Undergrad Sean Carroll podcast on many worlds interpretation

[Minnesota Joe]

Exactly. I will post the link again but many worlds, rightly viewed, is quite reasonable

It's now a lot like decoherent histories but without considering the other histories as 'real' or if like Gell- Mann you take real to mean one equal footing, there is actually not much if any difference..

Interesting, but Gell-Mann's doesn't really dig very deeply in the video. He just asserts that which branch is real "is determined only probabilistically".

The MWI people might agree that the two branches are indeed on "equal footing" to the point of being indistinguishable. The wave function doesn't collapse. So it is only bias to call one real (the one I'm in now) and the other fake. They might invoke a sort of Copernican principle. I don't know for sure how the would respond though and I could be wrong. And I don't think MWI has a good explanation for probabilities either however.

But some legitimate scientists who really should know better like to sensationalize things and even introduce consciousness which of course there is no reason to. Of course like all interpretations it may be true but I fear Sean has succumbed to populism.

That's the thing. I don't think Carroll is sensationalizing anything. I think he is dead serious about MWI as a research project and seeing how far he can take it. He agrees it is radical relative the physics community. He is quite up front about the possibility it will crash and burn, even though he believes it is the "correct" interpretation in the sense of the most probably true. His book is supposed to explain why (I haven't read it yet). It is also--again, according to him--the most popular interpretation his field of cosmology.

These days I only recommend two popularizations - Feynman - QED and Susskind's book

Hey, thanks for the reminder. I'd forgotten about QED and need to read it. Which Susskind book is that?

 

[bhobba]

I invoke the Bohm model because I do not know of any other way of defining what a branch is in MWI.

Formally a branch is simply a series of projection operators. Its similar in Consistent Histories:
http://quantum.phys.cmu.edu/CHS/histories.html
Thanks
Bill

 

[Minnesota Joe]

I was responding to Vanhees.

Oh, ha, my comment wasn't aimed at you or anyone here! I was originally complaining about something Sean Carroll claimed. He claims interpretations other than MWI have to "work hard" to get rid of the many worlds.

 

[bhobba]

Which Susskind book is that?

https://www.amazon.com/dp/0465062903/?tag=pfamazon01-20

Thanks
Bill

 

[Minnesota Joe]

https://www.amazon.com/dp/0465062903/?tag=pfamazon01-20

Ok thanks, I read that one. It is good, but it seems mostly like standard "shut up and calculate" QM to me. Susskind takes the collapse as a postulate and doesn't give it enough discussion in my opinion. He's very dismissive or indifferent about quantum foundations too (not in the book where I can't remember him even mentioning it, but elsewhere). But it does contain an interesting discussion of EPR and locality that I need to reread.

 

[DarMM]

I was known to argue a lot that collapse is not required by QM. I spent a lot of posts discussing what collapse was. But then I realized really it doesn't matter. We know from QM that when you observe a system you get an eigenvalue of the observable

If you measure a PVM, for a POVM in general you do not. I don't think that affects your main point much, just a technicality.

I'm not sure what you mean by QM not needing collapse. I mean after an observation you update the state, right?

 

[Minnesota Joe]

I invoke the Bohm model because I do not know of any other way of defining what a branch is in MWI.

I think it means something more radical to MWI proponents. Ones like Carroll and the late DeWitt I mean.

A branch is a continuous function of its initial value settings. There are a continuum many almost exact copies of Sean Carroll in different branches, if we like to think in that way.

The Copenhagen interpretation speaks of "a collapse of the wave function". If the scientist works in an isolated laboratory, no collapse can happen. We cannot discard parts of the whole wave function of an isolated system. Why does the scientist observe a collapse?

In the Bohm model he does measure one exact outcome from a quantum experiment. He did not know the values of the hidden variables and could not predict the outcome, only a probability distribution. If he is in the Copenhagen camp, then in his mind he has a metaphor that a wave function has collapsed.

Thus, both the branching and the collapse are just metaphors in the mind of a scientist.

You are using it more metaphorically I guess, which I wasn't complaining about. I was complaining about Sean Carroll's position which--I think--are that worlds are real. The are actual. Whatever you want to call it. And then he wants to argue that all the other interpretations have them too, but they just work hard to get rid of them.

 

[Michael Price]

@akvadrako,
There is no splitting [in MWI].

