’t Hooft on the Foundations of Superstring Theory

[Quantumental]

Physics has never in its entire history answered a question of "why", only "how", or "how much". "why" is purely philosophical, and will never get a scientific answer. Until you can relate to experiment, observation, etc, arguing between determinism and indeterminism has no connection (or relevance) to science.

Find just one situation where indeterminism fails and everyone will agree QM is lacking something.

I am not asking WHY, I am asking HOW, how does the Universe obey the Born Rule if there are no causes that makes it?

 

[tom.stoer]

"why" was definately an important question during the development of quantum mechanics; read Heisenberg, Schrödinger, Weitzsäcker et al. But I agree that even if "why" is interesting, only the "how" can be answered (partially) by physics.

Just as an example: there is no satisfactory answer to "why are people buying Hawking's books?", whereas "how they are buying them?" can be answered quite easily.

 

[Quantumental]

"why" was definately an important question during the development of quantum mechanics; read Heisenberg, Schrödinger, Weitzsäcker et al. But I agree that even if "why" is interesting, only the "how" can be answered (partially) by physics.

Just as an example: there is no satisfactory answer to "why are people buying Hawking's books?", whereas "how they are buying them?" can be answered quite easily.

I don't think it's that hard to answer why people are buying Hawking books. He is one of the most prolific scientists of our time due to the media.

I'm not even sure if you can really make a firm distinction between how and why in some of these cases.
Why are quantum particles obeying Born rule? vs. How are quantum particles obeying Born Rule? Doesn't seem to be an important distinction

 

[marcus]

Just as an example: there is no satisfactory answer to "why are people buying Hawking's books?", whereas "how they are buying them?" can be answered quite easily.

 

[martinbn]

Wtf how could you ever make such a far fetched comparison?
You do not have a quantum theory to explain quantum phenomena. All you've got is math.
You do not have a way of reconciling QM with GR, all you have is math that ends up not working.

Unless you can explain exactly what is going on in QM, you should really ask yourself why you dismiss 't Hoofts attempts at finding the true answer.

You misunderstood me. I don't dismiss anything. All I am saying is that even if he can get things to work, his alternative to QM will be a bad explanation because the explanation it provides is that QM is not complete, there are hidden variable, and things are set up in such a way so that the outcomes of any experiment to be so that it appears that QM is right even though it is not. How can you consider this a satisfactory answer to your favorite "why" question!

 

[Quantumental]

You misunderstood me. I don't dismiss anything. All I am saying is that even if he can get things to work, his alternative to QM will be a bad explanation because the explanation it provides is that QM is not complete, there are hidden variable, and things are set up in such a way so that the outcomes of any experiment to be so that it appears that QM is right even though it is not. How can you consider this a satisfactory answer to your favorite "why" question!

Because he'd be solving the GR + QM problems at the same time. This is afterall his motivation. He didn't sit down like "**** that Schroedinger Cat, I need to find a solution", this was just a side effect of his work.

Yeah it'll be "hidden variables" or rather a fundamental theory that is sub-quantum. Nothing wrong with that, it'll solve the problems of QM and be a complete theory (if it works), how is this bad?
Yeah the theory will make QM *seem* right, but only approximately. How is this a problem? This is the story of all theories. QM make classical physics seem right too, until you take a deeper look. It would be the same here.

 

[Haelfix]

The reason people are skeptical, is that his CA model is a little bit hokey. The dynamics are decidedly classical (alla Bohm), but the states are essentially quantum. Eg he puts Quantum mechanics in by hand, including the projection postulate.

Further he does tremendous violence to the Hilbert space, and requires highly non local conspiracies to take place in order to bypass Bells theorem and similar thought experiments. These conspiracies in his model seems to me to not be between typical QM correlations but actual classical objects, which almost surely implies a violation of Lorentz invariance as well.

