Reality Conditions

Wednesday, July 25, 2007

Loops 07: Conference report (part 2)

This is the second part of my conference report on Loops 07. The first part was here. Remember that you can download the slides or the audio for most of the talks from the conference webpage.

Wednesday 27/06
Plenary talks

Moshe Rozali started by giving an excellent talk about background independence in string theory, a topic that has been subject of legendary long discussions on Cosmic Variance and other blogs. The main points of his talk were: a) Perturbative string theory is in fact background independent, being a generalization of GR in a background field gauge; it's just that the perturbative framework makes the background independence non-manifest. b) Holographic dualities provide a way of archiving background independence in a more explicit way. In AdS/CFT, a gauge theory on the boundary can be manifestly diffeomorphism invariant and be equivalent to quantum gravity in the bulk of AdS. Rozali stressed that only the asymptotics of AdS (i.e. a particular negative value of the cosmological constant) need to be fixed; the interior geometry is completely dynamical. Ashtekar seemed to disagree about the extent of this statement and tried to press for a discussion in the question session, but it was interrupted for lack of time.

Klaus Fredenhagen talked about QFT in curved spacetime as a route quantum gravity. He extolled the virtues of Algebraic Quantum Field Theory and the techniques of microlocal analysis to provide a sound axiomatic foundation to QFT in curved spacetime, and explained some recent results proven in this area. Then he discussed the application of this formalism to the graviton field treated as a perturbation around a classical background, and wondered about its relation to the methods of effective field theory.

My namesake and compatriot Alejandro Perez (with whom I have been confused a couple of times for those two reasons, though we look nothing alike) gave a rather technical talk involving strings, BF theory and path-integral sum over topologies. I wish I could say more about it, but I got lost soon after the introduction.

Martin Reuter gave an exceptionally clear and compelling presentation on Asymptotic Safety in Quantum Gravity. It covered more or less all the ground that he had covered in the Zakopane lectures in March, which I will not summarise again (click on the link), but also a few tantalizing new implications for cosmology. If the results of the "Einstein-Hilbert truncation" are accepted as approximately true, then the physical cosmological constant "runs" with the scale in the following way: it is constant (at its currently observed tiny value) at lengthscales larger than 10^(-3) cm, and then starts growing as the fourth power of momentum (inverse length) until the Planck scale is reached, and from there on it grows quadratically. This means that in the early universe it was much larger than in the present but decreasing as the universe increased its scale. This provides a natural mechanism for inflation without any driving field. The inflation was driven by the same cosmological "constant" that we see today, and was due to the intrinsic running with scale of this parameter. Reuter had some calculations that seemed to show his model gives good results for the entropy of the universe, as well as a scale-invariant perturbation spectrum.

This is obviously the kind of thing that is either brilliantly right, or completely wrong. The "dark matter + small cosmological constant + inflation" model that is accepted in conventional cosmology gives predictions of extraordinary accuracy for many different observations (at least with respect to its first two elements). A lot of care would be needed to examine if Reuter's model can really emulate all the confirmed predictions, and whether it can make new ones that are testable. But if Reuter is right, then his talk was by a large margin the most important in the conference.

There were no parallel sessions on Wednesday afternoon, which was a free afternoon.

Thursday 28/07
Plenary talks

Daniele Oriti talked about Group Field Theory (GFT). According to him, GFTs (nonlocal field theories on group manifolds) can be interpreted as "second-quantized quantum gravity". They can be used as a general framework in which to rewrite discrete quantum gravity approaches such as LQG and spin foams. Oriti hopes that the elusive semiclassical limit of these theories may be more tractable with GFT methods. Instead of studying e.g. coherent semiclassical spin network superpositions, take a hugely populated "multi-particle state" of the GFT. The techniques of statistical field theory, used for the semiclassical limit of quantum mechanics in condensed matter theory, are suited to be applied to GFTs. By this way one may hope even to define notions of "temperature" and "phases" as they apply to quantum spacetime. One interesting result that he mentioned by the end, without much explanation, is that GFTs must be Fermi-quantized in the Lorentzian case and Bose-quantized in the Riemannian. Can anyone explain to me what he meant by this?

By this time I was feeling ill and with a bit of temperature (I had been warned against the local food, but...), so I went back to my hotel room to have some medicine and rest an hour or so. I thus missed David Rideout's talk on supercomputers and came back for Martin Bojowald's on effective field theory applied to LQG, on which I had put high expectations. Bojowald rewarded these expectations by dedicated one slide of his talk to quoting this blog…

…well, not exactly. The idea of the talk was to replace exact equations for quantum states by semiclassical, effective equations for a finite number of moments of a state (expectation value, fluctuation, etc.) This method is applied successfully to quantum cosmology. He hinted at the end at possible observable consequences in the inflation perturbation spectrum and at computable corrections to the Newtonian potential (meaning the 00 component of the metric in FRW cosmology). These do not seem to match those computed in Donoghue's ordinary effective field theory, but I'm not sure if this isn't because this is a different meaning of "Newtonian potential".

I kept feeling ill and missed almost all the other talks of the day, and didn't take notes in the few I attended. These included talks in the parallel sessions by Sundance Bilson-Thomson and fellow blogger Yidun Wan on models in which spin network braids are standard model particles. Next day would see Lee Smolin champion the same idea in a plenary talk. I returned early to rest in my hotel room and watch Argentina beat USA by 4-1 at football.