It depends on how you define splitting, of course. At the end of the day you have the evolution of the universal wavefunction, and all possible worlds or histories are embedded within it. There are histories with advanced non-avian dinosaurs and victorious Nazis. There are none of these in our world, so they must have split off somehow, by any sensible definition of splitting.

 

[DarMM]

Very interesting paper. I only don't understand, how possibility (a) (which is the standard assumption if I understand it right since a measurement always extends the system by coupling it to the measurement device + "environment") prevents the heat from being infinite given the arguments before

First, yes (a) is the standard assumption, i.e. Copenhagen/Minimal.

As for why, Cabello summarizes it well on page 3:

In contrast, if probabilities are not determined by intrinsic properties of the system, then measurement out-comes are created randomly when the observables are measured, without any need to overwrite information in the system and therefore without the system dissipating heat due to Landauer’s principle

Basically the extra heat cost is from overriding the information stored in pre-existent determined variables. No such variables means no extra heat cost. That's how Copenhagen/Minimal gets out of it.

Now since this cost comes from overriding pre-existent determined information one can also avoid it by supposing that the amount of information in the system is infinitely large, since then you never need to override the information. This is what happens in Bohmian Mechanics and Many-Worlds where even a qubit has an infinite amount of deterministic information in it.

Since I imagine you'll find this is kind of strange it's easiest to illustrate with a dice. Obviously for a dice we have $\Omega = \left\{1,2,3,4,5,6\right\}$ as the sample space and some function $p\left(\omega\right)$ as the probability distribution satisfying:
$$\sum_{i}p\left(\omega_{i}\right) = 1$$
(All this is known to you of course, just to set it up)

Many Worlds and Bohmian mechanics essentially correspond to saying $p\left(\omega\right)$ is a physical degree of freedom, rather than just being a set of probabilities. If you claim that, then the dice goes from having a finite set of physical states to an infinite amount. As each $p\left(\omega\right)$ is a physical configuration.

 

[Michael Price]

[Lubos] makes one interesting claim in that post that I have not seen discussion of in the literature: he claims that interpretations that say the wave function is real (his "R2" category) predict that heat capacities of quantum systems should be huge, many orders of magnitude larger than we actually measure them to be, because the wave function even for a single molecule would include so many physical degrees of freedom (for example, the wave function of a single qubit has to encode enough information to specify a point on the Bloch sphere). Does anyone know of any more rigorous treatment of such a claim in the literature?

Lubos' heat capacity claim is just so bonkers it is hard to see why he pushes it. The problem is that Lubos is so far gone into confirmation bias against MWI that even the most nutty statements seem reasonable to him if they are anti-MWI. MWI is an interpretation and interpretations don't change physical results, such as specific heat capacities.

 

[PeterDonis]

Lubos' heat capacity claim is just so bonkers it is hard to see why he pushes it.

Why do you think it is bonkers?

MWI is an interpretation and interpretations don't change physical results

But Motl's argument is basically that, if "MWI" includes the claim that "the wave function is physically real", then it can't be just an interpretation; it must actually make a different prediction from standard QM, the one about heat capacities. His argument might be wrong, but it can't be rebutted by just saying "MWI is an interpretation", because whether or not the MWI actually is just an interpretation is one of the points at issue.

 

[akvadrako]

But some legitimate scientists who really should know better like to sensationalize things and even introduce consciousness which of course there is no reason to. Of course like all interpretations it may be true but I fear Sean has succumbed to populism.

I'm not sure where Sean invokes consciousness (I would be happy to get a link), but I do think it unfortunately does have some role to play in MWI. That's because while the objective ontology of MWI is clear, conscious agents are just subsystems embedded in a larger system. So if you care about connecting the objective description to what is measurable, what some would call science, you need to understand the dynamics of consciousness and how it relates to the environment.

The big question for me is how much of a role self-locating uncertainty plays.

 

[Heikki Tuuri]

It depends on how you define splitting, of course. At the end of the day you have the evolution of the universal wavefunction, and all possible worlds or histories are embedded within it. There are histories with advanced non-avian dinosaurs and victorious Nazis. There are none of these in our world, so they must have split off somehow, by any sensible definition of splitting.

Let us think about that using the Bohm model. We first pick a wave function for the > 10^80 particles that explode in the Big Bang. The wave function stays fixed in this thought experiment.

