does not demonstrate "spooky action at a distance" in a relativistic causal sense, but rather the failure of a classical conception of causality applied to quantum phenomena.

 

[pines-demon]

I don't know what to say... based on Barandés's words, I'd say his interpretation doesn't correspond to the textbook. Since its correspondence excludes any result that, in a standard way, implies non-locality.

In the terms you used, Pines Demon, local causality is not causally local.

It seems that for Barandés, the violation of Bell's inequality does not demonstrate "spooky action at a distance" in a relativistic causal sense, but rather the failure of a classical conception of causality applied to quantum phenomena.

As you say, he is saying that there is nothing "spooky" (I disagree). He is trying to focus the conversation on his redefinition of causality. My problem is that all that he has shown is already known. I don't get what is interesting of his causal locality. The real deal is Bell theorem, which discusses Bell's local causality and that has not necessarily to be broken (super-determinism and many-worlds try to get away by dropping other assumptions). If Barandes is just dropping Bell's local causality then his interpretation is not necessarily interesting to understand entanglement specifically (and thus for me it is still "spooky").

 

[PeterDonis]

It seems that for Barandés, the violation of Bell's inequality does not demonstrate "spooky action at a distance" in a relativistic causal sense, but rather the failure of a classical conception of causality applied to quantum phenomena.

But he doesn't give any new "conception of causality" to replace the classical one. Perhaps he thinks that isn't required. But that doesn't help at all if one thinks (as many do) that that is required.

 

[javisot]

As you say, he is saying that there is nothing "spooky" (I disagree). He is trying to focus the conversation on his redefinition of causality. My problem is that all that he has shown is already known. I don't get what is interesting of his causal locality. The real deal is Bell theorem, which discusses Bell's local causality and that has not necessarily to be broken (super-determinism and many-worlds try to get away by dropping other assumptions). If Barandes is just dropping Bell's local causality then his interpretation is not necessarily interesting to understand entanglement specifically (and thus for me it is still "spooky").

But he doesn't give any new "conception of causality" to replace the classical one. Perhaps he thinks that isn't required. But that doesn't help at all if one thinks (as many do) that that is required.

Completely agree with both.

 

[Fra]

But at least he attempts to give us a new handle: this is where i see a progressed stance, althogh without full solutions.

IMO the missing link to make his new conception of causal microphysical law (transition probabilities) is their construction/origin/emergence(*). IF one consider them global constraints (that are effecitvely inferred from QM, hilberspace and hamiltionians) via the stochastic correspondence, then yes it really has not explain anything.

The new thing is that instead of having to explain hamiltonians and hilber spaces, we can try to explain the configuration spaces and the transition probabilities of the stochastic process.

If one finds the new handle adding no new intuition, then I understand the critique. But I see it does add intuition that mates my thinking better than the causal notion from system dynamics. Barandes replaces it with microcausal transition probabilites. Ie changing global constraints to local rules. The only problem with Barandes exposition is that the local rules are not explained yet by first principles, they are only shown to follow from correspondence.

(*) This is I think a much harder problem, on par with explaning the structure the phenomenology of the standard model - without fine tuning.

/Fredrik

 

[iste]

I don't get what is interesting of his causal locality

Its informative in one rather niche sense that he views his theory as a hidden-variable interpretation. The two major hiden variable theories I am aware of, Bohmian mechanics and the easily confused stochastic mechanics both violate his causal locality, or perhaps some close analog, so it is noteworthy from that standpoint, especially when compared to stochastic mechanics which is in some sense a competitor against his as a stochastic interpretation of quantum mechanics.

Edit: I should probably remove Bohmian mechanics because its deterministic so I guess it doesn't really fit in the kind of stochastic causal locality Barandes is talking about it but my impulse to put it in was because stochastic mechanics and Bohmian mechanics are nonlocal for more-or-les sthe same reasons, and ofcourse because Bohmian mechanics is more popular hidden variable theory by far.

At the same time, I don't think Barandes theory is a true hidden variable theory to the same extent those ones are.

