Many-worlds: When does the universe split?

[greypilgrim]

Hi,

I don't quite understand some central points of the many-world interpretation: When does the splitting happen, and is there superposition in a single universe? I see two alternatives:

1. There is superposition in a single, and this universe splits when we measure. This means we can start with only one universe which branches out more and more.
2. There is no superposition in a single universe. We just find out about which universe we're in when we measure.

Both of these alternatives are not satisfactory to me at all. In the first, I can't see the "advantages" over Kopenhagen, because the measurement is still a very special process, namely it causes a whole universe to split!
The second is even weirder, it would imply that from the beginning there are already enough universes (an enormous number) for all possible superpositions to come. But how do these universes know in which basis we are going to measure? We would have to give up every bit of free will. On the positive, a measurement would really be nothing special at all.

Which of the two is more accepted (or are there more interpretations)? Both seem very implausible to me.

 

[kith]

In the first, I can't see the "advantages" over Kopenhagen, because the measurement is still a very special process, namely it causes a whole universe to split!

In the MWI, decoherence is what splits the worlds and this is neither restricted to measurements nor is it special in the sense that it is introduced by hand like the measurements in Copenhagen. It can be derived from the Schrödinger equation.

What's special in the MWI is the interpretation of the decohered state and not the kind of processes which occur. MWI is the most literal interpretation of QM. It is what you get if you apply QM to the measurement process itself. However, not everyone agrees that it is a valid interpretation.

 

[phinds]

Both seem very implausible to me.

Personally, I think "implausible" is an extremely generous description of the various many-worlds / multiverse concepts. I tend to use words that are less acceptable in polite company.

 

[kith]

I think the basic idea is pretty compelling but it is not clear to me how it explains the probabilities we observe in experiments. Or if it is even meaningful to request such an explantion within the context of the MWI.

 

[The_Duck]

I don't quite understand some central points of the many-world interpretation: When does the splitting happen, and is there superposition in a single universe?

In many-worlds, the central object is the wave function of the universe. The wave function evolves with time according to the Schrodinger equation. The wave function tends to contain localized lumps of probability. We can call these localized lumps "worlds." Each lump corresponds to a possible configuration of the universe, and each lump evolves independently.

It can happen that a localized lump of probability amplitude within the wave function splits in two. Then we can say that one of the worlds has split into two worlds. For example, one lump might describe a configuration of the universe in which a measuring device readout displays "electron with spin up," while the other lump might describe a configuration of the universe in which a measuring device readout displays "electron with spin down."

The point is that measurement is not special: the wave function of the universe simply evolves according to the Schrodinger equation at all times. Sometimes a lump of probability amplitude will split into two; if we like we can call this a "measurement."

 

[meBigGuy]

I'm trying to understand what MWI is is really about. The sentence "MWI is the most literal interpretation of QM. " leaves me completely blank. I don't get the need for lumpy splitting to explain decoherence and the apparently classical world that results. What am I missing?

 

[kith]

The starting point of Everett was to wonder what happens if an observer A, who performs a measurement on a physical system, is observed himself by an external observer B. After collapse, A has a different wavefunction than B. This implies either that wavefunctions have no physical reality or that QM can not be applied to big systems like observers.

The MWI is the most literal interpretation of QM in the sense that it doesn't introduce additional elements (like dBB) but still disagrees with both statements. The wavefunction is the real thing (and not just knowledge about more fundamental unknown properties), and QM applies to everything.

Also note that the MWI doesn't explain decoherence. Decoherence is a measureable feature of QM which is present regardless of the interpretation.

 

[bhobba]

I don't get the need for lumpy splitting to explain decoherence and the apparently classical world that results. What am I missing?

You have it about face. MWI doesn't explain decoherence - its part of the formalism of QM regardless of interpretation - it makes use of it.

Thanks
Bill

 

[kaplan]

To me, the MWI is obviously the best interpretation of QM, because it is the simplest. It is nothing more or less than the idea that everything is described by QM and the Schrodinger equation, including measuring devices. Given the (remarkable) absence of any evidence to the contrary, I see no reason to doubt that, and I would say anyone that does has the burden of proof squarely on their shoulders.

As for when the universe splits - first, this is really a continuous process, so there's no perfectly well-defined "when". But more or less it happens whenever macroscopic objects (such as measuring devices or people) interact with anything. In other words, pretty much all the time.

It's also worth bearing in mind that given what we know about physics, all interactions are local, meaning that the split affects only the lightcone of an event.

 

[.Scott]

I'm trying to understand what MWI is is really about. The sentence "MWI is the most literal interpretation of QM. " leaves me completely blank. I don't get the need for lumpy splitting to explain decoherence and the apparently classical world that results. What am I missing?