There is a reason QM has survived decades of precision tests and thousands of thought experiments done by skeptical physicists... It's structure is very rigid and doesn't seem amenable to change. Even changing it at the smallest of scales very quickly leads even the greatest theorists into a world of pain, which is what 'T Hooft is struggling with at the moment.

 

[Demystifier]

Why is there virtually no interest whatsoever when one of the greatest physicsts of all time is speaking his mind and providing several technical papers about his ideas?!

Because Witten didn't say that these stringy papers are cool.

 

[negru]

Because he'd be solving the GR + QM problems at the same time.

It's not immediately obvious to me that a new theory of QM will solve these problems.

Identical problems appear for other theories as well, like chiral perturbation theory, not just GR. They are all solved by realizing the theory is not complete and adding new degrees of freedom.

If a new theory of QM will solve the GR issues in some other way, I think it is safe to assume it will just be wrong.

 

[martinbn]

Because he'd be solving the GR + QM problems at the same time. This is afterall his motivation. He didn't sit down like "**** that Schroedinger Cat, I need to find a solution", this was just a side effect of his work.

Yeah it'll be "hidden variables" or rather a fundamental theory that is sub-quantum. Nothing wrong with that, it'll solve the problems of QM and be a complete theory (if it works), how is this bad?
Yeah the theory will make QM *seem* right, but only approximately. How is this a problem? This is the story of all theories. QM make classical physics seem right too, until you take a deeper look. It would be the same here.

I am not sure if it is clear that he will be solving QM+GR, but what is clear is that his motivation is completely different. He has said that he is discussed by the many world interpretation and the Bohmian mechanics, I guess he doesn't like the other interpretations very much either, and that's his motivation.

You still misunderstand me. I am not against people trying to do what he is attempting. I am for it, if nothing else a no-go theorem may come out of it. I am only saying that it is not a good explanation. If you have a hidden variable theory that explains better than QM it should explain the apparent quantumness of the world, and as far as I can see his does not.

 

[mitchell porter]

So what are the unresolved issues in QM that need extra understanding? Non philosophical ones please, just ones that make sense.

The deficiency of QM, that I think ought to be most accessible to someone reared in the Copenhagen interpretation, is that it doesn't tell you which observables are the ones that actually take definite values. This is most familiar from consistent histories, where you have infinitely many choices about how to coarse-grain. Quantum theory gives you a choice of observables, but they don't commute, and the user of QM just chooses which subset of observables they care about on a given occasion.

It might seem that the physical analysis of measurement processes explains this. The microscopic observables we care about on any given occasion, are the ones which get magnified into macroscopically observable quantities, and the physics of the measurement interaction explains why and how it's one particular set of microscopic observables, rather than another, which is tracked by the macroscopic apparatus.

However, this is only an explanation at the level of wavefunctions. You still have the usual freedom to say which microscopic observables actually take values, and the usual limitation that they can't all take values at the same time (and still obey the rules of QM). State preparation is a macroscopic procedure, reading out measurements is also a macroscopic procedure, and the correlations between these macroscopic observables will still be the same, regardless of how you microscopically coarse-grain what happens between one and the other.

Here I have assumed the usual Copenhagen view, that observables are what's real, not wavefunctions. If you want to reify wavefunctions, that's a different choice and leads to different problems. But I think the usual Copenhagen view gives you the clearest perspective on quantum mechanics, so long as you add that quantum mechanics is obviously incomplete. It's the attempt to rationalize quantum mechanics as a complete description of reality, which makes people tie themselves in philosophical knots.

This is different from saying that quantum mechanics is the best predictive algorithm possible. That may be true. But you can still ask what model of reality best explains quantum mechanics. One way to ease into that question, is to consider maximally coarse-grained consistent histories, something that Gell-Mann and others studied in the late 1980s I believe. That line of inquiry foundered because there were too many possibilities and no criterion for favoring one coarse-graining over another. There's also the more subtle "problem" that you still need the wavefunction of the universe to explain or make sense of the coarse-graining, which runs against the wavefunction antirealism of Copenhagen.