The wave function allows also extremely improbable developments.

Each branch is distinct. It is a deterministic function of the initial values. Is there a branch where Earth develops almost exactly like our own, but then 100 million years ago dinosaurs start building railroads?

If the movement of particles were truly random, then, of course, we could shuffle the particles as we like and conjure up such dinosaurs.

But we picked a fixed wave function at the Big Bang. It is not clear at all that we can choose initial values such that intelligent dinosaurs arise.

We can construct a suitable wave function by making one for the dinosaur civilization and then calculating back to the Big Bang. But we may have picked a different wave function at the start.

I had thought that MWI allows all kinds of strange worlds, but that might not be the case.

 

[Michael Price]

Exactly. I will post the link again but many worlds, rightly viewed, is quite reasonable


It's now a lot like decoherent histories but without considering the other histories as 'real' or if like Gell- Mann you take real to mean one equal footing, there is actually not much if any difference..

But some legitimate scientists who really should know better like to sensationalize things and even introduce consciousness which of course there is no reason to. Of course like all interpretations it may be true but I fear Sean has succumbed to populism.

These days I only recommend two popularizations - Feynman - QED and Susskind's book

Thanks
Bill

Murray GM is just playing with words and evading the issue: are the Everett world's real? MWI says they are. It is not popularization to state this. Indeed it is condescending not to say this - or else you are saying the public need protection from the truth because they can't handle it?

Why do you think it is bonkers?
But Motl's argument is basically that, if "MWI" includes the claim that "the wave function is physically real", then it can't be just an interpretation; it must actually make a different prediction from standard QM, the one about heat capacities. His argument might be wrong, but it can't be rebutted by just saying "MWI is an interpretation", because whether or not the MWI actually is just an interpretation is one of the points at issue.

The putative reality of the wavefunction does not change any measurements made - it won't move a needle on a gauge, for example. It changes whether the needle is ghostly or real, but not its dynamics, and is the dynamics that determines the needle's position, on how it is correlated with whatever it is measuring.

There is universal agreement in the literature that interpretations do not differ about empirical results. Our experiences are embedded in the correlations within the universal wavefunction. These correlations do not depend on ontological status of the UW. (I do not think there is any published claim to the contrary.)

 

[Minnesota Joe]

I'm not sure where Sean invokes consciousness (I would be happy to get a link), but I do think it unfortunately does have some role to play in MWI. That's because while the objective ontology of MWI is clear, conscious agents are just subsystems embedded in a larger system. So if you care about connecting the objective description to what is measurable, what some would call science, you need to understand the dynamics of consciousness and how it relates to the environment.

To my knowledge Carroll doesn't invoke consciousness. Nor does he endorse a special role for consciousness. I think he has argued that consciousness emerges from physics, not the other way around. (I don't want to speak for him but I seem to remember him talking along those lines.)

 

[Michael Price]

Let us think about that using the Bohm model. We first pick a wave function for the > 10^80 particles that explode in the Big Bang. The wave function stays fixed in this thought experiment.

The wave function allows also extremely improbable developments.

Each branch is distinct. It is a deterministic function of the initial values. Is there a branch where Earth develops almost exactly like our own, but then 100 million years ago dinosaurs start building railroads?

If the movement of particles were truly random, then, of course, we could shuffle the particles as we like and conjure up such dinosaurs.

But we picked a fixed wave function at the Big Bang. It is not clear at all that we can choose initial values such that intelligent dinosaurs arise.

We can construct a suitable wave function by making one for the dinosaur civilization and then calculating back to the Big Bang. But we may have picked a different wave function at the start.

I had thought that MWI allows all kinds of strange worlds, but that might not be the case.

Personally I find the Schrödinger picture, where the wavefunction evolves with time, easier to use. It does not really matter, but I was thinking of worlds where dinosaurs survived to the modern day because the asteroid missed the Earth. Whether such a scenario is possible, nay inevitable, reminds of the discussions we had about Dr.Chinese being president. I say yes, but the discussion became very weird and heated and I will leave it at that.

 

[PeterDonis]

The putative reality of the wavefunction does not change any measurements made

It depends on what "reality of the wavefunction" implies. But phrasing it that way might be misleading; let me try rephrasing what I take to be Motl's argument (or at least a "steelman" version of it that appears to me to be worth considering) without using that terminology, at least at the start.

