User talk:Byrgenwulf/RQM

Procedure
If I see an ambiguity I'll edit the article directly to correct or leave a question. --Michael C. Price talk 20:43, 28 July 2006 (UTC)


 * Excellent...I see you caught one already; it needs a thorough scouring for those sorts of things. As you have no doubt picked up, I am trying to keep the first few sections relatively accessible to the mathematical layperson, and moving on to the more technical bits later.
 * One thing I have been wondering about, and perhaps you have some ideas, is just how far to take the formal derivation of the theory. I have provided a very rough summary here, sticking to Rovelli's paper but cutting out as much as possible in the interests of conciseness; should the extra bits come in, for the sake of rigour and completeness?
 * Also, to what extent do you think the cosmological applications be included? The material is scant enough that Smolin's thoughts don't merit a separate article, but at the same time they may be a bit off topic in a discussion of the interpretation of quantum mechanics. Byrgenwulf 20:51, 28 July 2006 (UTC)

OK, some general impressions. It's not obvious, at this stage in the development of the exposition, how RQM differs from Everett's RSI (which he originally called the "Correlation Interpretation", did you know?). That may be intentional (I suspect it is) but it also means that I am having problems seeing what advantages it has over, or differences from, the RSI/MWI.

As to how far to take the theory -- no easy answers there (I have the same problem with the MWI!), but I would include the cosmological stuff (heck, why not?) if it can be made concise and coherent. --Michael C. Price talk 21:28, 28 July 2006 (UTC)


 * I understand what you mean about the similarity to MWI...and the two are very similar in many respects. The difference, though, is this.  Everett gave us the structure and behaviour of a universal wavefunction and the MWI ascribes the attribute of "reality" to this wavefunction.  In RQM, on the other hand, there can be no well-defined meaning to a "universal wavefunction", because a wavefunction is an expression of the correlation between two objects: if the universe is one such object, we would need another object to give a universal wavefunction meaning; what would this other object be? God? But then God is material and part of the universe, and so on.
 * One could say that while MWI creates multiple "realities", each corresponding to an entire universe and a new "branching off" from the wavefunction, RQM shatters the idea of one giant "reality" across the whole universe into a myriad tiny ones: objects only have well-defined properties in the light cones of other objects; the ultimate in "local realism" one could say: everything local is real, but spacelike separated events are not defined relative to one another. I visualise it in my head as being like trying to tile a sphere with flat circular tiles...a problem in cohomology, really.
 * As for advantages, I suppose it comes down to what one regards as being desirable in a physical theory (as with any interpretation!)...to me, RQM makes sense because it is more ontologically austere than MWI, while simultaneously not suffering the problems of the Copenhagan Interpretation, not having hidden variables like Bohm's, and filling out the gaps in the Consistent Histories view (with which it accords perfectly well). The formal derivation of the theory can also be done from the most elementary of postulates, because the assumption of the observer-dependence of state yields enough structure (linear subspaces in a complex Hilbert space) to recreate the formalism.
 * I hope that answers your query...I still have to write more on the implications/interpretation (all those empty headings); and in the "relationships to other interpretations" I shall obviously discuss things like this: how it is similar and different to all the "big" ones. And I shall include the cosmology bits, because I used them above to differentiate RQM from MWI, and hence would be of benefit to the reader, I think.
 * It's good you're asking questions like this, because obviously the article will only work if it can take someone who is familiar with quantum mechanics, but not this particular view of it, to the point where they fully understand it. Byrgenwulf 07:10, 29 July 2006 (UTC)

RQM seems to take the relationships between objects as the base existential reality without assigning reality to the objects as well. I confess this seems like a contradiction to me -- objects are empirically defined by their relationships and so must inhereit any associated reality -- no doubt it will seem contradictory to others as well and your exposition will need to address this. I don't see any mechanism that denies reality to the other Everett worlds.

The term "post-Everett" is sometimes applied to DeWitt's MWI, Gell-Mann/Hartle's consistent histories, Zurek's existential interpretation and whatever decoherent approaches I've misssed. I guess you see RQM as a post-Everett interpretation?