Friday 29/07
Plenary talks

As I was still not feeling perfectly well, I slept till late and attended only the last two morning talks. The first was by blogfriend Sabine Hossenfelder, on Phenomenological Quantum Gravity. She has written up the introduction to the talk in this post, so I can do nothing better than recommend you to read it. The rest of the talk examined the generic predictions made by models such as Minimal Length, Generalised Uncertainty Principle and Deformed Special Relativity. According to her, the main problem with all these models is an insufficient connection with fully developed fundamental theories.

Lee Smolin, as I said, talked on braided QG structures as elementary particles. He started making the point that for LQG and related models of quantum spacetime to work, it is needed to explain how low-energy excitations (gravitons, photons, etc.) can propagate through the spacetime foam without decohering with it. That is, one needs to identify "noiseless subsystems" and a ground state on which they propagate coherently, protected by an emergent symmetry. Then he presented the main result: a class of spin network models exists whose simplest coherent excitations (braided, embedded framed graphs) match the quantum numbers of Standard Model 's first generation of fermions. Higher generations can can be included, at the cost of some exotic states. Interactions can be included. (But he did not say the crucial thing: if these "interactions" match, or can be made to match, the U(1)xSU(2)xSU(3) gauge structure of the Standard Model.) Open problems are to include symmetry breaking and masses (all these degrees of freedom are massless), find momentum eigenstates and conservation laws.

I can understand Smolin's excitement about these ideas, but for the moment I remain highly skeptical about them. The Standard Model is a lot more than a table with quantum numbers, and without much more development it will be hard to convince me that the behaviour of some pretty knots can reproduce the rich mathematical structure of Quantum Field Theory.

Parallel sessions

I chose to go to the sessions centred on black holes. William Donelly gave a talk on entanglement entropy of spin networks, and its use in calculation of black hole entropy. Ashetekar expressed skepticism, saying that those calculations did not include the fact that the surface used is a black hole horizon; Donelly answered that he assumes that any surface will have entropy for some observer accelerating in a way so that the surface is a horizon to him. Daniel Termo talked on how the bulk entropy of a graph scales with its boundary, hoping to identify a "holographic regime" of LQG. The conclusion is that LQG will not be holographic, unless the Hamiltonian constrain reduces dramatically the allowed graph complexity. Bad news, I guess. Yidun Wan talked a second time, this time giving the talk of his colleague Mohammed Ansari who couldn't make it to the conference. It was on an alternative framework to the "isolated horizons" one for dealing with quantum black holes. By a reasoning I could not follow, macroscopic corrections to Hawking radiation were predicted; Ashtekar was again skeptical. Another talk worth mentioning was Jacobo Diaz-Polo's on the old problem of the black hole area spectrum in LQG. Jacobo and his collaborators did exact numerical calculations of the area degrees of freedom, without the approximations used for analytical calculations. They obtain, as usual, the Bekenstein-Hawking entropy as leading term (up to a choice of the Imirizi parameter) and a universal logarithmic correction with prefactor -1/2. The number of states as a function of the area has an interesting structure with evenly spaced peaks of degeneracy. If as a first approximation one considers only the states on the peaks, one gets an equidistant area spectrum and the Bekenstein – Mukhanov effect. Of course, all of this is purely kinematical (Jacobo himself stressed it) and the question of how to incorporate the dynamical constraint seems to remain as elusive as always.

This will be enough for today. My next and last post on the conference will describe the last day's two plenary talks and the discussion session that closed the conferemce. As always, if anyone has anything to add to my summaries, thinks I forgot something important, or wants to correct some egregious mistake, they are more than invited to do so.



  • Thanks so much for writing this excellent summary. Do appreciate.

    By Anonymous Anonymous, at 11:18 PM, July 25, 2007  

  • Thank you... whoever you are.

    By Blogger Alejandro, at 11:21 PM, July 25, 2007  

  • It's great that you are writing these. Although I attended many of the same talks it's really interesting to hear commentary from another person's perspective.

    It occurred to me only after my talk that I had not included enough background material about why the entanglement entropy is interesting as an interpretation of black hole entropy, and why we should _not_ impose boundary conditions at the horizon. Unfortunately I was much too nervous to explain these points well during the question period.

    By Blogger William, at 4:48 PM, July 26, 2007  

  • Hi William, thanks for the feedback. I had forgot that you had a blog as well! Can you give me a quick idea on why not to impose boundary conditions at the horizon?

    By Blogger Alejandro, at 10:13 PM, July 26, 2007  

  • The short version is that the horizon is not locally identifiable. The distinguishing feature of a horizon the property of acting as a causal barrier preventing signals from propagating out to an observer at infinity. Thus the event horizon is "teleological" - it can't be identified without first solving the dynamics. To an observer at the horizon it looks locally just like a piece of flat space, so there is no reason to treat it specially.

    By Blogger William, at 2:19 AM, July 27, 2007  

  • I think that the Isolated horizon framework only work on an effective level. In this sense, we should remember that an isolated horizon is a quasilocal concept. But, the problem with de dynamics is an interesting issue. We need to solve the Hawking radiation within LQG as soon as posible.

    By Anonymous Anonymous, at 7:59 PM, July 27, 2007  

  • Thank you for the article, very worthwhile material.
    employee schedule template | canned pumpkin | eyebrow rings

    By Anonymous Dominick, at 12:16 AM, November 24, 2011  

Post a Comment

Links to this post:

Create a Link

<< Home