Consider a single qubit. If we don't know its state, the usual assumption in quantum statistical mechanics is that its entropy is of order unity (in "natural" units where "unity" means $k_B$). This means it has approximately 1 bit's worth of information in its unknown state.

However, specifying the exact quantum state of a qubit takes much more than 1 bit of information, since it is equivalent to specifying an exact point on the Bloch sphere. So the question is, does that exact quantum state refer to real, physical degrees of freedom? In other words, does it refer to real, physical information which, if we don't know the qubit's state, should be reflected in the entropy?

Motl's argument is basically that the MWI, or indeed any interpretation that treats the wave function as real, must answer "yes" to the above question. If that is true, then, whether the proponents of such interpretations realize it or not, or have even considered it or not, their interpretation actually isn't just an interpretation; it makes a testable physical prediction about entropy that is different from standard quantum statistical mechanics.

There is universal agreement in the literature that interpretations do not differ about empirical results.

There is agreement that that is the definition of an interpretation. But you can't establish that a particular thing that claims to be an interpretation, actually is one, by just quoting the definition.

 

[Minnesota Joe]

Just to put money to my claim that Sean Carroll takes MWI very seriously, I looked up a paper he mentions in on of his podcasts.

Mad-Dog Everettianism: Quantum Mechanics at Its Most Minimal, Sean M. Carroll, Ashmeet Singh

 

[Quanundrum]

The main annoyance I have (besides the obvious problem of probability) is how people like Sean Carroll make these extravagant claims in vivid language. Hyperbole is littered throughout the book and interviews he has done in the press. He adamantly claim that you live in multiple worlds and extrapolates wild claims that just confuses the general public. Even if you make the assumption that the Born Rule will one day be derived and MWI is correct, you don't know whether you are living in a divergent (all worlds exist from big bang in a separated fashion) multiverse or a splitting one (where you actually branch off from overlapping worlds). At first this may seem like a technical nuance, but it sits at the heart of all these radical claims. Quantum Immortality automatically goes out the window in the former, as does virtually all talk of you going into multiple futures at every instant.

Furthermore, the claim that "MWI is simply taking the equations seriously man" is lazy. To quote Mitchell Porter from Sean's latest blog post comment section

Mitchell Porter says:
September 21, 2019 at 6:04 pm
So when an observable has two eigenvalues, how many worlds are there – two? Two million? A continuum? Does the number of worlds relate to the probability of an observation? If yes, please explicitly exhibit the decomposition of the wavefunction into the appropriate worlds. And if the probabilities don’t come from the number of worlds, please explain where they do come from.

This is never answered by any of these dogmatic Everettians

 

[Michael Price]

The main annoyance I have (besides the obvious problem of probability) is how people like Sean Carroll make these extravagant claims in vivid language. Hyperbole is littered throughout the book and interviews he has done in the press. He adamantly claim that you live in multiple worlds and extrapolates wild claims that just confuses the general public. Even if you make the assumption that the Born Rule will one day be derived and MWI is correct, you don't know whether you are living in a divergent (all worlds exist from big bang in a separated fashion) multiverse or a splitting one (where you actually branch off from overlapping worlds). At first this may seem like a technical nuance, but it sits at the heart of all these radical claims. Quantum Immortality automatically goes out the window in the former, as does virtually all talk of you going into multiple futures at every instant.

Furthermore, the claim that "MWI is simply taking the equations seriously man" is lazy. To quote Mitchell Porter from Sean's latest blog post comment section
This is never answered by any of these dogmatic Everettians

https://www.quora.com/How-does-the-...ding-to-the-Born-rule/answer/Michael-Price-29

 

[Elias1960]

He claims interpretations other than MWI have to "work hard" to get rid of the many worlds.

This sounds like an echo of the "BM is many worlds in denial" argument made by Deutsch iirc. It presupposes, of course, that MWI makes sense at all (which most of those preferring other interpretations doubt). Then it presupposes that the other interpretations use the ideas of MWI, which they don't.