The whole point of MWI is to explain the apparent random results you get from decoherence. Instead of looking for a cause for QM decoherence results, MWI allows us to say that there is no answer. From my way of looking at it, it's bad scientific attitude.

What is normally done in science is to presume that there is a modelable explanation for a measured result, hypothesize, and then test. So if you presume MWI, you have stopped doing science because you are resting on an untestable "theory".

A better attitude is to presume that decoherence isn't in principle random - presumably that there are discoverable non-local causes that can be modeled. This is not to say that such a model could be used to actually predict the outcome of decoherence event, but that model could be used to make other testable predictions.

 

[kaplan]

The whole point of MWI is to explain the apparent random results you get from decoherence. Instead of looking for a cause for QM decoherence results, MWI allows us to say that there is no answer. From my way of looking at it, it's bad scientific attitude.

I don't recognize what you're saying as the MWI, at least not its modern incarnation. The modern view is that MW is simply a consequence of Schrodinger evolution. The Schrodinger evolution of systems with many particles is decoherent, full stop. The "cause" is, well, Schrodinger's equation.

What is normally done in science is to presume that there is a modelable explanation for a measured result, hypothesize, and then test. So if you presume MWI, you have stopped doing science because you are resting on an untestable "theory".

Not so, as above.

A better attitude is to presume that decoherence isn't in principle random - presumably that there are discoverable non-local causes that can be modeled.

That's precisely what the MWI assumes. Differential equations (like the Schrodinger equation) are deterministic. So decoherence in MW is certainly not random.

 

[kith]

Today, we have all kinds of explanations for the randomness of QM: hidden variables, many worlds, retrocausality, etc. However, all of them have in common that the entities which are supposed to explain the radnomness can't be measured. My guess is that this will be true also for future theories and interpretations.

 

[.Scott]

It's also worth bearing in mind that given what we know about physics, all interactions are local, meaning that the split affects only the lightcone of an event.

I hope that I am allowed to agree and disagree with you on the same point at the same time.
To a certain extent, the light cone statement is true. But for this discussion, I would flip it around: Everything contributing to the outcome of a decoherence event will be contained within the historic light cone of that event.

However... you still need a form of non-locality to make it happen because the light cone isn't really a cone. It doesn't come to a point. There is no instant in time or point in space where the decoherence takes place. Instead, the transition occurs over a region of time and space.

 

[.Scott]

Today, we have all kinds of explanations for the randomness of QM: hidden variables, many worlds, retrocausality, etc. However, all of them have in common that the entities which are supposed to explain the radnomness can't be measured. My guess is that this will be true also for future theories and interpretations.

If you can't, in principle, test a theory, then it isn't a theory.

 

[.Scott]

That's precisely what the MWI assumes. Differential equations (like the Schrodinger equation) are deterministic. So decoherence in MW is certainly not random.

There is a big difference between the implications of differential equations and MWI. With the differential equations, you're allowing all of the "many worlds" to interact. With MWI, at decoherence they stop interacting with each other completely.

Put it another way, if Scroedinger's Cat was really subject to macroscopic QM effects, it wouldn't just come out dead or alive, it would escape with the experience of having interacted with it dead self. That's what the differential equations suggest.

 

[kith]

If you can't, in principle, test a theory, then it isn't a theory.

Yes, that's why they are called interpretations. But my guess is that also future theories won't be able to introduce observable elements which get rid of quantum randomness.

 

[.Scott]

Yes, that's why they are called interpretations. But my guess is that also future theories won't be able to introduce observable elements which get rid of quantum randomness.

I am also confident that they "won't be able to introduce observable elements which get rid of quantum randomness" - at least not in the sense that they will make it any more predictable. To do so would be a direct violation of Heisenberg Uncertainty. But they'll be able to demonstrate other side effects that are consistent with there being no net synthesis of information - that is, that the results, though random, can be modeled as fully determined by their historic light cone and are independent of anything like a many-world selection parameter.

Ahh, that's true. In the many-world theory, there would be a continuous influx of information into the universe - and as a result, there would be a continuous minimum physical size to the universe.

That would be, at least in principle, testable.

It would also mean that if you run time backwards, there would be a destruction of information. I'm not sure that's good for the MW theory.

 

[kith]

But they'll be able to demonstrate other side effects that are consistent with there being no net synthesis of information - that is, that the results, though random, can be modeled as fully determined by their historic light cone and are independent of anything like a many-world selection parameter.

I am very sceptical about this as well because I don't think that future theories will get rid of quantum non-seperability.

