Posted: Jun 06, 2011 7:38 am
by twistor59
I got this from here. It's such a nice summary that I reproduce it in full:

Re: What exactly is loop quantum gravity?
It's difficult w/o math.

The problem with quantum gravity is that naive mechanisms to quantize gravity (which have been applied successfully to other fields) fail for gravity. That means that something fundamental has to be changed for quantum gravity.

There are different approaches to solve these problems, e.g.
a) string theory
b) asymptotic safety
c) loop quantum gravity (LQG)

I don't want to comment on a) and b) here.

Essentially LQG does the following: it introduces new variables which replace the (in GR) well-known metric that describes spacetime + curvature. This is pure math, so I don't want to go into details here, but what happens is that these new variables are rather close to fields that we know from gauge theory like QED and QCD. Indeed in a certain sense gravity looks rather similar to QCD, but there is one additional property of gravity that allowes one to apply a second mathematical trick which essentially replaces the fundamental fields with something like "fluxes through surfaces" or "fluxes along circles". These surfaces and circles are embedded into spacetime.

The next step is again rather technical and it becomes possible due to so-called diffeomorphism invariance: one can get rid of the the embedding of circles and surfaces into spacetime. Instead one replaces these entities with a so-called spin network, i.e. a graph with nodes and links between nodes where each link and each node carries some numbers which represent abstract entities from which certain properies of spacetime can be reconstructed. You can think about spacetime as made of cells (I will soon tell you that you can't :-); each cell has a certain volume carried by a node; each cell has certain surfaces and the link between different nodes (sitting inside these cells) carry the areas of the surfcaes.

The problem with this picture is that one might think about these cells as sitting in spacetime - but this is fundamentally wrong: this picture is only due to the construction, but basically there is no spacetime anymore; all there is are nodes and links (and certain numbers attributed to nodes and links). Spacetime is no longer fundamental but becomes an entity emerging from the more fundamental graphs with their nodes and links. The graphs are called spin networks b/c the numbers they are carrying have properties well-known from spins. But this is a mathematical property only, it does not means that there are real spinning objects.

Compare this emerging spacetime to a water surface of a lake. We know that it consists of atoms, and as soon as we get this picture it is clear that there is no water between the atoms; the surface is only an emerging phenomenon, the true fundamental objects are the atoms. In the same sense the spin networks are the fundametal entities from which spacetime, surfaces etc. and their properties like volume, area, curvature etc. can be constructed. Dynamics of spacetime (which was curvature, gravitational waves etc. in GR) is replaced by dynamics of spin networks: within a given graph new nodes with new links can appear (there are mathematical rules, but I don't want to go into detail here).

The last puzzle I have for you is the fact that such a spin network is not a mechanical object which "is" spacetime. Instead quantized spacetime is a superposition of (infinitly many) spin networks. This is well-known in quantum mechanics; there is no reason why an atom should be in a certain state; we can achieve that via preparation or measurement, but in principle a single atom can be in an arbitrary complex quantum state which is a superposition of "an atom sitting here, an atom moving in a certain direction over there, an atom moving in this or that direction, ...".

So classical spacetime is recovered by two averaging process: first there seems to be a regime were this superposition of spin networks is peaked around a single classical spacetime, i.e. were one networks dominates the superposition of infinitly many spin networks; second from this single spin network one can reconstruct spacetime in the same sense as one can reconstruct the water surface from the individual atoms. But there may be different regimes (e.g. in black holes or closed to the big bang) where is classical picture and this averaging does no longer work. It may be that in these regimes all there is are spin networks w/o any classical property like smooth spacetime, areas, volume etc. It's like looking at a single atom: there is no water surface anymore.

Eventually this is why one started with this stuff: the classical picture of spacetime seems to become inconsistent when one tries to quantize it, i.e. when one defines these superpositins etc. These inconsistencies do not bother us as long as we talk about spacetime here, in the solar system etc. But they become a pain in the a... when we talk about spacetime near a singularity like a black hole or like the big bang. In order to understand these new non-classical regimes of spacetime a fundamentally new picture is required. This is what LQG (and other approaches) are aiming for: construct a new mathematical model from which well-known classical spacetime (like in GR) can be reconstrcuted, but which does not break down in certain regimes but remains well-defined and consistent.