So, the MWI guys completely miss the role of the configuration trajectory $q(t)\in Q$ in BM. It defines what we are ourselves, it defines what we see in the macroscopic domain, and the effective wave function of a quantum system is defined by that trajectory of the measurement device,
$$ \psi(q_{sys},t) = \psi_{all}(q_{sys},q_{device}(t),t). $$
Instead, MWI seems to think that BM guys would have a problem to observe any trajectories at all, all that can be measured are wave functions, as it has to be in MWI, while in BM the wave function can be defined only based on observing the trajectories.

 

[vanhees71]

The maximum entropy of one bit is $k_B \ln 2$ in the standard definition. But a qubit is not a bit. A bit can only have two values, $0$ or $1$. A qubit's wave function can have any value on the Bloch sphere. Measuring a qubit can only result in one of two values, but a single measurement on a qubit is not sufficient to tell you its exact wave function. Strictly speaking, it takes an infinite number of measurements (on an ensemble of identically prepared qubits) to do that.

That, at least, is how I understand Motl's argument, and it seems at least worth enough consideration for somebody to have written a paper on it at some point; that's why I asked if anyone knows of such a paper.

The state of maximum entropy for a qubit is $\hat{\rho}=\hat{1}/2$ and the von Neumann entropy is $S=-k_{\text{B}} \mathrm{Tr} \hat{\rho} \ln \hat{\rho}=k_{\text{B}} \ln 2$.

Of course, state determination can usually not achieved by a single measurement, but what has this to do with the (useful) definition of von Neumann entropy?

 

[vanhees71]

If you measure a PVM, for a POVM in general you do not. I don't think that affects your main point much, just a technicality.

I'm not sure what you mean by QM not needing collapse. I mean after an observation you update the state, right?

The question is for what purpose you are "updating the state". Collapse in the sense of some Copenhagen flavors is a clear contradiction of the standard local relativistic QFT constructions, which don't allow for an instantaneous action at a distance. So it cannot be part of any interpretation that pretends to interpret standard local QFTs.

But indeed, there's no need for a collapse. All there is are probabilities calculated using Born's rule, given the state preparation. These probabilities refer to PVM's, i.e., precise measurements. To blur the subject with weak measurements, which can be understood when needed using the minimal interpretation, but are not important for the interpretational argument at all.

Now, when do you need an "update of the state"? It's when you do a preparation. A preparation, as any other manipulation we can do, is a local interaction between some "object" with some matter set up by us to prepare it. Since these manipulations are local, it's no collapse either, and the collapse is not needed to understand the preparation procedure.

Then there's simply a (PVM) measurement, where the measurement device provides a pointer reading, we get notice of (maybe decades later after the experiment is dismantled already) by reading off some measurement protocol (nowadays usually a computer file from some detector). There's no more mystery in that than taking notice of the outcome of the lottery drawing each Saturday. Nothing collapses here at all.

 

[DarMM]

Certainly, it's more so that I'm wondering by "collapse" does @bhobba mean just plain old state reduction, which I'd be surprised if he were saying we don't need it, or a physical nonlocal collapse which certainly we don't need.

These probabilities refer to PVM's, i.e., precise measurements. To blur the subject with weak measurements, which can be understood when needed using the minimal interpretation, but are not important for the interpretational argument at all

POVMs aren't weak measurements. Could you explain how they blur the subject? To me they add a lot. For example that most measurements are not actually quantizations of classical quantities, also POVMs are not equivalent to PVMs but are in fact more general. They're the same relation to PVMs that density matrices are to pure states. So it would be equivalent to saying density matrices "blur" the subject or are irrelevant. Rather I think they are an important part of the modern formalism.

Also I should say I was just pointing the fact out to @bhobba that the results of measurements are not always related to the eigenvalue of an observable. That's a fact, not really tied to the interpretation.

 

[akvadrako]

To my knowledge Carroll doesn't invoke consciousness. Nor does he endorse a special role for consciousness. I think he has argued that consciousness emerges from physics, not the other way around. (I don't want to speak for him but I seem to remember him talking along those lines.)

It depends what you mean by physics. If you mean objective reality then consciousness is irrelevant. If you mean what's measurable then it depends how subjective experience emerges. We have a rough idea how this works in day to day life, but things get hairy when you consider the experience of Wigner's friend or other circumstances when subjective experience differs greatly from two different viewpoints. Maybe Wallace was on the right path when we tried to derive the Born rule from decision theory, though he only considered one aspect of experience.

