Why is Bell’s theorem still considered controversial?

[danieltanfh95]

TL;DR

How did we end up with faster-than-light communication, superdeterminism or other exotic interpretations when Bell's theorem is something quite trivial to understand

Taken at face value, Bell’s theorem seems to say that entangled particles are one shared non-separable joint wavefunction, and measurements of entangled particles (A,B) at two different spacetime coordinates produce correlated outcomes because they’re reading one joint object via the Born rule. We don’t expect any information to propagate between the two sites, because the information is contained within the underlying joint wavefunction instead (Bob’s reduced density matrix is mathematically independent of Alice’s measurement choice, also known as the no-signalling theorem).

However, I see “thought experiments” like https://arxiv.org/abs/1110.3795 that confuse wavefunctions with labels and then conclude that there must somehow be superluminal signalling between local hidden states otherwise, and then make the subsequent claim that “this falsifies direct causation in quantum mechanics.”

However, Reichenbach’s Principle of Common Cause (PCC) is commonly interpreted using commutative probabilities, which already suggests that its classical definition cannot be used to explain “common cause” in quantum mechanics, even if objectively, entanglement preparation is common cause by any rational description of the term. Cavalcanti and Lal (2014, J. Phys. A 47, 424018) https://arxiv.org/abs/1311.6852 decomposed PCC into independent assumptions and showed how dropping the factorization condition (route c, Leifer-Spekkens 2013) accommodates quantum cases cleanly, which also leads to a general definition of common cause. This still does not imply that there must be superluminal signalling, just that we need to update our priors on how to define common cause. One does not say that because matter particles have wave-like behaviour the particle view is completely out of the picture as an effective description.

Additionally, the term “local hidden state” is a huge gaffe on the part of the science influencers and pedagogy. That is simply not what Bell meant with “shared classical seed consumed by two independent per-site functions”. It is quite straightforward to understand the characteristic of quantum mechanics as described in Bell’s theorem using macroscopic behaviour. We do that all the time with computers. Take for example the Gameboy, where a shared memory layout can be read differently by some generator function to produce a phenomenon like in-game items that are situated in very different spacetime coordinates in the game. There are three layers, the actual memory layout, the code describing the memory that translates it to electrical signals to the screen and sound card, and the perceived game itself. What we interact with (or “measure” or “observe”) within the game is only an observable produced by some function or procedure, which hides its internals from access by other functions, like using closures. The function itself can have local (scoped to the function) hidden (inaccessible by others, only observed outputs) state, even if joint evaluation produces CHSH-violating outcomes.

So my question really is this: why and how did it all get so confusing with faster-than-light communication, superdeterminism or other exotic interpretations when it is something quite trivial to understand, or does my explanation above constitute some new theorem (because if so, I’m not even sure what the core thesis is).

 

[danieltanfh95]

Simple python code to illustrate my point

"""Bell violation from an immutable joint state via closure, with decoherence."""

import numpy as np


Y = np.array([[0, -1j], [1j, 0]], dtype=complex)


def eigenstate(theta, sign):
    if sign == +1:
        return np.array([np.cos(theta / 2), np.sin(theta / 2)], dtype=complex)
    return np.array([-np.sin(theta / 2), np.cos(theta / 2)], dtype=complex)


def projector(theta, sign):
    v = eigenstate(theta, sign)
    return np.outer(v, v.conj())


def make_independent_generators():
    def generator_A(theta, rng):
        return int(rng.choice([+1, -1]))

    def generator_B(theta, rng):
        return int(rng.choice([+1, -1]))

    return generator_A, generator_B


def couple_to_singlet():
    v = np.array([0, 1, -1, 0], dtype=complex) / np.sqrt(2)
    return np.outer(v, v.conj())


def make_joint_generator(density):
    def joint_generator(theta_a, theta_b, rng):
        outcomes = [(+1, +1), (+1, -1), (-1, +1), (-1, -1)]
        probs = np.array([
            np.real(np.trace(
                density @ np.kron(projector(theta_a, s_a), projector(theta_b, s_b))
            ))
            for s_a, s_b in outcomes
        ])
        probs = np.clip(probs, 0.0, None)
        probs = probs / probs.sum()
        return outcomes[rng.choice(len(outcomes), p=probs)]