What would be interesting, is a maximal coarse-graining (or at least, a relatively fine-grained one) where the sequence of values taken by observables could be motivated by a dynamical rule that didn't require the wavefunction as an independent causal agent, the way it normally is e.g. in Bohmian mechanics. This is one reason why 't Hooft's new work is interesting, because the "histories" have a simple causal rule and it's even deterministic.

I agree substantially with what Haelfix said; in particular, I think 't Hooft is just wrong thinking that he can get around Bell's theorem somehow. If you're going to get QM from local determinism, that can't be locality in phenomenological space-time, it would have to be locality in some other "space" related to this one by a nontrivial transformation. But 't Hooft's 2012 work is nonetheless a step in a new direction for subquantum realism, potentially a step towards reality, and definitely noteworthy for anyone keeping track of possible realist explanations of QM.

 

[John86]

you could also say that deterministic theories explain classical world best.
But maybe if there is ever to be a next theory it will transcend Quantum physics in indeternism.
Like the relative locality framework

i don't see our future theories to be in any way more classical

 

[atyy]

People interested in that paper might also be interested in those:
http://arxiv.org/abs/hep-th/0702060
http://arxiv.org/abs/hep-th/0605250
http://arxiv.org/abs/hep-th/0512186

From your Bohmian QFT stuff, can you reconstruct QM in the bulk from Bohmian boundary QFT, analogous to what these guys are trying for bulk observables?
http://arxiv.org/abs/1102.2910
http://arxiv.org/abs/1201.3666

BTW, I see a bunch of papers by Struyve about Bohmian QFT too. But you don't cite him in your recent review (the chapter for Oriols and Mompart's book), nor does he cite you. Are the approaches completely different?

 

[Demystifier]

From your Bohmian QFT stuff, can you reconstruct QM in the bulk from Bohmian boundary QFT, analogous to what these guys are trying for bulk observables?
http://arxiv.org/abs/1102.2910
http://arxiv.org/abs/1201.3666

I don't know, essentially because my understanding of AdS/CFT is quite deficient.

BTW, I see a bunch of papers by Struyve about Bohmian QFT too. But you don't cite him in your recent review (the chapter for Oriols and Mompart's book), nor does he cite you. Are the approaches completely different?

Yes, they are very different. In my review I study particle beables, while he studies field beables. Roughly, this is like comparing perturbative string theory with string field theory; they have different sets of starting assumptions and deal with different types of technical problems.

 

[atyy]

I don't know, essentially because my understanding of AdS/CFT is quite deficient.Yes, they are very different. In my review I study particle beables, while he studies field beables. Roughly, this is like comparing perturbative string theory with string field theory; they have different sets of starting assumptions and deal with different types of technical problems.

Can BM handle condensed matter models like those in http://arxiv.org/abs/1210.1281? They are non-relativistic spins on a lattice, but they produce emergent relativistic QFTs.

There is some effort, which I find very interesting, to connect these with string theory via AdS/CFT.
http://arxiv.org/abs/0905.1317
http://arxiv.org/abs/1209.3304

 

[Demystifier]

Can BM handle condensed matter models like those in http://arxiv.org/abs/1210.1281? They are non-relativistic spins on a lattice, but they produce emergent relativistic QFTs.

Yes it can. If that does not convince you, then you would help me to give a better answer by explaining why exactly do you think that it might not?

 

[unusualname]

I am not asking WHY, I am asking HOW, how does the Universe obey the Born Rule if there are no causes that makes it?