In the many-world theory, there would be a continuous influx of information into the universe

Why do you think so? The entropy of the universe doesn't change under unitary time evolution.

 

[The_Duck]

There is a big difference between the implications of differential equations and MWI. With the differential equations, you're allowing all of the "many worlds" to interact. With MWI, at decoherence they stop interacting with each other completely.

What? You seem to think that in MWI something special happens when "worlds" split. Not so. The splitting of worlds, and their evolution before and after the splitting, is described by the Schrodinger equation.

Put it another way, if Scroedinger's Cat was really subject to macroscopic QM effects, it wouldn't just come out dead or alive, it would escape with the experience of having interacted with it dead self. That's what the differential equations suggest.

No! Stop! You don't understand MWI as well as you think you do. Schrodinger's equation is linear, which means that the two elements of a superposition evolve entirely independently of one another, without affecting each other. If you were placed into a superposition of two states, neither version of you would be able to detect or interact with the other version, precisely because of this linearity.

they'll be able to demonstrate other side effects that are consistent with there being no net synthesis of information - that is, that the results, though random, can be modeled as fully determined by their historic light cone and are independent of anything like a many-world selection parameter.

Pretty sure you've just postulated local hidden variables, which are ruled out by Bell's theorem. If you want the results of measurements to be fully determined by their past light cone, this is the same as having some hidden variable at each point in space with the hidden variables evolving in time according to local interactions. Any such model is ruled out by the observed violations of Bell's inequality.

In the many-world theory, there would be a continuous influx of information into the universe - and as a result, there would be a continuous minimum physical size to the universe.

What? The fact that worlds are constantly splitting does not imply a minimum size for the universe.

It would also mean that if you run time backwards, there would be a destruction of information. I'm not sure that's good for the MW theory.

The fact that worlds merge instead of splitting when you run time backwards is not a bad thing. It's a reflection of the second law of thermodynamics.

 

[jfizzix]

The many worlds interpretation makes lots of sense if we do away with any notion that observers are somehow special, or that we are anything more than the total of all the atoms and their interactions making us up

That we observers happen to see a random chain of events is simply due to the fact that our memories are material accounts written in patterns of connected neurons, but written with materials made out of atoms nonetheless.

As far as probabilities go, keep in mind we never determine a probability experimentally with a single measurement. We have to measure many many times and look at the ensemble as a whole. Also for a large number of measurements, these random outcomes overwhelmingly conform to specific probability distributions.

As for what determines those distributions in the first place, I haven't figured that out yet.

 

[bhobba]

As for what determines those distributions in the first place, I haven't figured that out yet.

Then you might be interested in Gleason's Theorem:
http://kof.physto.se/theses/helena-master.pdf

Thanks
Bill

 

[craigi]

One the one hand, the Many Worlds Interpretation is just a way to visualise QM rather than a theory, so the criticism that it makes no testable predictions seems to miss the point.

On the other hand, as a member of a class of multiverse theories. There is, in principal at least, some way to distinguish them.

If it is demonstrated that our universe has evolved in such a way that didn't require particularly fortunate, chance events then we would have no evidence to support a multiverse theory.

This sort of analysis seems a long way off and it's not clear that it could distinguish between different multiverse arrangements, but we do have access to information in order to perform a test.

 

[kaplan]

There is a big difference between the implications of differential equations and MWI. With the differential equations, you're allowing all of the "many worlds" to interact. With MWI, at decoherence they stop interacting with each other completely.

No, that's not the case. They do not "stop interacting with each other completely". The wavefunction evolves according to the Schrodinger equation, and it's not possible under Schrodinger evolution for two branches to form and then entirely cease to interfere.

What does happen is that the wavefunction becomes very, very close to a diagonal density matrix in the appropriate basis, and interference between the branches is very small.

Put it another way, if Scroedinger's Cat was really subject to macroscopic QM effects, it wouldn't just come out dead or alive, it would escape with the experience of having interacted with it dead self. That's what the differential equations suggest.

No, it's not. Again - decoherence is a feature of Schrodinger evolution. It's not some extra assumption.

 

[.Scott]

Let me address the issue of information and minimum universe size.
If you assert the many-world interpretation, then in order to predict the outcome of certain QM events, you not only need to have information from the historical light cone of the event, but you also must know which universe you end up in. (I realize I just use grammatical tense in an odd way - but you know what I mean.)

So if an event splits into 200 possibilities, you get 200 possible worlds each "tagged" with a different result, one of 200 possible results. That means that that universe is now 7+ bits bigger than it was before the split, the amount of information it would take to make that universe different from all of its 199 siblings.

So how does this affect the minimum size of the universe? It increases the Berkenstein bound.