    return joint_generator


def dephase(density, gamma, t):
    decay = np.exp(-gamma * t)
    rho = density.copy()
    for i in range(4):
        for j in range(4):
            if i != j:
                rho[i, j] *= decay
    return rho


def concurrence(density):
    yy = np.kron(Y, Y)
    product = density @ (yy @ density.conj() @ yy)
    roots = np.sort(np.sqrt(np.abs(np.linalg.eigvals(product))))[::-1]
    return max(0.0, float(roots[0] - roots[1] - roots[2] - roots[3]))


def partial_trace_B(density):
    rho = np.zeros((2, 2), dtype=complex)
    for i in range(2):
        for j in range(2):
            for k in range(2):
                rho[i, j] += density[2 * i + k, 2 * j + k]
    return rho


def pointer_select(density, theta_a, theta_b, rng):
    outcomes = [(+1, +1), (+1, -1), (-1, +1), (-1, -1)]
    probs = np.array([
        np.real(np.trace(
            density @ np.kron(projector(theta_a, s_a), projector(theta_b, s_b))
        ))
        for s_a, s_b in outcomes
    ])
    probs = np.clip(probs, 0.0, None)
    probs = probs / probs.sum()
    return outcomes[rng.choice(len(outcomes), p=probs)]


def correlation(density, theta_a, theta_b, n=20_000):
    rng = np.random.default_rng(0)
    return float(np.mean([
        a * b for a, b in (
            pointer_select(density, theta_a, theta_b, rng) for _ in range(n)
        )
    ]))


def chsh(density, n=20_000):
    a, a_prime = 0, np.pi / 2
    b, b_prime = np.pi / 4, 3 * np.pi / 4
    return (
        correlation(density, a, b, n)
        - correlation(density, a, b_prime, n)
        + correlation(density, a_prime, b, n)
        + correlation(density, a_prime, b_prime, n)
    )


if __name__ == "__main__":
    rng = np.random.default_rng(42)

    print("Independent generators:")
    gen_A, gen_B = make_independent_generators()
    products = [gen_A(0, rng) * gen_B(0, rng) for _ in range(5000)]
    print(f"  mean(a * b) = {np.mean(products):+.3f}")

    print("\nCoupling:")
    density = couple_to_singlet()
    print(f"  concurrence = {concurrence(density):.3f}")
    print(f"  |CHSH| = {abs(chsh(density)):.3f}")

    print("\nDecoherence:")
    gamma = 1.0
    print(f"  {'t':>5}  {'coherence':>10}  {'concurrence':>12}  {'|S|':>8}")
    for t in [0.0, 0.3, 0.7, 1.5, 3.0, 7.0]:
        rho_t = dephase(density, gamma, t)
        print(
            f"  {t:>5.1f}  {np.exp(-gamma * t):>10.3f}  "
            f"{concurrence(rho_t):>12.3f}  {abs(chsh(rho_t)):>8.3f}"
        )

    print("\nReduced density matrix at A:")
    print(f"  before decoherence: {partial_trace_B(density).real.diagonal()}")
    print(f"  after decoherence:  {partial_trace_B(dephase(density, gamma, 10.0)).real.diagonal()}")

 

[Nugatory]

TL;DR: How did we end up with faster-than-light communication, superdeterminism or other exotic interpretations when Bell's theorem is something quite trivial to understand

So my question really is this: why and how did it all get so confusing with faster-than-light communication, superdeterminism or other exotic interpretations when it is something quite trivial to understand, or does my explanation above constitute some new theorem (because if so, I’m not even sure what the core thesis is).

There’s not a lot of controversy about what Bell’s theorem says, nor about how it emerges from the mathematical formalism of quantum mechanics as a theory about the probabilities of measurement results. But that just invites another question: what is it about a physical system that makes observations behave according to this mathematical formalism and not some other?

Compare with the ideal gas law. Once we accept that there is are observable quantities of pressure and temperature, we find that $PV=nRT$ works really well, it’s easy to understand, it’s not controversial. But what is it about a volume of gas that makes it behave according to that law and not some other? That’s easily answered: we do a bunch of statistical stuff based on the well-understood and even less controversial Newton’s laws… and we see that that’s how a gas composed of classically interacting particles is expected to behave.