 

[javisot]

Its informative in one rather niche sense that he views his theory as a hidden-variable interpretation. The two major hiden variable theories I am aware of, Bohmian mechanics and the easily confused stochastic mechanics both violate his causal locality, or perhaps some close analog, so it is noteworthy from that standpoint, especially when compared to stochastic mechanics which is in some sense a competitor against his as a stochastic interpretation of quantum mechanics.

Edit: I should probably remove Bohmian mechanics because its deterministic so I guess it doesn't really fit in the kind of stochastic causal locality Barandes is talking about it but my impulse to put it in was because stochastic mechanics and Bohmian mechanics are nonlocal for more-or-les sthe same reasons, and ofcourse because Bohmian mechanics is more popular hidden variable theory by far.

At the same time, I don't think Barandes theory is a true hidden variable theory to the same extent those ones are.

Personally, and for the time being, I would characterize Barandés' interpretation as a middle ground between objective collapse and the standard interpretations. It may well explain more results than those interpretations that include objective collapse, but I don't believe it reaches the level of a standard interpretation.

 

[pines-demon]

The two major hiden variable theories I am aware of, Bohmian mechanics and the easily confused stochastic mechanics both violate his causal locality

Where does he claim that?

 

[Morbert]

Sure but we already know quantum mechanics, in any of its interpretation does not violate non-signaling so if by causal locality he means that then it is pointless definition.

@Sambuco said it well here:

but that the difference lies in the fact that, in his formalism/interpretation, non-communication does not apply only to measurements, but to any physical system, at any time, according to the assumed ontology.

In Barandes's reformulation, the derived conditional probabilities can pertain directly to microphysical configurations of the measured system. This permits a stronger inference from the no-signalling principle: That microphysical systems, spacelike-separated for the duration of a physical process, do not influence eachother during that process.

Bell-inequality violations still follow from dynamics, but these dynamics, being non-Markovian, don't imply nonlocal influence.

 

[pines-demon]

Bell-inequality violations still follow from dynamics, but these dynamics, being non-Markovian, don't imply nonlocal influence.

This is the part were I remain skeptic, the dynamics remains nonlocal in the Bell sense but local in Barandes sense. We are repeating the same claims with different wording.

Edit: I would be interesting to see which other interpretations are local in Barandes sense, probally every interpretation? Why is Barandes causal locality interesting anyway?

 

[Morbert]

This is the part were I remain skeptic, the dynamics remains nonlocal in the Bell sense but local in Barandes sense.

Yes. The question is whether the Bell sense actually captures the notion of nonlocal influence if dynamics are non-Markovian. From "La Nouvelle Cuisine"

A theory will be said to be locally causal if the probabilities attached to values of local beables in a space-time region 1 are unaltered by specification of values of local beables in a space-like separated region 2, when what happens in the backward light cone of 1 is already sufficiently specified, for example by a full specification of local beables in a space-time region 3.

If dynamics are non-Markovian, then the full specification of beables in region 3 will not sufficiently specify what happens in the backward light cone of 1, but not because of missing information contained in some spacelike-separated region. Rather, because of missing information in a timelike-separated region. These non-Markovian dynamics take the form of sparse conditional probabilities, and so a region further down the light cone of 1 where these conditional probabilities pertain to, is needed.

 

[javisot]

Bell-inequality violations still follow from dynamics, but these dynamics, being non-Markovian, don't imply nonlocal influence.

"Our work implies that a violation of the Tsirelson bound requires either a violation of the principle of causal locality, and thus manifestly non-local dynamical laws, or requires laws that cannot be formulated in terms of unistochastic processes."

I cannot see in all of Barandés' work where he translates the violation of Bell's inequalities in the terms he proposes; all he says about the violation of Bell's inequalities is that nobody is understanding it properly, simply.


