Tuesday, January 25, 2005

Bubbling AdS space

I have not written anything about this paper by Lin, Lunin, and Maldacena:
Because I believe that this is one of the best papers in the last 6 months, let me say a couple of words.

Take type IIB string theory on AdS5 x S5 or its pp-wave limit. Both of them have the maximal number of 32 supercharges. Is there some interesting generalization of these two geometries?

The answer is: yes, there is. Both of these geometries have at least the SO(4) x SO(4) x R isometry. The pp-wave is a limit of the the anti de Sitter space. Moreover, the pp-wave limit has a Z_2 symmetry exchanging the two SO(4) factors - this symmetry is broken by the anti de Sitter space. Is there some geometric heuristic picture how to visualize these two geometries?

Yes, there is. You can imagine
  • the AdS5 x S5 space as a black disk drawn on a white paper
  • the pp-wave limit of it is a paper whose lower half-plane is black
Note that the lower half-plane is a limit of a very large disk. Also, the two half-planes filled with different color have a Z_2 symmetry, exchanging these two colors. This Z_2 symmetry does not exist for the disk whose interior and exterior look different. So far, it sounds ridiculous, of course. But the reason why I say it in this way is the following:

For any black-and-white picture that you can draw on the plane, there exists a solution of type IIB string theory - more or less, it's a geometry - with 16 supercharges. How do you construct it? Well, parameterize the ten-dimensional space by the following coordinates:
  • time "t"
  • three coordinates labeling the three-sphere S^3 number one
  • three coordinates labeling the three-sphere S^3 number two
  • a coordinate "y" which is kind of "radial"
  • two coordinates "x_1, x_2" spanning a plane that you imagine to be analogous to the "x-p" phase space
You may count that the total number of dimensions is 10. One can see that the AdS5 x S5 geometries satisfy this general Ansatz. How do we get the black-and-white pictures to the game? Draw any black-and-white pictures on the "x_1, x_2" plane. This picture expresses the behavior of the geometry near "y=0". In the regions of the "x_1, x_2" plane that are black, the S^3 number one is filled and becomes locally a flat space R^4. In the white regions of the "x_1, x_2" plane, the S^3 number two is filled and becomes a R^4. One can see that there are only two ways how to regularize the geometry near "y=0": black and white. Moreover, LLM have showed that one not only obtains a nice smooth geometry in the bulk of the black region - or, analogously, in the bulk of the white region. One also has a smooth geometry at the boundary between them.

The black regions in the "x_1, x_2" plane represent the Fermi liquid known from the matrix description of two-dimensional string theory. You may imagine that the two-dimensional string theory is embedded into the ten-dimensional type IIB string theory as a subsector. Analogous constructions, although possibly slightly less exciting ones, exist for other geometries - like the Anti de Sitter space solutions of M-theory.

So how many SUSY solutions of type IIB string theory did they obtain by this construction? A huge number. First of all, for every different topology of the black-and-white picture (a different number of "droplets" etc.), one obtains a different topology of the spacetime. If all droplets are large and their boundaries kind of straight, the curvature of the spacetime will also be small. The spacetime curvature becomes large if the droplets approach one another - a droplet eaten by a bigger droplet on the black-and-white picture describes topology changing transitions.

Even if you fix the topology, the shape of the droplet can be anything you want - and you obtain different geometries. In this sense, their Ansatz has infinitely many parameters. If you describe a boundary of a droplet as a function "x_2(x_1)" of one variable, for every function of one variable you will obtain one solution. A huge number. Of course, all these solutions have different asymptotics.

This continuously infinite number of parameters of the class of the solutions is analogous to Mathur et al. who construct their revolutionary solutions that are meant to describe the black holes, although they have neither horizon nor singularity. In that case, the solutions are also parameterized by a function of one variable - describing a shape of a string - that is dualized by various dualities to obtain a solution that looks like a black hole outside, but whose interior is very different.

Is any black-and-white picture allowed? One can see that the areas of all droplets must be actually integers (in some proper units of areas on the "x_1, x_2" plane). This requirement arises from quantization of the fluxes. In the "AdS_5 x S_5" solution, for example, the black-and-white picture is a black disk. Its area is proportional to "N", the five-form flux through the five-sphere. The classical geometry is only appropriate if the droplets are large and their curvature is small.