But when we ask the analogous question of a quantum system, there’s nothing well-understood and non-controversial underneath. Bell’s Theorem tells us that whatever the answer, we must either accept non-locality, reject counterfactual definiteness, or appeal to something inherently implausible like superdeterminism.

And that’s why we’re still talking about Bell’s Theorem and likely will be for a long time to come.

 

[Matterwave]

Cavalcanti and Lal (2014, J. Phys. A 47, 424018) https://arxiv.org/abs/1311.6852 decomposed PCC into independent assumptions and showed how dropping the factorization condition (route c, Leifer-Spekkens 2013) accommodates quantum cases cleanly, which also leads to a general definition of common cause.

Is your proposal to follow their lead and drop facorizability? And are you going down the LS route (using conditional quantum states) or dropping factorizability altogether?

This still does not imply that there must be superluminal signalling, just that we need to update our priors on how to define common cause.

Note that "no superluminal signaling" is not enough to maintain relativistic causal structure according to Bell and those pushing the non-locality thread. Bohmian Mechanics is, as far as I can see, also "local" in the no superluminal signalling sense--though there were a few papers in the late 2010s and early 2020s debating this point. Most however would consider BM a grossly nonlocal interpretation.

Additionally, the term “local hidden state” is a huge gaffe on the part of the science influencers and pedagogy.

Fwiw, I've actually not seen this particular terminology and I do consume a decent amount of online pedagogy about QM. "Local hidden variables" abounds, but this is the first time I've seen "local hidden state".

The function itself can have local (scoped to the function) hidden (inaccessible by others, only observed outputs) state, even if joint evaluation produces CHSH-violating outcomes.

The sense of "local" used here seems unrelated to the discussion to me. Other than both using the same word "local", I don't see the connection between local variables in the sense of namespaces in computer science and local in spacetime.

So my question really is this: why and how did it all get so confusing with faster-than-light communication, superdeterminism or other exotic interpretations when it is something quite trivial to understand, or does my explanation above constitute some new theorem (because if so, I’m not even sure what the core thesis is).

There are various no-go theorems in QM, Bell being perhaps the most famous one. Each valid no go theorem comes with assumptions, at least one of which has to be denied. Pick one (say, factorizability if that's the one you prefer), deny it, and see others come to debate it. Then you will see why there's debate on this subject. :)

 

[jeffn1]

The interesting thing is, when we measure it/detect locally, the entanglement ends immediately for both the local and nonlocal particle, seemingly "causing" an effect billions of miles away (even though the termination of the correlation cannot be confirmed until the non-local particle is measured.) [Caveat: Maybe this is only accurate under some interpretations].

[Moderator's note: Unnecessary link deleted]

 

[PeterDonis]

when we measure it/detect locally, the entanglement ends immediately for both the local and nonlocal particle

Only in certain interpretations.

Caveat: Maybe this is only accurate under some interpretations

Yes, indeed.

 

[Matterwave]

Source: https://www.physicsforums.com/threads/why-is-bells-theorem-still-considered-controversial.1086607/

Did you mean to link back to this current thread?

 

[jeffn1]

Did you mean to link back to this current thread?

No, that was user error.

 

[PeterDonis]

No, that was user error.

Now fixed.

 

[Demystifier]

The function itself can have local (scoped to the function) hidden (inaccessible by others, only observed outputs) state, even if joint evaluation produces CHSH-violating outcomes.

In QM, the Bell theorem tells us that this hidden state cannot be local, in the sense that it cannot obey purely local laws of physics.

 

[danieltanfh95]

Is your proposal to follow their lead and drop facorizability? And are you going down the LS route (using conditional quantum states) or dropping factorizability altogether?

I don’t think it’s a matter of routes. It’s a matter of stopping the assumption that classical probabilities apply to non-commutative probabilities. The community should instead work on a generalised way to describe common causes, instead of treating common cause as non-existent. Here, I use route c as an example of my own exploration.

Note that "no superluminal signaling" is not enough to maintain relativistic causal structure according to Bell and those pushing the non-locality thread. Bohmian Mechanics is, as far as I can see, also "local" in the no superluminal signalling sense--though there were a few papers in the late 2010s and early 2020s debating this point. Most however would consider BM a grossly nonlocal interpretation.

The problem with Bohmian is adding more questions instead of reducing questions. What is the physical mechanism of how the particle moves with the waves? Why are there relativistic violations in the waves themselves? Being non-local is fine as long as you can show that it obeys the rest of the physics.