More specific comments, as I'm sure you've noticed, are left scattered through the text with a "-MCP" tag. --Michael C. Price talk 07:48, 29 July 2006 (UTC)


 * I addressed a couple of the MCPs in the text; do put them back if there is still an outstanding ambiguity.
 * I know what you mean about the "non-existence of objects"...I see it as a form of nominalism; although, the philosophy behind the interpretation is actually far more complex than the physics, hence the weight given to it in the article - I may end up needing to fix some of the philosophy articles as well, so that the whole "net" hangs together. But let me put it like this, for the moment.  There is no a priori reason that we must expect objects to have intrinsic properties.  In RQM, as in nominalism and some other philosophies (e.g. Nietzsche's), objects have no intrinsic properties whatsoever: all properties are relational...and that's really what RQM is saying. Weird, yes, but impossible, no.
 * There's no mechanism to deny reality to the other Everett worlds, but nor is there a reason for asserting them, unless one assigns reality to all possible relations between objects, in which case there would be many spheres with circular tiles - but one doesn't have to make this assumption to keep the interpretation consistent, because "wave collapse" events happen simply when two objects interact and a particular instantiation of a relationship is stochastically chosen; because RQM uses the Heisenberg picture, Schroedinger evolution doesn't describe anything real, it is simply another mathematical way of getting the same results. Byrgenwulf 08:36, 29 July 2006 (UTC)

I have no problems with "all properties being relational" because whether the properties are associated either with objects or with relationships is a "meaningless" distinction (= "metaphysical" in A. J. Ayerian terminology); the two viewpoints are empirically indistinguishable.

I have a more serious problem with the statement "wave collapse" events happen simply when two objects interact and a particular instantiation of a relationship is stochastically chosen;. In the widely accepted decoherence view collapse "occurs" only with thermodynamically irreversible interactions, not with any interaction in general. Any interpretation that claims otherwise is making falsifiable (nay, falsified) predictions.

--Michael C. Price talk 10:27, 29 July 2006 (UTC)
 * Hence the scare quotes around "wave collapse", because it is not wave collapse in the Copenhagen sense...and the decoherence view has evolved into the consistent histories approach, with which the interpretation accords perfectly. Let's put it like this.  In the Copenhagen view, interaction with a macroscopic object causes collapse.  In the consistent histories/decoherence view, a thermodynamically irreversible interaction causes collapse.  In RQM, any interaction causes a wavefunction to take a particular state.  This has not been falsified, and will not be falsified, because we cannot observe a superposition (because of decoherence!). Our description of states might be in a superposition (so a system is in a superposition relative to us) but once the system interacts with us, it will no longer be in a superposition.  And, assuming the system does not interact with any other system, it will continue to evolve according to normal laws.

If it does interact with another system, we have two possibilities. Either we have a description of that system as well, or we don't. If we do have such a description, we can then describe the states of both systems after the interaction, according to normal laws. If we don't have such a description, then our initial information becomes meaningless anyway, and whatever measurements we make will be a surprise. How does that violate decoherence empirically? Byrgenwulf 11:08, 29 July 2006 (UTC)


 * My scare quotes around "occurs" were an acknowlegdement of yours; we seem to be somewhat taking past each other somehow. I am confused by your frequent use of "description" which implies the wavefunction is just a classical repository of our knowledge of the world -- such an understanding is incompatible with the occurance of Interference effects; the disappearance of interference cross-terms can be understood (as opposed to axiomatically conjectured) via the operation of Quantum_decoherence.  I can't see how this is handled in RQM  --Michael C. Price talk 11:55, 29 July 2006 (UTC)

Ha, I see where the problem comes in. I have been using "description" very loosely, in such a way that it could imply some "anthropic" influence: knowledge we (humans!) have. No, "x's description of y" is exactly the same as "x's specification of y's wavefunction" is exactly the same as "the information x has on y". An electron x can have a description of another electron y, in this sense, if they have interacted and x has information on y: x's state is correlated in some sense with y's state. So, decoherence still works exactly as it does, really...if the system y (on which x has information, or x can describe, or for which x knows the wavefunction etc.) has not interacted with the environment, then the Born rule applies in a straightforward sense. If the system y has interacted with anything (be it "environment" or another particle or anything), then the unitary evolution of y has broken down, and x's information on y is no longer "current". Which means that if x has some information on the system with which y has interacted (the environment), then x must take account of that information when determining the new state of y: which is done using the normal method: i.e. summing over the states that that information might have and applying the decoherence condition. Of all interpretations, RQM is perhaps closest to the consistent histories interpretation...which, after all, was born of decoherence. Maybe one could call it a post-Everett interpretation (but I hadn't heard that term until you used it earlier). Byrgenwulf 12:33, 29 July 2006 (UTC)


 * "Post Everett" dates back to Gell-Mann and Hartle in the early 1990s, I think. I know Zurek and Baez use the term often. Googling on "post Everett" and "quantum" pulls up some interesting material.


 * "x's specification of y's wavefunction" seems to be the same as "x's relative state with respect to y". It might be less confusing to use the relative state terminology to avoid anthropic confusions.


 * Assuming the relative state/decoherent issues are handled coherently within RQM that still leaves us with the central philosophical problem I mentioned earlier. Drawing a distinction between the existence of objects vs relationships seems a purely non-empirical, metaphysical, and hence meaningless, dicotomy.  --Michael C. Price talk 13:10, 29 July 2006 (UTC)

Well, to avoid confusion with Everett's work, Alice's state relative to Bob (like Alice's velocity relative to Bob), since the inspiration is from relativity.