 

[akvadrako]

Motl's argument is basically that the MWI, or indeed any interpretation that treats the wave function as real, must answer "yes" to the above question. If that is true, then, whether the proponents of such interpretations realize it or not, or have even considered it or not, their interpretation actually isn't just an interpretation; it makes a testable physical prediction about entropy that is different from standard quantum statistical mechanics.

Motl claims this is a testable prediction but how do you propose to test it? Even if by some measure the entropy is different, it's inaccessible.

 

[vanhees71]

Certainly, it's more so that I'm wondering by "collapse" does @bhobba mean just plain old state reduction, which I'd be surprised if he were saying we don't need it, or a physical nonlocal collapse which certainly we don't need.POVMs aren't weak measurements. Could you explain how they blur the subject? To me they add a lot. For example that most measurements are not actually quantizations of classical quantities, also POVMs are not equivalent to PVMs but are in fact more general. They're the same relation to PVMs that density matrices are to pure states. So it would be equivalent to saying density matrices "blur" the subject or are irrelevant. Rather I think they are an important part of the modern formalism.

Also I should say I was just pointing the fact out to @bhobba that the results of measurements are not always related to the eigenvalue of an observable. That's a fact, not really tied to the interpretation.

For the interpretational issues we discuss you don't need the very complicated description of weak measurements by POVMs. Of course they are useful, when needed, but not for the purpose to discuss the interpretational foundations.

 

[DarMM]

For the interpretational issues we discuss you don't need the very complicated description of weak measurements by POVMs. Of course they are useful, when needed, but not for the purpose to discuss the interpretational foundations.

I don't know about that.

First I wouldn't say they are very complicated, or at least more so than PVMs. Both are ultimately just decompositions of the identity operator on a Hilbert space. A POVM is just the more general case when the elements of the decomposition don't commute. Also I wasn't speaking of weak measurements, but POVMs. POVMs are not weak measurements.

Secondly I think they do have a major impact on interpreting QM as they show you most measurements are not associated to the quantization of a classical quantity. So in many cases what we are measuring cannot be called angular momentum, position or any other quantity that appears in classical physics. Indeed in most cases we are simply measuring that POVM for which we have no name in general. That's quite a big deal in my opinion.

 

[vanhees71]

We have zillions of real-lab measurement devices described by PVMs. Recently we had the debate of POVMs, and I still have not a single example for a real-lab apparatus and its description by a POVM. The math is indeed simple enough to understand it, but the link to real measurement devices is still not clear to me.

I'm also not sure what POVMs have to do with "quantization of a classical quantity" or not. The observable algebra cannot be derived from classical physics of course. One has to rely on mathematical arguments like symmetry principles. Then you build a model and check, whether it describes real-world phenomena correctly. An example for an observable which for sure is not derivable by some "quantization of a classical quantity" is spin.

I think to make physical sense of POVMs you need no more and no less than the standard postulates and the minimal interpretation including Born's rule for the measurement in the usual PVM sense.

 

[Minnesota Joe]

This sounds like an echo of the "BM is many worlds in denial" argument made by Deutsch iirc. It presupposes, of course, that MWI makes sense at all (which most of those preferring other interpretations doubt). Then it presupposes that the other interpretations use the ideas of MWI, which they don't.

Yes, that attitude is what I had in mind. Not just BM though but all interpretations QM except MWI are in denial according to Carroll.

So, the MWI guys completely miss the role of the configuration trajectory $q(t)\in Q$ in BM. It defines what we are ourselves, it defines what we see in the macroscopic domain, and the effective wave function of a quantum system is defined by that trajectory of the measurement device,
$$ \psi(q_{sys},t) = \psi_{all}(q_{sys},q_{device}(t),t). $$
Instead, MWI seems to think that BM guys would have a problem to observe any trajectories at all, all that can be measured are wave functions, as it has to be in MWI, while in BM the wave function can be defined only based on observing the trajectories.

In de Broglie-Bohm, as I understand it, the particles are actually what you observe and you can retrodict their trajectories though you can't predict them because of ignorance of initial conditions. So if a particle goes through a Stern-Gerlach (SG) device, it matters which side of the direction of propagation axis it was on: above or below. The the SG separates the wave function into two packets, one containing the particle and one empty. Does that sound right so far? So is it the empty packet that the MWI people are calling a "world" or is it something else do you suppose?