Fwiw, I've actually not seen this particular terminology and I do consume a decent amount of online pedagogy about QM. "Local hidden variables" abounds, but this is the first time I've seen "local hidden state".

Sure, I meant local hidden variables. This does not change the argument.

The sense of "local" used here seems unrelated to the discussion to me. Other than both using the same word "local", I don't see the connection between local variables in the sense of namespaces in computer science and local in spacetime.

Correct me if I’m wrong but Bell’s theorem doesn’t actually assume a specific spacetime location or volume in its formal statement either. The locality condition is structural (outcomes at one site don’t depend on the setting at the other site given λ), which is why I’m pointing out that “local” is specific to the layer the observable or condition is in.

There are various no-go theorems in QM, Bell being perhaps the most famous one. Each valid no go theorem comes with assumptions, at least one of which has to be denied. Pick one (say, factorizability if that's the one you prefer), deny it, and see others come to debate it. Then you will see why there's debate on this subject. :)

Yes, I’m not saying to pick one to deny. That’s a sort of thinking that comes from trying to preserve classical mechanics in the first place, and the wrong conclusion to derive from quantum mechanics. It’s like saying that GR allows you to pick and choose. It doesn’t.

 

[Matterwave]

I don’t think it’s a matter of routes. It’s a matter of stopping the assumption that classical probabilities apply to non-commutative probabilities.

Simply slapping a "classical" label on something doesn't make it false.

Reichenbach's principle of common causes says that when two events $A,B$ are correlated and not independent, i.e. $p(A,B) > p(A)p(B)$, then they are explained by common causes $C$ such that once all common causes are screened off, the probabilities become independent:

$$p(A,B|C) = p(A|C)p(B|C)$$

In the context of Bell, where $A,B$ are Alice and Bob's measurement outcomes respectively, $a,b$ their detector settings respectively, and $\lambda$ all other possible common causes (hidden variables), we get a factorization $p(A,B|\lambda, a, b) = p(A|\lambda, a)p(B|\lambda, b)$. The fact that the right hand side values are not conditioned on $A,B$ is called "outcome independence" (it is a kind of locality constraint). The fact that the probability $p(A|\lambda, a)$ is not conditioned on $b$ and vice versa is called "parameter independence" (this is the no-signaling kind of locality constraint). The fact that we can perform this factorization at all is due to Reichenbach (though Bell formulated it in his own way, without reference to RPCC).

So when you say you want to stop assuming "classical probabilities apply to non-commutative probabilities" what are you saying exactly?

If what you are saying is to get rid of the above analysis wholesale and just stick with $p(A,B) = |\langle A,B|\Psi\rangle|^2$ where $|A,B\rangle$ is the simultaneous eigenstate of $\sigma_A\cdot a$ and $\sigma_B\cdot b$ then you're really not saying anything more than orthodox QM. If you're saying something else, then please state so clearly and preferably with some mathematics to illustrate what you mean.

The community should instead work on a generalised way to describe common causes, instead of treating common cause as non-existent.

I would just say, think hard about what you really mean by "common cause". It's not that I can say with certainty Reichenbach's principle is the only way to formulate a common cause principle; however, it is quite intuitive and seems to match much of the understanding of probability.

The problem with Bohmian is adding more questions instead of reducing questions. What is the physical mechanism of how the particle moves with the waves? Why are there relativistic violations in the waves themselves? Being non-local is fine as long as you can show that it obeys the rest of the physics.

@Demystifier would be the best person to address these concerns. But from what I know, BM gives an ontology (the particles + the wave function), a dynamical equation for each (of the) object(s) in the ontology (the guidance equation and Schroedinger's equation), and provably reproduces the standard predictions of (at least non-relativistic) QM.

Sure, I meant local hidden variables. This does not change the argument.

Got it. In that case, local hidden variable theories, as a class, were disproved (excepting certain loopholes such as super determinism) by the experimental validation of violations of Bell Inequalities.

Correct me if I’m wrong but Bell’s theorem doesn’t actually assume a specific spacetime location or volume in its formal statement either. The locality condition is structural (outcomes at one site don’t depend on the setting at the other site given λ), which is why I’m pointing out that “local” is specific to the layer the observable or condition is in.