(Barandés suggests that defining "locality" in terms of classical causality is inappropriate for the quantum regime, as it leads to non-locality. His view is that quantum causality prevails in this regime, the non-signaling theorem still holds)

 

[Morbert]

"Our work implies that a violation of the Tsirelson bound requires either a violation of the principle of causal locality, and thus manifestly non-local dynamical laws, or requires laws that cannot be formulated in terms of unistochastic processes."

Note that causal locality here refers to Barandes's principle, not Bell's principle of local causality.

 

[Fra]

Why is Barandes causal locality interesting anyway?

For me it is interesting as a first step in preparing for a pardigm change from where the heart of causality, the dynamical law, lives at global system level, to where it lives at microphysical/subsystem level.

The two paradigms I envision here are system dynamics, and agent based modelling.

What prevents the paradigm change in Barandes view is that the conditional transition matrices, take the role of dynamical law, but Barandes does not provide a clear ontology of where these are encoded. I'd say that in the normal paradigm, dynamical law is encoded at "statistical level" as seen from an an external observer - or possible you can also imagine that they "just are" some mathematical facts about nature, fined tuned and requires no explanation.

To get an interesting notion of the causal locality, the "constraints" of the stochastic processes that occure in parallell at each configuration of every subsystem, we need clarity of how thse constraints are physically supported.

I would say that Baranders has no answer in his papers, but I find it interesting as as I can at least imagine some possible answers.

Baranders elaborations exactly on that the notion of "causality" that is implicit in the system dynamics paradigm is problematic, simple because there is an implicitly timelessness, causality becomes ambigous, as the future and the past are constrained in a deterministic system dynamics. This is IMO why the classical notion of causality is not useful and should be replaced by something better. Barandes made a first step here, but something is indeeed still missing. I think it was in one of the first yotube clips that was posted in some of the barandes disucssions but i lost track of which one.

/Fredrik

 

[Sambuco]

Note that causal locality here refers to Barandes's principle, not Bell's principle of local causality.

I must say that it would not have been bad if Barandes had chosen other words for his principle of "causal locality", instead of simply swapping the order in "local causality".

Lucas.

 

[Sambuco]

If dynamics are non-Markovian, then the full specification of beables in region 3 will not sufficiently specify what happens in the backward light cone of 1, but not because of missing information contained in some spacelike-separated region. Rather, because of missing information in a timelike-separated region. These non-Markovian dynamics take the form of sparse conditional probabilities, and so a region further down the light cone of 1 where these conditional probabilities pertain to, is needed.

This is a key point! In one of his many talks, Barandes said that, in his formulation, the typical spatial non-locality that arises from the usual interpretations of Bell's theorem can be reinterpreted as a "temporal non-locality".

Lucas.

 

[javisot]

I must say that it would not have been bad if Barandes had chosen other words for his principle of "causal locality", instead of simply swapping the order in "local causality".

Lucas.

Completely agree. Barandés could have simply spoken of classical causality (cause first, then effect) and quantum causality (something different from classical causality).

Bell's theorem, according to Bell, tests local causality (classical causality) in Bell-type experiments. If Bell's inequalities are experimentally violated, it would be proven that classical causality is inadequate to explain these experiments. Meanwhile, the Tsirelson bound, according to Barandés, It would test quantum causality; if experimentally proven, quantum causality would not be appropriate to explain that experiment. Let's not expect the Tsirelson bound to be violated in nature.

The non-violation of the Tsirelson bound is a consequence of accepting the standard quantum structure (Hilbert + Born), and therefore holds true in both quantum mechanics and relativistic QFT.

 

[iste]

Where does he claim that?

Signalling in terms of the evolution of particles' Markovian trajectories is the stochastic mechanical manifestation of Bohmian nonlocal influence.

It is briefly mentioned the Nelsonian stochastic mechanics nonlocality and how Barandes avoids it in that clip Sambuco linked in post #298, soon after the timetag, in the question by Peter Morgan iirc.

 

[iste]

Just a thought: in his arxiv:2402 paper, it looks like Baranded implies that the subject systems (not measuring devices) do nonlocally causally influence each other, but he explains this away due to the initial interaction even though he cannot "plug it in" as a conditioning variable.