Unless there was a way of remerging universes, the result would be a universe that was continuously growing.

Here's another way of looking at it. The many worlds view hasn't been worried about creating an arbitrarily large number of universes. However, in order for there to be an arbitrarily large number of universes, there needs to be something different about each of those universes. That means that each universe must hold at least as many bits as it would take to count all of its siblings, cousins, second cousins, etc. Since that is the binary logarithm of a number that is allowed to grow arbitrarily large, it is a number of bits that is also allowed to grow arbitrarily large. And since the number of bits in our universe will grow arbitrarily large, our universe must also, over time, grow arbitrarily large.

 

[bhobba]

That means that that universe is now 7+ bits bigger than it was before the split, the amount of information it would take to make that universe different from all of its 199 siblings.

Run that by me again. Cant follow that at all.

Pour a glass of water into two separate containers. No information lost or gained but the water in each container can evolve in two totally different and independent ways.

The many worlds view hasn't been worried about creating an arbitrarily large number of universes.

There is only one universe. When the universal wave-function encounters decoherence it becomes an improper mixed state and each 'element' of that mixed state is considered a separate world - nothing is created or destroyed - we still only have one universe, but it is simply partitioned differently as it evolves.

Thanks
Bill

 

[.Scott]

No! Stop! You don't understand MWI as well as you think you do. Schrodinger's equation is linear, which means that the two elements of a superposition evolve entirely independently of one another, without affecting each other. If you were placed into a superposition of two states, neither version of you would be able to detect or interact with the other version, precisely because of this linearity.

If you're going to look at it in that detail, then bear in mind that there are not just two results by even the simplest count. If particle decay occurs, it can occur over any "moment" over a span of time. So our cat may be alive, just dead, dead, long dead, etc.

Next: There are a couple of ways of using the Schrodinger's cat story. I always took it as a way of illustrating "measurement" as if the cat was subject to QM collapse - the notion that the cat's destiny was not determined until the box was opened. Since I never expected anyone would seriously think that the cat was really ever both dead and alive, I extended the cats predicament to what a particle would do when it crosses through both the left and right slit. Since the left and right particle "interact", so would the dead and alive cat.

 

[.Scott]

There is only one universe. When the universal wave-function encounters decoherence it becomes an improper mixed state and each 'element' of that mixed state is considered a separate world - nothing is created or destroyed - we still only have one universe, but it is simply partitioned differently as it evolves.

Thanks
Bill

Okay. Then instead of an arbitrarily large number of universes, you have an arbitrarily large number of partitions. The arithmetic still holds.
Unless you allow those "partitions" to eventually remerge, you will have partitions that continue to grow.

 

[bhobba]

Okay. Then instead of an arbitrarily large number of universes, you have an arbitrarily large number of partitions. The arithmetic still holds. Unless you allow those "partitions" to eventually remerge, you will have partitions that continue to grow.

So your partitions continue to grow? What of it?

Thanks
Bill

 

[bhobba]

There are a couple of ways of using the Schrodinger's cat story.

More than a couple - pretty much each interpretation has its own view. The MWI take is we experience one world after decoherence creates one with a live cat, and another with a dead cat - opening the box changes nothing - zero zilch.

Thanks
Bill

 

[.Scott]

So your partitions continue to grow? What of it?

Thanks
Bill

It makes it a testable theory. It's not an interpretation.
How many QM collapses occur in a cup of water every millisecond? You could calculate the rate at which information is being pumped into the water. I suspect it is way too much to go unnoticed. That means you will need a mechanism for "remerging" to destroy the information. That puts MWI in conflict with conservation of information - a theory that's been looking mighty good recently.

 

[bhobba]

It makes it a testable theory. It's not an interpretation.

Sorry - it doesn't - that's why its an interpretation.

If you think there is a way to tell the difference, write it up, get it published in a peer-reviewed journal and a certain Nobel prize is in store.

I suspect you don't understand the modern view of the MWI and its relation to decoherence:
http://philsci-archive.pitt.edu/5439/1/Decoherence_Essay_arXiv_version.pdf

I suggest becoming familiar with it.

Thanks
Bill

 

[DrChinese]

To me, the MWI is obviously the best interpretation of QM, because it is the simplest. It is nothing more or less than the idea that everything is described by QM and the Schrodinger equation, including measuring devices. Given the (remarkable) absence of any evidence to the contrary, I see no reason to doubt that, and I would say anyone that does has the burden of proof squarely on their shoulders.

As for when the universe splits - first, this is really a continuous process, so there's no perfectly well-defined "when". But more or less it happens whenever macroscopic objects (such as measuring devices or people) interact with anything. In other words, pretty much all the time.