Can you clarify? What do you mean by "layer"?

The strongest argument I know from Bell about what his theorem actually means is found in his paper The Theory of Local Beables. In it, he shows how his inequalities (in the CHSH form) arises from relativistic locality conditions.

Yes, I’m not saying to pick one to deny. That’s a sort of thinking that comes from trying to preserve classical mechanics in the first place, and the wrong conclusion to derive from quantum mechanics. It’s like saying that GR allows you to pick and choose. It doesn’t.

Unfortunately, you have to pick one (or more) to deny. The only way around a (valid) theorem is to deny one of its assumptions (premises) since the logical chain from premises to conclusions can not be broken. Bell's theorem, for example, is not about preserving classical mechanics, rather, it lays out clearly what assumptions went into it so that you know that if violations of his inequalities are observed in the laboratory, then (at least) one of the assumptions must be false.

Pick your poison:

1. Locality / relativistic causal structure
2. RPCC
3. Statistical measurement independence
4. Single measurement outcomes

A valid way to escape picking one of those 4 is to show, formally, another load-bearing assumption in Bell's theorem and violate that. MWI went this way and that's why 4 got added to that list.

 

[danieltanfh95]

Simply slapping a "classical" label on something doesn't make it false.

Reichenbach's principle of common causes says that when two events $A,B$ are correlated and not independent, i.e. $p(A,B) > p(A)p(B)$, then they are explained by common causes $C$ such that once all common causes are screened off, the probabilities become independent:

$$p(A,B|C) = p(A|C)p(B|C)$$
...

I’m not proposing anything new. My point is that orthodox QM with wavefunction realism is already a (somewhat) complete description. The problem is that folks revert to classical mental models (demanding CFD, expecting classical factorization, treating particles as labelled objects) and then generate pseudo-questions like FTL propagation to resolve puzzles that only arise under those classical commitments.

Like your issue on me using “state” vs “variables”, I take issue with your treatment of “classical” as just a general category rather than a specific mathematical structure. To be more precise, Reichenbach’s PCC is formulated in classical commutative probability theory (Kolmogorov axioms, joint distributions over all observables, well-defined classical conditional probability). QM is obviously built on non-commutative probabilities that generalise into C*-algebras. Classical probability is a special case, and we can expect a similar case to be made for common causes.

Which brings me to the second point, which I think is the crux of the problem. To me, Reichenbach's PCC in terms of classical probabilities is a poor representation of common cause as a philosophical concept since it does not cover the QM case. However, people seem to be hung up on the mathematical representation and then use that classical representation to somehow capture the whole meaning of common cause.

Just because something is intuitive doesn't mean it's right. One would think that we would have already moved past that with quantum mechanics, but perhaps this is the problem. People are still in denial over the fact that we need to generalise or update our priors on every mathematical representation of philosophical content we have (as long as the logic is rational).

 

[Demystifier]

The problem with Bohmian is adding more questions instead of reducing questions. What is the physical mechanism of how the particle moves with the waves? Why are there relativistic violations in the waves themselves? Being non-local is fine as long as you can show that it obeys the rest of the physics.

Let me answer your questions in the reverse order.

1. Bohmian mechanics (BM) makes the same measurable predictions as standard quantum theory, in this sense it obeys "the rest of physics". Some principles, like the Born rule, the uncertainty principle and the principle of Lorentz invariance, may in BM be emergent rather than fundamental, but it does not violate any existing empirical knowledge. It can only violate theoretical prejudices of some physicists about which principles are supposed to be fundamental, but so much the worse for the prejudices.

2. The pilot waves in BM obey exactly the same equations of motion as the corresponding waves in standard quantum theory, so when there are no relativistic violations of waves in standard quantum theory, then there are no relativistic violations of waves in BM either.

3. As for the physical mechanism of how the particle moves with the wave, this is like asking for the physical mechanism of how the classical particle moves with the Hamilton-Jacobi function $S({\bf x},t)$ in the Hamilton-Jacobi formulation of classical mechanics. The Hamilton-Jacobi theory does not provide any additional such mechanism, except the equations that describe this motion, and yet, nobody complains that it is any sort of deficiency of the classical theory. The Bohmian equations of motion are very similar to the classical Hamilton-Jacobi equations of motion, so questions about the "physical mechanism", in addition to the precise equations that BM provides, are off the target.