Suppose you have a single particle (with no internal structure) in the entire universe. What would it do? You'd probably say nothing, but let's describe it by exp(ix), where x is real, and allow x to spontaneously change (otherwise nothing ever happens)

We could perhaps just describe it by x, and suppose that spontaneously changes instead, but that may not describe nature (see below)

Now suppose a second particle is added to the universe. What happens? Shall we allow both particles to spontaneously change in an unlawful manner? That wouldn't be a very interesting universe. INSTEAD, let's ASSUME that when one of the particles SPONTANEOUSLY changes the universe evolves to a new state according to a rule connecting both particles, eg a 2x2 matrix equation.

Now add ~10^80 particles with similar rules.

The Born rule falls out as conserved quadratic form (which is why we need to represent states by complex numbers - real representations don't match our observations) - and experiment shows it's not a quartic or higher power.

't Hooft just wants a deterministic system - but at the individual/tiniest particle level what possible deterministic law can there be - I mean, what's an individual particle meant to do? Hence we assume no causality at the level of individual particles/cells/states - and wow, you just need the Schrödinger Equation and a Hamiltonian matrix to make it all work.

That's QM, formulated ~1927

 

[Quantumental]

Suppose you have a single particle (with no internal structure) in the entire universe. What would it do? You'd probably say nothing, but let's describe it by exp(ix), where x is real, and allow x to spontaneously change (otherwise nothing ever happens)

1st stop right here: so.. we are going to assume that it spontaneously changes, because otherwise nothing happens?
So you look at the universe and see that things happen, so therefore you conclude that change occurs, but how can you say that it is spontaneous?

Now suppose a second particle is added to the universe. What happens? Shall we allow both particles to spontaneously change in an unlawful manner? That wouldn't be a very interesting universe. INSTEAD, let's ASSUME that when one of the particles SPONTANEOUSLY changes the universe evolves to a new state according to a rule connecting both particles, eg a 2x2 matrix equation.

2nd stop: again why do you assume this?
This is my entire argument: if they are "random/indeterministic/spontaneous" they should not follow laws. Laws need explanation.
The law of gravity means there is this thing called gravity affecting the universe... The Born Rule cannot just *be* without actually *being*.

't Hooft just wants a deterministic system - but at the individual/tiniest particle level what possible deterministic law can there be - I mean, what's an individual particle meant to do? Hence we assume no causality at the level of individual particles/cells/states - and wow, you just need the Schrödinger Equation and a Hamiltonian matrix to make it all work.

Gerard 't Hooft and others think that at the Planck scale / or somewhere around that length a cellular automata is taking place. Then we suddenly have a mechanism (whatever rule the CA is playing) that is deterministic which everything in the universe emerges from. Just like Game of Life

Hell you could even say de-Broglie came before 1927 and formulated his pilot wave theory, it's not hard to assume some deterministic law.

 

[unusualname]

Gerard 't Hooft and others think that at the Planck scale / or somewhere around that length a cellular automata is taking place. Then we suddenly have a mechanism (whatever rule the CA is playing) that is deterministic which everything in the universe emerges from. Just like Game of Life

Hell you could even say de-Broglie came before 1927 and formulated his pilot wave theory, it's not hard to assume some deterministic law.

The problem is, what stops the universe happening "all at once"? With a deterministic CA you have to make an additional assumption that each iteration takes some finite time-step. (Wolfram's computational speed limit of the universe idea)

Whereas if you have spontaneous "jumps" seeding each evolution step you already have a mechanism which prevents the universe all happening instantaneously. ie Nothing happens until there is a spontaneous jump, then the Universe updates every other particle/state/cell in response to whatever the spontaneous change was (via schrodinger evolution of the universe wave function)

That way you have quantum superpositions without MWI splitting, and you also have a natural delay to the universe's evolution.

However, I agree that's rather easy speculation, just pointing out that determistic CAs have to do extra work to get superpositions and a cosmic speed limit.

 

[audioloop]

Wtf how could you ever make such a far fetched comparison?
You do not have a quantum to explain quantum phenomena. All you've got is math.
You do not have a way of reconciling QM with GR, all you have is math that ends up not working.