It's also worth bearing in mind that given what we know about physics, all interactions are local, meaning that the split affects only the lightcone of an event.

One minor detail that is not covered by most discussions of MWI: that light cone MUST include the future as well as the past. If you don't consider that, the results of Bell tests on entanglement-swapped pairs cannot be explained. Since the swapping can occur after the detection, and also alternately the detection correlations can occur far outside the light cone to the past.

So my point is that if MWI is really such a simple explanation, it must also match any and all QM setups simply too. And that requires acknowledgment of a future context.

 

[jambaugh]

To me, the MWI is obviously the best interpretation of QM, because it is the simplest.

Don't equate "simple" with "easy" some may find MWI easier to rap ones head around but that doesn't make it simpler... you are, in the theory, creating a continuum of multiverses, and one has not even yet given the mechanism for the causing of their interactions and such,... one only gives an ad hoc description of how they behave... which is what one is trying to avoid in choosing it over the (truly simpler IMNSHO) Copenhagen Interpretation.

CI is simpler as it rejects the necessity and ability to provide an ontological role to the wave function. It is not as easy to understand though due to our history of though and classically trained intuition. I would ask you to consider the same argument as you make here, but instead supporting the aether theory of light (empirically equivalent to SR but with the Lorentz's object contraction and clock slowing time dilation taken as literally now as they were originally proposed.) It is easier to get one's head around a theory where the aetheric wind messes up our clocks and measuring rods just enough to hide its existence from our ability to observe. But it is a complicated attempt to reconcile our wrong intuitions with the right evidence. CI simplifies in the same way as SR does... simply trimming via Occam's razor that unnecessary component... in CI's case the "reality" of the "state vector" as in SR's case it was the absoluteness of time.

 

[.Scott]

I suspect you don't understand the modern view of the MWI and its relation to decoherence:
http://philsci-archive.pitt.edu/5439/1/Decoherence_Essay_arXiv_version.pdf

I suggest becoming familiar with it.

Thanks
Bill

I just read the article you cited, scanning Chapter 1, quickly reading most of sections 2 through 4. Of course I paid particular attention to certain sections, including section 3.2.3.
An essential part of this concept has not changed since I first read about it quite some time ago. The essential part is that "branches" or "worlds" or "partitions" or whatever you wish to call them are continuously generated - each eventually affecting the macroscopic world and each continuing on without ever remerging with the others.

Sorry - it doesn't - that's why its an interpretation.

If you think there is a way to tell the difference, write it up, get it published in a peer-reviewed journal and a certain Nobel prize is in store.

Your article describes it as an interpretation - and it does not include the expanding information issue. I do believe the expanding information issue is what puts this into the category of theory.

As for getting it published, I have little connection to the science community. I am a Software Engineer who hasn't any notion of what process to follow to become published. And, of course, it would be a lengthy process for me to form the paper in terms acceptable for publication.

Nevertheless, there is clearly a potential issue with information density during superpositioning. If superpositioning is allowed to compound without limit, as with the MWI, the capacity of local space to hold anyone of the "worlds" will be exceeded.

 

[bhobba]

Your article describes it as an interpretation - and it does not include the expanding information issue.

That's because information is not expanding. That's simply a misconception on your part, like you believe you can tell the difference between MWI and other interpretations. There have been proposals put forward on how to do it, but (of the ones I am aware of) the technology required to do it is of dubious practical, or possibly even theoretical, realization. Anyone actually coming up with a way to do it that was realizable, and actually got it carried out, and the results proved, or disproved it, would get a Nobel prize for sure - it would be something of fundamental importance.

For what its worth I believe its inherently impossible to tell the difference. As Murray Gell-Mann replied when asked the difference between Decoherent Histories, which he champions, and MW, he said - basically none - MW is simply the interpretation where an observation doesn't single out one history - but all occur simultaneously. Since the worlds are assumed to not communicate there is no way to tell the difference - the proposals put forward to do it are flawed - like there were proposals put forward to tell DBB from other interpretations - careful analysis showed they were flawed - and in hindsight it should have been obvious - DBB was cooked up from the start to be indistinguishable.

I do believe the expanding information issue is what puts this into the category of theory.

Exactly why you believe that is beyond me.

Nevertheless, there is clearly a potential issue with information density during superpositioning. If superpositioning is allowed to compound without limit, as with the MWI, the capacity of local space to hold anyone of the "worlds" will be exceeded.

It may be clear to you, but it far from clear to me. 'Superpositioning' - have zero idea what you mean by that. The number of worlds compounds without limit. So?

Thanks
Bill