So BM answers all the additional questions that you asked.

 

[Matterwave]

I’m not proposing anything new. My point is that orthodox QM with wavefunction realism is already a (somewhat) complete description.

I see some hedging with the (somewhat), but it seems your view is basically that of orthodox QM? It is called "orthodox" because it is the most popular (afaik it still is) view of QM.

The problem is that folks revert to classical mental models (demanding CFD, expecting classical factorization, treating particles as labelled objects) and then generate pseudo-questions like FTL propagation to resolve puzzles that only arise under those classical commitments.

I'm gonna say it again, slapping a "classical" label on something doesn't make it false or wrong. If your contention is only that orthodox QM explains the world then plenty of people (including Bohr, Heisenberg, etc.) held that view. Not much for me to debate there.

On my part, I enjoy solving puzzles and I think one way we make progress in physics (and in general) is by thinking deeply about various issues, problems, paradoxes and puzzles.

Like your issue on me using “state” vs “variables”, I take issue with your treatment of “classical” as just a general category rather than a specific mathematical structure.

I don't see the analogy here. I didn't have an issue with you using "state", I just hadn't seen that terminology "local hidden state" used before.

To be more precise, Reichenbach’s PCC is formulated in classical commutative probability theory (Kolmogorov axioms, joint distributions over all observables, well-defined classical conditional probability). QM is obviously built on non-commutative probabilities that generalise into C*-algebras. Classical probability is a special case, and we can expect a similar case to be made for common causes.

Ok. But I'm not sure what case you are trying to argue here. If your case is simply "orthodox QM is fine as-is" then again, that's a viewpoint held by many (possibly the majority).

Which brings me to the second point, which I think is the crux of the problem. To me, Reichenbach's PCC in terms of classical probabilities is a poor representation of common cause as a philosophical concept since it does not cover the QM case. However, people seem to be hung up on the mathematical representation and then use that classical representation to somehow capture the whole meaning of common cause.

Is this a specific argument against Bell's theorem? Are you making the claim, for example, that violations of Bell Inequalities do not constitute a proof of non-locality because RPCC is not a valid step?

Just because something is intuitive doesn't mean it's right. One would think that we would have already moved past that with quantum mechanics, but perhaps this is the problem. People are still in denial over the fact that we need to generalise or update our priors on every mathematical representation of philosophical content we have (as long as the logic is rational).

Is the claim here to simply abide by the rules of orthodox QM and stop trying to dig deeper?

 

[Sambuco]

I’m not proposing anything new. My point is that orthodox QM with wavefunction realism is already a (somewhat) complete description.

Does "wavefunction realism" imply a $\psi$-ontic interpretation? In that case, there are interpretations consistent with this, such as many-worlds, but many people remain unsatisfied and, therefore, opt for other alternatives.

Lucas.

 

[danieltanfh95]

...
So BM answers all the additional questions that you asked.

I think BM actually plays into my original point of trying to hold onto classical particles and then introducing a non-physical equation that cannot be justified empirically beyond its ability to reproduce QM. “The equations describe the motion” ends up functioning as “let’s just hide the ugly stuff behind postulated mathematics”.

I see some hedging with the (somewhat), but it seems your view is basically that of orthodox QM? It is called "orthodox" because it is the most popular (afaik it still is) view of QM.

Mostly yes, and specifically with the wavefunction treated as physical ($\psi$-ontic in Harrigan-Spekkens’ sense, though whether it’s primary or emergent from something deeper is a further question). Even when people say they’re “orthodox QM”, they drift towards $\psi$-epistemic when pressed on interpretation, and that’s where the paradoxes come in when they try to reconcile the resulting tension.

Are you making the claim, for example, that violations of Bell Inequalities do not constitute a proof of non-locality because RPCC is not a valid step?

Yes. Bell violations prove no classically-factorizing LHV model reproduces QM. The factorization failure is a structural property of non-separable wavefunctions under the Born rule, not evidence of a dynamical propagation mechanism. Interpreting it as “non-locality” in the dynamical sense is importing a classical causal story onto a structural fact.

Is the claim here to simply abide by the rules of orthodox QM and stop trying to dig deeper?