Unless you can explain exactly what is going on in QM, you should really ask yourself why you dismiss 't Hoofts attempts at finding the answer.

agree, not dismiss for dismiss.

 

[audioloop]

It's not immediately obvious to me that a new of QM will solve these .

Identical problems appear for other theories as well, like chiral perturbation theory, not just GR. They are all solved by realizing the theory is not complete and adding new degrees of .

If a new theory of QM will solve the GR issues in some other way, I think it is safe to assume it will just be wrong.

maybe not QM, rather beyond QM, i.e. other theory that predict the results of QM and beyond (maybe QM, GR, SR at once and more) as relativity supersedes Newtonian physics.

 

[atyy]

Yes it can. If that does not convince you, then you would help me to give a better answer by explaining why exactly do you think that it might not?

Thanks for your answers. I'm still learning about BM so just have a bunch of stupid questions to ask, and will probably take a long time to understand the details. It is definitely helpful to have expert opinion like yours to guide my "homework".

Anyway, if you are right, then 't Hooft if wrong, ie. there is no foundations problem in QM, QFT, string theory?

 

[Demystifier]

Anyway, if you are right, then 't Hooft if wrong, ie. there is no foundations problem in QM, QFT, string theory?

I think the main property of the Bohmian approach which 't Hooft does not like is non-locality. There are very general theorems which say that non-locality is unavoidable, but 't Hooft tries to avoid some assumptions of these theorems. See also
https://www.physicsforums.com/blog.php?b=3622

 

[RUTA]

I think the main property of the Bohmian approach which 't Hooft does not like is non-locality. There are very general theorems which say that non-locality is unavoidable, but 't Hooft tries to avoid some assumptions of these theorems. See also https://www.physicsforums.com/blog.php?b=3622

There are two things one can abandon to avoid the conclusion of the Bell inequality -- locality and statistical independence. So, if he doesn't like non-locality, then he must abandon statistical independence, e.g., as is done with time-symmetric interpretations.

 

[atyy]

Is cosmology a problem for Bohmian mechanics, since if there is only one universe, there is no distribution of initial conditions?

The Valentini essays that Ilya mentioned in the other thread seem to provide a way out, ie. non-equilibrium Bohmian mechanics, so that equilibrium BM = standard QM emerges from the dynamics. But I don't think there's any concrete proposal for this at the moment.

 

[Demystifier]

Is cosmology a problem for Bohmian mechanics, since if there is only one universe, there is no distribution of initial conditions?

This is like asking is cosmology a problem for classical mechanics, because in classical statistical mechanics applied to one universe there is no distribution of initial conditions. The answer, of course, is that it is not a problem. Even though statistical reasoning plays a role in both classical and Bohmian mechanics, both theories are fundamentally deterministic, not statistical.

 

[atyy]

This is like asking is cosmology a problem for classical mechanics, because in classical statistical mechanics applied to one universe there is no distribution of initial conditions. The answer, of course, is that it is not a problem. Even though statistical reasoning plays a role in both classical and Bohmian mechanics, both theories are fundamentally deterministic, not statistical.

But just as we still don't know how statistical mechanics arises from classical mechanics, then we still don't know how quantum mechanics arises from Bohmian mechanics?

In trying to get stat mech from classical mechanics, there's usually some coarse graining, and there have been proposals for chaos to be involved, or involving canonical typicality or eigenstate thermalization (but I think those assume the Born rule). What are the corresponding ideas for getting QM from Bohmian mechanics?

 

[mitchell porter]

But just as we still don't know how statistical mechanics arises from classical mechanics

This is a peculiar statement. It becomes even more peculiar when I look at the links you provided to back it up, because two of them are about quantum mechanics. (And the third one, after the part devoted to study of Boltzmann's words, is about technicalities to do with chaos and defining entropy away from equilibrium.)