As for the "philosophical objection", that holds for absolutely any interpretation of quantum mechanics: ultimately, empirically, the theory works no matter what interpretation we put on it. We know this by now. However, it often sits uncomfortably with common-sense notions about how the world must work, and it seems that we must ultimately make some "philosophical sacrifice" to reconcile the two. Copenhagen it is consistency of world-view; MWI it is ontological paucity; Bohm it is accepted norms in philosophy of science; RQM it is observer-independent state of a system (bearing in mind "observer" here is used like it is in relativity).

RQM does account for "quantum non-locality" very well: there is no quantum non-locality. So that, I think, is a "plus". But of course, an encyclopaedia article's purpose isn't to convince the reader that a given point of view is the one they should adopt; it's merely to lay it out. I shall, however, be discussing the philosophical aspects of the theory in detail. Byrgenwulf 13:32, 29 July 2006 (UTC)


 * I couldn't've expressed myself clearly with my "philosophical objection". Some interpretations of QM are easy to distinguish from each other: MWI has splitting observers and parallel universes; many-minds is the same as MWI except that the observers' minds are already split; Bohm has "hidden variables" subject to the forces of a quantum potential; Copenhagen has collapsing wavefunctions.  But RQM (and consistent histories) I can't distinguish from MWI.  --Michael C. Price talk 16:14, 29 July 2006 (UTC)

What distinguishes RQM from MWI, then, is that instead of an observer splitting into many worlds, the world is split into many observers, so to speak. This is not like the many-minds interpretation, though, because minds have nothing to do with it. Only local events are real, to any given system. MWI assigns absolute reality to the wavefunction in its entirety. RQM assigns absolute reality only to correlation, information: a participatory universe, in Wheeler's sense; "it from bit". MWI has no wave collapse; in RQM all superposed "states" (correlations) instantly take on an eigenstate upon any interaction, and the eigenstate chosen is entirely random (which isn't wave collapse in the strict Copenhagen sense, because the system itself wasn't ever in a superposition, only a relationship was). There are no parallel universes in RQM -we could still imagine them, I suppose, but there's no physical reason to; the only reason to do that would be as a result of a modal realist philosophical bent, which isn't inconsistent with RQM, but isn't a necessary part of it. Are those differences enough? Byrgenwulf 16:32, 29 July 2006 (UTC)


 * As I've previously indicated I don't buy into the whole "relationships not objects" tack as scientifically or philosophically meaningful, so I think we are condemned to mutual confusion. (Reading Rovelli's preprint it's clear he doesn't understand Everett's approach, raising some sort of "basis problem" strawman, so his whole motivation is wrong - he's trying to fix what ain't broke. I think my efforts would be better directed into improving the MWI article.)  --Michael C. Price talk 17:01, 29 July 2006 (UTC)

Well, you're more than welcome to do whatever you wish, of course. The problem of the preferred basis is not a strawman, but an accepted and recognised problem with the Everett approach. Either we have to accept that any interaction causes branching (in which case I cannot imagine what your problem with any interaction establishing a fixed correlation can be), or we have to establish exactly what does cause branching, which has not been clearly and satisfactorily achieved. It is not unique to Rovelli, either; and his motivation for coming up with the theory was not because of a deficiency in Everett's approach (he never claimed it was). Rather, just as you might find the "relationships not objects" tack unsatisfying, many people find the idea of an infinity of parallel universes similarly unsatisfying...as I said above, it really comes down to one's personal ideas of what does and doesn't constitute an ontologically acceptable physical theory. But I certainly appreciate the ambiguity clearing you did, and this exchange has given me some ideas for what to say about the MWI in the appropriate section! Byrgenwulf 17:13, 29 July 2006 (UTC)


 * You are certainly correct that the preferred basis problem is widely perceived as a problem with the MWI, but I believe this is one of many other misunderstandings of the MWI that need addressing with a clearer presentation of the metatheory. I don't agree with your chracterisation of Rovelli's motivation: his preprint expresses his unhappiness with the interpretational side of QM today, which includes MWI.


 * Saying that I find the "relationships not objects" tack unsatisfying rather understates my position. Suppose someone claims the moon is made of green cheese.  You show them evidence that it is made of grey rocks or whatever.  They respond that the evidence merely demonstrates that the "green cheese" looks as if it is rock, but still affirm that the moon is still really made of green cheese.  Yes, I would find this "unsatisfying", to put it mildly.  Similarly with the claim that objects don't "really" exist, it just "looks" as if they exist....  Yes, it comes down to what we find ontologically plausible.