No. I’m asking why we’re generally digging in the direction of classical mechanics (BM, and the extensive literature on FTL signalling / causation puzzles, especially in quantum information) instead of digging in directions consistent with QM like non-commutative probability, QFT foundations, quantum common-cause frameworks, decoherence and einselection, and so on.

Since I can’t share personal work here, I can’t propose specific new theories.

Does "wavefunction realism" imply a $\psi$-ontic interpretation? In that case, there are interpretations consistent with this, such as many-worlds, but many people remain unsatisfied and, therefore, opt for other alternatives.

Lucas.

Yes, but do note that spontaneous collapse theories like GRW are also $\psi$-ontic. I think the primary problem is how single outcomes are handled, and I’m considering whether QFT implies the measurement problem is instead a problem of branch selection rather than collapse.

 

[Sambuco]

Yes, but do note that spontaneous collapse theories like GRW are also ψ-ontic. I think the primary problem is how single outcomes are handled, and I’m considering whether QFT implies the measurement problem is instead a problem of branch selection rather than collapse.

Okay, I see. What would be different in QFT compared to the non-relativistic case? I’m referring to the many-worlds interpretation and the branch selection.

As an aside, it often happens that one perceives interpretations other than one’s own preferred one as being too closely tied to classical prejudices. For instance, I tend to favor $\psi$-epistemic interpretations such as relational quantum mechanics and I often view many-worlds as being too classical.

Lucas.

 

[Matterwave]

Mostly yes, and specifically with the wavefunction treated as physical ($\psi$-ontic in Harrigan-Spekkens’ sense, though whether it’s primary or emergent from something deeper is a further question). Even when people say they’re “orthodox QM”, they drift towards $\psi$-epistemic when pressed on interpretation, and that’s where the paradoxes come in when they try to reconcile the resulting tension.

This is interesting. Sometimes one hears from the orthodox QM crowd that a lean towards a $\psi$-epistemic reading avoids paradoxes since the wave function "is just book keeping".

For example, probably one of the more hardline orthodox QM papers I've come across (from a different thread here on PF actually) is this one from Englert: https://arxiv.org/pdf/1308.5290

The abstract is literally:

Quantum theory is a well-defined local theory with a clear interpretation. No “measurement problem” or any other foundational matters are waiting to be settled.

And yet, if you read section 6, he obviously ascribes the statistical operator $\rho$ a book keeping only role:

The statistical operator is not a physical object or a property of a physical object, it describes the object by encoding what we know about it, and state reduction is the book keeping device for updating the description. An electron carries its mass and charge around, but not “its” statistical operator.

It is interesting that you seem to agree with Englert that QM is not in need of further interpretation but you reach the opposite conclusion in terms of the nature of $\psi$.

I must admit that I myself am quite skeptical of Englert's view and that paper in particular. However, the paper does seem to cut against your claim that $\psi$-epistemic views lead to paradoxes and tensions. At least, Englert, it seems to me, would disagree.

Yes. Bell violations prove no classically-factorizing LHV model reproduces QM. The factorization failure is a structural property of non-separable wavefunctions under the Born rule, not evidence of a dynamical propagation mechanism. Interpreting it as “non-locality” in the dynamical sense is importing a classical causal story onto a structural fact.

What does this mean for you -- "non-locality" in the dynamical sense? Do you mean a violation of the no-signaling theorem? Note that Bell violations do not constitute violations of the no-signaling theorem.

No. I’m asking why we’re generally digging in the direction of classical mechanics (BM, and the extensive literature on FTL signalling / causation puzzles, especially in quantum information) instead of digging in directions consistent with QM like non-commutative probability, QFT foundations, quantum common-cause frameworks, decoherence and einselection, and so on.

Since I can’t share personal work here, I can’t propose specific new theories.

Well, you're the one who started this thread. :P If you do not wish to discuss these matters, you could choose not to.

But historically speaking, how much time do you think has been spent "digging in the direction of classical mechanics" vs "digging in directions consistent with QM"? For example, I view the entirety of the work developing QFT to be "digging in directions consistent with QM". Further, all the work in applied physics ranging from condensed matter physics to plasma physics to optics/photonics to quantum computing -- all of this work is done consistently with QM and generally without regard for any deeper interpretation.

 

[Demystifier]

The equations describe the motion” ends up functioning as “let’s just hide the ugly stuff behind postulated mathematics”.