If we have a classical equation of motion, then we can also figure out how a probability distribution over classical states will evolve. So where is the foundational problem? Consulting Wikipedia, it seems the basic challenge is to physically motivate the probability distributions that one might use. On what grounds do I say that a uniform distribution over a certain set of microstates is an appropriate description for a system at equilibrium? Well, that's a bit like the general problem in probability theory, of where you get your prior from.

Anyway, I would like you to explain the difference between "statistical mechanics of classical systems" and "mathematics of probability distributions over classical systems". I hope we can agree that there's no deep mystery about the latter, if you already have the classical equation of motion. So any foundational problem of "classical statistical mechanics" must arise somewhere else. But where, exactly?

In the case of what I call "cosmological Bohmian mechanics", what one would need (in my opinion) is a hypothesis about cosmic initial conditions (both for the pilot wave and for the classical system it guides), such that, if you looked at the "reduced pilot waves" associated with small sets of "classical" degrees of freedom, in the subsequent history of the Bohmian universe, the "demographics" of this association would resemble the Born rule. E.g. if you picked out a random electron, from somewhere in the space-time history, and looked at the reduced density matrix associated with that degree of freedom (derived from the universal pilot wave at that time), you should expect a Born-like probability relation between its position and its reduced density matrix. I have no idea how far people like Valentini have gone towards such a goal. But the need to gauge-fix in Bohmian gravity (the shift and lapse functions) seems a far more important difficulty, anyway.

 

[atyy]

This is a peculiar statement. It becomes even more peculiar when I look at the links you provided to back it up, because two of them are about quantum mechanics. (And the third one, after the part devoted to study of Boltzmann's words, is about technicalities to do with chaos and defining entropy away from equilibrium.)

If we have a classical equation of motion, then we can also figure out how a probability distribution over classical states will evolve. So where is the foundational problem? Consulting Wikipedia, it seems the basic challenge is to physically motivate the probability distributions that one might use. On what grounds do I say that a uniform distribution over a certain set of microstates is an appropriate description for a system at equilibrium? Well, that's a bit like the general problem in probability theory, of where you get your prior from.

Anyway, I would like you to explain the difference between "statistical mechanics of classical systems" and "mathematics of probability distributions over classical systems". I hope we can agree that there's no deep mystery about the latter, if you already have the classical equation of motion. So any foundational problem of "classical statistical mechanics" must arise somewhere else. But where, exactly?

In the case of what I call "cosmological Bohmian mechanics", what one would need (in my opinion) is a hypothesis about cosmic initial conditions (both for the pilot wave and for the classical system it guides), such that, if you looked at the "reduced pilot waves" associated with small sets of "classical" degrees of freedom, in the subsequent history of the Bohmian universe, the "demographics" of this association would resemble the Born rule. E.g. if you picked out a random electron, from somewhere in the space-time history, and looked at the reduced density matrix associated with that degree of freedom (derived from the universal pilot wave at that time), you should expect a Born-like probability relation between its position and its reduced density matrix. I have no idea how far people like Valentini have gone towards such a goal. But the need to gauge-fix in Bohmian gravity (the shift and lapse functions) seems a far more important difficulty, anyway.

Sure, I did note the latter two links were about QM. The general question is how does stat mech arise? And yes, the question is how does one justify the initial distribution. Valentini's approach is indeed in the same spirit of my question - have any concrete examples been worked out?

 

[Demystifier]

But just as we still don't know how statistical mechanics arises from classical mechanics, then we still don't know how quantum mechanics arises from Bohmian mechanics?

In trying to get stat mech from classical mechanics, there's usually some coarse graining, and there have been proposals for chaos to be involved, or involving canonical typicality or eigenstate thermalization (but I think those assume the Born rule). What are the corresponding ideas for getting QM from Bohmian mechanics?

While there are some uncertainties regarding how statistical mechanics arises from classical mechanics, I think those uncertainties are not very serious. Anyway, the situation is very similar with Bohmian mechanics.