Saturday, September 27, 2014

Hartle, Srednicki on foundations of QM

In August, there was a workshop on the foundations of quantum mechanics (see 12 videos) where lots of nonsense has been said but there were smart and reasonable people, too. Santa Barbara is a nice and sunny place in California but it's also a source of almost reliably sensible comments on quantum mechanics.

Because we discussed Murray Gell-Mann's comments about quantum mechanics a few days ago, I think it's natural to start with Gell-Mann's co-father of the Consistent Histories, Jim Hartle. He gave a talk about the emergence of classical physics.

I could obviously describe the talk in the usual detailed TRF way but let me avoid it this time. The main points is that the default behavior of physical systems is the quantum behavior, classical physics is just a limit, and whether this limit is relevant must be judged by the approximate validity of the classical dynamical laws as extracted from the probabilities of some coarse-grained histories. The degree of "classicality" may be quantified and Hartle was discussing some relationships between classicality and complexity.

There are lots of somewhat new ideas and tons of nice pedagogical memes to present ideas that are not so new and Jim has organized it nicely. You may notice one very general point that I consider a very good one: it is really a talk about the interpretation of classical physics. Because the quantum mechanical laws are the right ones, they don't really need special interpretations; they define the primary rules and postulates one simply has to learn at the beginning. If something needs extra interpretations and "ifs" and "buts", it's some approximate laws that aren't fundamentally true, e.g. the laws of classical physics. Most people unsuccessfully trying to study the foundations of quantum mechanics err in this very basic point – the large number of people who misunderstand the point is the reason why the deeply problematic phrase "interpretation of quantum mechanics" remains so popular.

One may hear lots of questions during Jim's talk that aren't really terribly intelligent. Most of those who ask them weren't loud enough. But Sean Carroll started to speak at 46:00 and he's always louder than appropriate. He spoke about some e-mail from Seth Lloyd that Hartle hadn't read. Carroll interpreted the e-mail as an opinion that the physical system consisting purely of one simple harmonic oscillator in the ground state actually has lots of coarse-grained histories in it, one may ask lots of questions, there is actually bath of harmonic oscillators within the single one, and so on.

Jim has pretty much only stated that Sean Carroll's question was an unintelligable one and they should have discussed it later instead of wasting time with this stuff during Hartle's seminar. Yes, for another minute, Hartle was failing in his über-diplomatic efforts to make Carroll shut up (Carroll would also use the monologue to promote one of his crackpot papers). I am sure that Jim thinks the same things about these trivial issues as I do and it might be nice if he analyzed all these misconceptions and when he is asked after another talk, he could coherently and concisely summarize the basic mistakes that the likes of Carroll are making all the time.

If there is just one harmonic oscillator, there is just one harmonic oscillator, and there is no way how you could describe it as an "ensemble" or a "bath" with many oscillators. A single oscillator is simply not equivalent to many. People confusing themselves with fairy-tales about the "many worlds interpretation" may get totally messed up when it comes to distinguishing "one" from "a number larger than one" but if one analyzes things rationally, it's very important that "one" is not the same thing as "many" in general. The algebras of observables for "one" or "many" oscillators are not isomorphic, and so on, and that's the criterion meaning that the systems are physically different.

Also, if we only deal with a simple system like a single harmonic oscillator, there is clearly no decoherence at all. Because Gell-Mann's and Hartle's consistent histories are really "decoherent histories", there is no room for them here at all! Decoherence only occurs if there are many degrees of freedom and if some "irreversible loss of detailed information" becomes inevitable. It seemed pretty self-evident that Carroll does believe that there is decoherence even in a single harmonic oscillator that is not coupled to anything else at all. He is just totally confused about the very basics of quantum mechanics. Also, I am generally alarmed by his way of arguing. When he says something completely wrong or meaningless, he tries to make it sound important by references to e-mails from would-be authorities – while the e-mails are probably heavily misinterpreted and they were probably nonsensical to start with even without any misinterpretation. A proper scientist just ignores things that make no sense to him – that can't pass basic checks. People like Carroll amplify all the rubbish they find somewhere.

Incidentally, if one talks about just one harmonic oscillator, but not necessarily assuming that it was found in the ground state, there are various coarse-grained histories one may define. One may measure the positions and velocities at various moment, with some limited variable accuracy, and so on. But there will still be no decoherence. Decoherence is an irreversible process so it simply can't take place in integrable systems, especially as simple ones as the harmonic oscillator.

In the next question, Scott Aaronson argued that decoherence and complexity seem independent – one can have non-complex systems that decohere (large piece of a material) and complex ones that don't (quantum computer). Hartle's answer made it clear that Aaronson didn't get it, either. The relationship was supposed to be upside down: classicality is indeed correlated with the the simplicity (of the laws describing the chosen coarse-grained observables), and not the complexity. Large systems are still complex if you want to analyze them precisely but that's exactly what you're never doing if you analyze a system through its classical limit only.

I find it bizarre that Scott could have gotten this basic point upside down, especially because he was giving a talk about almost the same question and the talk wasn't "quite" wrong – the problem was mostly that it was filled with tons of unphysical flapdoodle about consciousness etc. But Scott, classicality is surely a property or emerging description of a system whose laws simplify rather than become more complex!

Yes, there is some sense in which this important point is the exact opposite of what Aaronson tried to describe in his talk. He would identify classicality – or consciousness – with "complexity" in the sense of having many degrees of freedom that are chaotically enough coupled. But what even semi-intelligent people such as Aaronson fail to understand, the term "classicality" doesn't refer to the intrinsic character of a physical system in the most accurate possible description. Instead, "classicality" is a property of a particular description itself. Intrinsically, all systems are quantum mechanical if one wants to be as accurate as possible – or allowing all kinds of physical questions that are legitimate. Systems with many degrees of freedom have the potential to yield laws for some limits that simplify (e.g. the classical laws) but the potential doesn't mean that these systems are intrinsically classical as a whole. They are not. No systems in Nature, if considered in their full depth and entirety, are "intrinsically" classical. Scott's misconception may be interpreted by saying that he suffers from some "misconceptions of the realists" because he wants to label "classicality" as an objectively real property of a physical system. But it can't be like that. "Classicality" is always a property of a description only (it may emerge when the appropriate relevant observables and their accuracy or coarse-graining are specified).

Another question was about the anthropic reasoning – whether it's a truism or a testable claim. He probably repeated some trivial things that were incorporated in Jim's talk. Jim also correctly stated that physics, when it's done right, must be done in the same way as news reporting on Fox News. ;-)

Mark Srednicki (who has incidentally often collaborated with Hartle over there in Santa Barbara, of course) gave a talk about the subjective Bayesianism and quantum mechanics. He would start describing Bayesian probabilities and axioms about probabilities in general and continued with claims that "probabilities of probabilities" are OK and quantum mechanics should be phrased in that way. Superficially, everything should have been OK but I would have serious problems with tons of "details" that Srednicki would say in between.

No comments:

Post a Comment