Unlike Copenhagen/orthodox type interpretations, which function as "let's just hide ugly stuff behind vague words, so when someone presses us to be more precise we can hide behind the shut-up-and-calculate doctrine, and when someone objects that we don't understand the meaning of those calculations we can play again with vague words."

 

[danieltanfh95]

Okay, I see. What would be different in QFT compared to the non-relativistic case? I’m referring to the many-worlds interpretation and the branch selection.

QFT is quite different from the non-relativistic case, notably that particles are localised excitations in quantum fields, which leads to a conclusion of universal entanglement. Pointer observables can be identified with specific QFT observables so we can derive an operator-specific einselection explanation rather than through generic environmental interactions.

It is interesting that you seem to agree with Englert that QM is not in need of further interpretation but you reach the opposite conclusion in terms of the nature of $\psi$.

Yes, because I do not subscribe to the "shut-up-and-calculate" view.

For example, I view the entirety of the work developing QFT to be "digging in directions consistent with QM".

I have only been referring to interpretations, especially those that import classical commitments.

If you do not wish to discuss these matters, you could choose not to.

I'm having trouble understanding the motivation for shoehorning classical commitments or classical probabilities into quantum mechanics.

Unlike Copenhagen/orthodox type interpretations, which function as "let's just hide ugly stuff behind vague words, so when someone presses us to be more precise we can hide behind the shut-up-and-calculate doctrine, and when someone objects that we don't understand the meaning of those calculations we can play again with vague words."

As mentioned above, I don’t subscribe to the shut-up-and-calculate view which, in my opinion, bears the smell of religious avoidance by subscribing to the view that physics, or quantum mechanics, is only a simulation of the real world (an instrumentalist position with roots going back to Osiander’s 1543 preface to Copernicus, where mathematical astronomy was explicitly framed as saving the appearances to avoid conflict with theological ontology). This means I’m also not in favour of hiding ugly details behind the mathematics.

However, it does seem that I occupy a weird niche in that I subscribe to orthodox quantum mechanics as the (mostly, considering GR) whole story but I also assume the reality of the wavefunction?

 

[Demystifier]

However, it does seem that I occupy a weird niche in that I subscribe to orthodox quantum mechanics as the (mostly, considering GR) whole story but I also assume the reality of the wavefunction?

So you are one of those who hides behind the vague words, just sayin.

 

[danieltanfh95]

...hides behind the vague words...

I can point you to this special issue on foundations where some of this work addresses the physical content behind what might otherwise look like vague mathematics: https://www.mdpi.com/journal/quantumrep/special_issues/EA3316RV40

 

[Morbert]

However, it does seem that I occupy a weird niche in that I subscribe to orthodox quantum mechanics as the (mostly, considering GR) whole story but I also assume the reality of the wavefunction?

Are you describing many-worlds here?

 

[danieltanfh95]

Are you describing many-worlds here?

No, I’m personally (and separately) arguing for single outcomes rather than many branches.

 

[zonde]

I'm having trouble understanding the motivation for shoehorning classical commitments or classical probabilities into quantum mechanics.

My answer would be that QM predictions are statistics based while real world observations are event based. Considering (at least for me) event is more primitive concept than statistics, it should be possible to have an event based model that reproduces statistical QM predictions.

 

[Sambuco]

QFT is quite different from the non-relativistic case, notably that particles are localised excitations in quantum fields, which leads to a conclusion of universal entanglement. Pointer observables can be identified with specific QFT observables so we can derive an operator-specific einselection explanation rather than through generic environmental interactions.

Do you have any reference for that last point?

Lucas.

 

[Matterwave]

I have only been referring to interpretations, especially those that import classical commitments.

Which would be?

 

[PeterDonis]

Since I can’t share personal work here, I can’t propose specific new theories.

That's correct. Also please bear in mind one of the things that is said in the guidelines for this subforum, that the kinds of discussions we have here are about things that can't be resolved by experiment, so they come down to people's opinions, and if different people have different opinions, there's no way to resolve that. So discussion often can't go any further than making sure everyone has a fair chance to state their opinions.

 

[PeterDonis]

@danieltanfh95 the reference you provided in response to @Sambuco is (a) your own work, and (b) only just published this year, meaning it is too new to be a valid reference, since there has not been sufficient time for the relevant scientific community to review and respond to it.