The edge has more to say

If a theory has no local gravitational waves, where could its states hide? One answer begins with an apparently dull task: specify what the metric is allowed to do far away.

Boundary conditions sound like the fine print beneath the physics. In anti-de Sitter gravity, they are part of the physics. They tell us what we hold fixed, which variations belong to the same problem, and which transformations can carry a charge.

The edge is not simply where the calculation stops. It helps decide what the calculation means.

A change of coordinates, with consequences

General relativity teaches us to be suspicious of coordinates. Two metrics related by a harmless change of labels should not be counted as two universes. But “harmless” depends on what happens at the boundary.

Think of moving every object in a room together with the ruler used to locate them. That changes no relative measurement. Now imagine the ruler is attached to a wall that the experiment holds fixed. A transformation that acts on the wall needs a different assessment. This is only an analogy, but it points toward the distinction between gauge redundancies with vanishing charges and asymptotic symmetries with nonzero charges.

Brown and Henneaux examined asymptotically AdS₃ metrics under specific boundary conditions in 1986. The resulting symmetry was much larger than the finite-dimensional isometry group of exact AdS₃: the charge algebra contains two Virasoro algebras.^[1] The boundary supports an infinite family of transformations.

There is an especially memorable number attached to them:

Here \(\ell\) is the AdS radius and \(G\) is the three-dimensional Newton constant, with \(\hbar=1\). The central extension already arises in the classical charge algebra. It is not something that must first wait for a quantum anomaly.

A clue, not a completed dictionary

If you know two-dimensional conformal field theory, “two Virasoro algebras” is a very familiar sound. It suggests left- and right-moving conformal symmetry. But hearing the symmetry is not the same as identifying the theory.

An analogy is recognizing the rules of chess without knowing the position on the board. The rules constrain every legal move. They do not tell you which pieces are present, how the game arrived here, or which legal position you are looking at. Similarly, a symmetry algebra does not specify the full spectrum and operator products of a CFT.

This distinction makes the Brown–Henneaux result more interesting, not less. It supplies a powerful piece of a future boundary description without pretending to supply every piece. It is a clue precise enough to calculate with.

Years later, the central charge would enter Strominger’s use of CFT state-counting asymptotics to reproduce black-hole entropy.^[2] We will take that step after meeting the black hole. For the moment, notice the shape of the argument: geometry fixes a symmetry datum, and that datum will help constrain a huge number of states.

The fine print is the plot

Different boundary conditions can define different physical problems. We should therefore avoid saying that every version of AdS₃ gravity automatically has exactly this boundary phase space. The celebrated result comes with a choice, and the choice is doing work.

This becomes a recurring habit in the series. Whenever someone says “the theory of three-dimensional gravity,” ask what they allow at infinity. A beautiful answer to one boundary-value problem need not answer another.

We started with a theory that seemed to have too few degrees of freedom. Now its boundary knows an infinite-dimensional algebra. Next comes another surprise: its bulk equations can be reorganized in the language of gauge theory.