Therefore it sounds reasonable to imagine that the "x_1, x_2" plane is noncommutative, like a phase space, and the quantum of the area is a single cell of this phase space. The function "z" that equals +1/2 in the black regions and -1/2 in the white regions could really be a function on a non-commutative space that satisfies "z*z=1/4" where "*" is the non-commutative star-product. Anyone has a way to see that such a description is possible? There could be some "dual" object - like a D3-brane that can wrap either of these spheres S^3. The coordinates "x_1, x_2" would be fields living on the worldvolume of this dual object, and one should be able to show that they don't commute and the commutator is the right c-number. Such an object could be in various states, and a lowest energy state would correspond to the field "z(x_1,x_2)" that describes the black-and-white picture. Note that in the normal picture of string theory, "z" parameterizes the geometry (and the RR field strengths) and therefore we treat it as a closed string field. Near y=0, however, there could be a dual way to describe physics in which geometry comes from quantization of this "new kind of object" that sees a non-commutative "x_1, x_2" plane.

Any comments related to this paper are welcome.


  1. Just a brief comment on your explanation and then I will look at article.

    Lubos:Is there some geometric heuristic picture how to visualize these two geometries?This is the first time for me I have seen you refer to the geometries this way, and the way you lead into the visualizations.

    This is quite productive to me, in understanding the environment, with which you lead us through.

    AS a laymen I thank you for the clearness with which you refer to the article.

    When one considers Lenny's insightual vision of the loop, we know such visualizations have to be clear to make a impression. A culmination of sorts(mathematically) of the natural world, where such mini blackholes could be described?:)

  2. This comment has been removed by a blog administrator.

  3. Dear Quantoken,
    I always mean what I say and vice versa.

    For example, now I say that the paper must definitely be uninteresting for simpletons - please, Quantoken, take this comment personally - but if one clicks "cited by", she will see that I am certainly not the only one who finds the paper highly interesting - it may be the most cited theoretical paper since September.


  4. I agree it is a very interesting paper, but certainly not the most cited since September. Currently it has 19 citations, but hep-th/0409245 (which appeared a week later) has 29 citations --- this would in fact be my bet for the paper with the most citations since the day LLM appeared. There are other papers which come close. For example hep-th/0410224 already has 18 citations, only one less than LLM even though it appeared five weeks later.

  5. High energy physicists are citation-crazy.

  6. "Currently it has 19 citations, but hep-th/0409245 (which appeared a week later) has 29 citations --- this would in fact be my bet for the paper with the most citations since the day LLM appeared. There are other papers which come close. For example hep-th/0410224 already has 18 citations, only one less than LLM even though it appeared five weeks later.
    Well, I sincerely hope that this will make Lubos stop and think. I'm sure he knows that these twistor papers are not even one-tenth as deep and significant as the LLM paper -- and yet they are ahead in citations. Which only adds one more data point showing that citations mean very little. It is very [self-] destructive for a young researcher to measure his worth by citations. Maldacena clearly doesn't care about citations -- a lot of his recent papers are on subjects that are certain never to attract them -- but that doesn't change the fact that he is a leader and not a follower. I'd rather have my name on any of his papers than write more junk about MHV amplitudes.

  7. This comment has been removed by a blog administrator.

  8. Some comments are obviously too sensitive in academia. I did not say that I measure the importance of papers by citations, nor I believe it. ;-)

    I just needed an independent argument against Quantoken's statement that LLM is an uninteresting paper - a statement that shows nothing else than Quantoken's ignorance.

    Twistors are now an industry, and the papers mentioned above are interesting and admirable, too. LLM is an original approach, however.

  9. Dear readers,

    could you please try to concentrate on the topic - which is Bubbling AdS space - instead of filling the comments section by rubbish? I know that the presence of Quantoken is, due to his or her mostly innate inability to think rationally, highly distracting, but I believe that it should not prevent others from using their brains.


  10. To the 5:03 AM anonymous poster: I find it interesting that you refer to Witten's paper hep-th/0409245 as "more junk". I certainly agree that Maldacena is a leader in the field, but he is clearly not as powerful a leader as Ed Witten (although, for several years I have been predicting that Maldacena will eventually take up the torch when Ed passes it on). So when Lubos speculated that LLM might be the most cited paper since September, it was completely obvious how I could check, and then disprove, this suggestion --- just find a contemporary paper by Witten!

  11. LLM make an analogy with topological string
    theory in their paper. There have been some
    suggestions that it has something to do with
    c=1. Any thoughts?