 * I also found this exchange illuminating. It was fun, and gave me a few ideas a well.  --Michael C. Price talk 17:46, 29 July 2006 (UTC)

Well, I'd certainly like to see how you resolve the preferred basis problem.

What I meant about Rovelli's motivation was that it was not solely concerned with MWI. And a relational (or "ultra-nominalist") ontology can be philosophically well justified, as I say - hopefully I can do that part of the article soon.

Since you've now read the pre-print (although it's technically a post-print), do you think I've captured the key moves in the derivation, just as a side note? Or do I need to add more?

Also, I was thinking of making the "Third Man" problem a separate article at some point, since both RQM and MWI could make use of it (if I recall, it is how Everett opens his thesis). That way, the problem could be presented with a more complete analysis from various approaches. Byrgenwulf 18:00, 29 July 2006 (UTC)


 * The preferred basis "problem" vanishes upon consideration that all empirical results, including our consistent memories, are derivatives of eigen values and transition amplitudes, which are themselves basis independent, in a similar way that all empirical results in special relativity are frame independent. Frame, basis -- same thing conceptually, with same resolution.  Re preprint, I thought you played up the philosophical side more than Rovelli, but that is just a matter of style -- and I quite liked it anyway, even though (or perhaps because) it is where my problems with the interpretation lie.  re "Third Man", yes an article would be nice, like Wigner's friend.  Isn't it really the Second Observer or Third System, though?  Howabout the Observer Observed?--Michael C. Price talk 18:39, 29 July 2006 (UTC)

Observer Observed works well. What did Everett call it? I think what I've written there, "story-wise", is somewhere between Rovelli and Everett, or applied Rovelli's analysis to Everett's story, to be more precise. Or something...I shall have to pull that 1973 book out of my paper compost heap and have a look - I have the terrible habit of citing from memory, which is not always consistent with my environment ;).

Regarding the preferred basis, how do we define "empirical results"? Because that, ultimately, is the problem, there: an empirical result is a measurement. But what counts as a measurement?

As for philosophy, I think I would tend to play it up a little because to me the philosophical aspects of the theory are what make it the most appealing. And, of course, it's what I spend my time thinking and writing about. But the Stanford Encyclopaedia article I referenced in the article goes into the philosophical aspects quite a bit, if I remember correctly. Byrgenwulf 19:21, 29 July 2006 (UTC)


 * Everett didn't call it anything. Somewhere I thought I saw it referred to as "Everett's friend" by way of Wigner's friend, but google doesn't help - probably my memory has merged the two tags.  Here it's called "Everett's Observer".
 * And here "Everett's Friends Paradox" --Michael C. Price talk 20:39, 29 July 2006 (UTC)
 * What is a measurement? - no snappy definition leaps to mind, but I'd be interested to hear of any properties beyond the eigenvalues and the probabilities of observing them. --Michael C. Price talk 20:16, 29 July 2006 (UTC)

I'm going with "Observer Observed" for the RQM article, for now; I also have a good image to use, which I am trying to upload - the problem is there are 'trick' options in the licensing drop-down list, and I keep choosing the wrong ones; I suppose I am going to have to read the GFDL agreement now, so that I can pick that one.

Yes, but whose probabilities of measuring eigenvalues, because we've seen how they can be different ;) Byrgenwulf 06:54, 1 August 2006 (UTC)
 * Yes, since the relative state is relative to the observer. But the point is that given a specified observer its relative state is precisely defined, and hence the probability of the specified observer observing any specified result is definite and basis independent.  Ergo, no basis "problem", just a badly formed question. --Michael C. Price talk 08:50, 1 August 2006 (UTC)

OK, I see your point about that: it works quite well; in essence, it's the same move as RQM (or is RQM the same move as that? :P ). But, given that once we have specified an observer, we have a definite probability, and that a measurement (presumably conducted by an observer) is what causes branching of worlds, I would like to know what an "observer" is...it surely cannot be "anything", if you have a problem with "any" interaction causing the formation of an eigenstate relative to the interacting parties. So it it a human? A macroscopic object? But what's a macroscopic object, anyway? And decoherence seems to suggest that it may be possible to prepare a macroscopic object in a superposition, given sufficient containment (it's already been done for so-called "mesoscopic" gases). Byrgenwulf 13:57, 1 August 2006 (UTC)
 * Actually I retracted my point about "any" interaction when I realised that this was in reference to the formation of relative states, not wavefunction collapse. Everett regarded measurement as any interaction that induced object-subject correlations  -- as I mentioned he originally called his interpretation the "correlation interpretation".  I'd have to check but I think Everett regarded anything as potentially an observer.  More recently I think the trend has been to identify observers with any thermodynamically irreversible process, i.e. observation induces decoherence by definition. --Michael C. Price talk 18:32, 1 August 2006 (UTC)