I also think Barandes' theory prima facie avoids signalling where others don't because I believe hs theory only gives you the non-signalling observable probabilities that exist in quantum mechamics, I believe. Whereas in stochastic mechanics the particle trajectories are hidden, unobservable and so have transition probabilities that don't exist in quantum mechanics but nonetheless rely on parts of quantum theory overtly connected to non-locality, i.e. the imaginary momentum appearing in some formulation (also related to quantum potential) which just so happens to go away with measurement (of momentum).

 

[Fra]

"Our work implies that a violation of the Tsirelson bound requires either a violation of the principle of causal locality, and thus manifestly non-local dynamical laws, or requires laws that cannot be formulated in terms of unistochastic processes."

I'll try to sharpen the exact place in Barandes exposition where my conceptual issues are and where the ontology is unclear, and the context where I think more research would be extremely interesting.

If we start by a more general stochastic process, that is NOT unistochastic, then tsirelson bound will not be respected. Thus, it seems to me that in a bigger theory space, tsireloon bound matches precisely when the stochastic game becomes a unistochstic one.

A physical mechanism and ontology for this, is lacking (this is the missing part). But in Barandes view, there are two parts

1. the stochastic-quantum that says that any general stochastic system can be EMBEDDED into a unistochastic bigger system. But this is an existence theorem, not a uniqueness.

2. the Stochastic–Quantum Correspondence is the actual dictionary from QM to a unistochastic system.

So if you start with QM, go via (2), then tsirelon bound is respected.

But if you just go via (1), tsirelon bound is also respected, but you have not yet found the actual unistochastic, all we know is "there exists an embedding" and is consistent for all subsystems as the interact? So for this to have deep explanatory value, the question I ask is: can we understand the "selection" or "emermgence" of unistochastic embedding, without cheating and using (2). And is this itself a physical process althoug so far from equilibrium that it violates currently known laws - or not?

/Fredrik

 

[Morbert]

Personally, and for the time being, I would characterize Barandés' interpretation as a middle ground between objective collapse and the standard interpretations. It may well explain more results than those interpretations that include objective collapse, but I don't believe it reaches the level of a standard interpretation.

Objective collapse and no-objective-collapse theories disagree about what happens in the realm of superobservers. Barandes's interpretation is staunchly in the no-objective-collapse camp in this regard.

And I wouldn't agree that it doesn't reach the level of a standard interpretation. The correspondence does appear to be exact.

 

[pines-demon]

And I wouldn't agree that it doesn't reach the level of a standard interpretation. The correspondence does appear to be exact.

A mathematical correspondence is not an interpretation. Is a Wick rotation an interpretation? for me Barandes vision is missing a toy model preferably unrelated to quantum mechanics.

 

[Morbert]

A mathematical correspondence is not an interpretation. Is a Wick rotation an interpretation? for me Barandes vision is missing a toy model preferably unrelated to quantum mechanics.

You just read off the ontology from the configuration space on the stochastic side of the correspondence.

 

[pines-demon]

You just read off the ontology from the configuration space on the stochastic side of the correspondence.

I can’t because I do not have any clear example of this kind of non-Markovian dynamics in other contexts.

 

[Morbert]

I can’t because I do not have any clear example of this kind of non-Markovian dynamics in other contexts.

The configuration space isn't Markovian or non-Markovian. It's the ordinary configuration space of classical physics.

 

[pines-demon]

The configuration space isn't Markovian or non-Markovian. It's the ordinary configuration space of classical physics.

Sure but I am talking about the dynamics (the indivisible part).

 

[Morbert]

Sure but I am talking about the dynamics (the indivisible part).

I just don't see the interpretational challenge. At each time the system is in a definite configuration, with a likelihood given by the dynamics.

 

[pines-demon]

I just don't see the interpretational challenge. At each time the system is in a definite configuration, with a likelihood given by the dynamics.

Can you provide a simple example of what would that look like?