---
title: Chiral Topologically Massive Gravity
url: https://www.emergentmind.com/topics/chiral-topologically-massive-gravity-tmg
type: topic
---

# Chiral Topologically Massive Gravity

Chiral Topologically Massive Gravity is the critical regime of three-dimensional Einstein gravity with negative cosmological constant \(-1/\ell^2\) deformed by a gravitational Chern–Simons term with coupling \(1/\mu\). In the standard asymptotically AdS\(_3\) setting with Brown–Henneaux boundary conditions, “chiral gravity” denotes the point \(\mu\ell=1\), where the left-moving Virasoro central charge vanishes, the right-moving one becomes \(3\ell/G\), the massive graviton degenerates with the left-moving sector, and BTZ black holes obey \(\ell M=J\), motivating a conjectural purely right-moving boundary theory [0801.4566]. With Compère–Song–Strominger boundary conditions, TMG also exhibits special points at \(\mu\ell=\pm1\), where the asymptotic symmetry reduces respectively to a chiral Virasoro algebra or a pure \(\hat u(1)\) Kac–Moody current algebra [2008.06793].

## 1. Dynamical definition

Topologically Massive Gravity in three bulk dimensions is defined by the sum of the Einstein–Hilbert action with negative cosmological constant and a gravitational Chern–Simons term,
$$
S=S_{EH}+S_{CS},
$$
with
$$
S_{EH}=\frac1{16\pi G}\int d^3x\,\sqrt{-g}\,\Bigl(R+\frac2{\ell^2}\Bigr),
$$
and
$$
S_{CS}=\frac1{16\pi G\,\mu}\int d^3x\;\varepsilon^{\lambda\mu\nu}\,\Gamma^\rho_{\lambda\sigma}\Bigl(\partial_\mu\Gamma^\sigma_{\nu\rho}+\tfrac23 \,\Gamma^\sigma_{\mu\tau}\Gamma^\tau_{\nu\rho}\Bigr)\,.
$$
Varying the metric yields
$$
E_{\mu\nu}+\frac1\mu\,C_{\mu\nu}=0,
$$
where
$$
E_{\mu\nu}\equiv R_{\mu\nu}-\tfrac12\,R\,g_{\mu\nu}-\frac1{\ell^2}\,g_{\mu\nu},
$$
and
$$
C_{\mu\nu}\equiv\varepsilon_\mu{}^{\alpha\beta}\,\nabla_\alpha\Bigl(R_{\beta\nu}- \tfrac14\,R\,g_{\beta\nu}\Bigr).
$$
The Cotton tensor is symmetric, traceless, and covariantly conserved [0801.4566].

The chiral regime is not obtained by changing the field equations themselves, but by tuning the dimensionless combination \(\mu\ell\). In the original Brown–Henneaux formulation the distinguished value is
$$
\mu\ell=1.
$$
At this point the left-moving central charge vanishes, and the degeneracy structure of the linearized bulk equations changes qualitatively. This is the sense in which “chiral” denotes a critical coupling rather than a separate action [0801.4566].

## 2. Asymptotic AdS\(_3\) structure and symmetry algebras

With Brown–Henneaux asymptotics, one imposes
$$
g_{tt}=-\frac{r^2}{\ell^2}+{\cal O}(1),\quad g_{t\phi}={\cal O}(1),\quad g_{\phi\phi}=r^2+{\cal O}(1),\quad g_{r\mu}= {\cal O}(r^{-3})\,.
$$
The asymptotic symmetry generators then form two copies of the Virasoro algebra with central charges
$$
c_L=\frac{3\ell}{2G}\Bigl(1-\frac1{\mu\ell}\Bigr),\qquad c_R=\frac{3\ell}{2G}\Bigl(1+\frac1{\mu\ell}\Bigr).
$$
At the chiral point,
$$
\mu\ell=1\quad\Longrightarrow\quad c_L=0,\qquad c_R=\frac{3\ell}{G}.
$$
The vanishing of \(c_L\) is the basic asymptotic signature of chiral gravity in the Brown–Henneaux phase space [0801.4566].

A different asymptotic structure arises with Compère–Song–Strominger boundary conditions. In that phase space, the asymptotic symmetry algebra is the semi-direct product of a Virasoro algebra and a \(\hat u(1)\) Kac–Moody algebra generated by charges \(L_m\) and \(P_m\),
$$
i\{L_m,L_n\} =(m-n)\,L_{m+n}\;+\;\frac{c}{12}\,m^3\,\delta_{m+n,0},
$$
$$
i\{L_m,P_n\} =-n\,P_{m+n},
$$
$$
i\{P_m,P_n\} =k\,m\,\delta_{m+n,0}\,.
$$
For TMG, the central extensions are
$$
c(\mu)=\Bigl(1-\frac{1}{\mu\ell}\Bigr)\,\frac{3\ell}{2G},\qquad
k(\mu)=\Bigl(1+\frac{1}{\mu\ell}\Bigr)\,\frac{\ell}{G}.
$$
At \(\mu\ell=+1\), \(k=0\) and \(c=3\ell/G\), so the \(\hat u(1)\) current algebra becomes trivial and only a single chiral Virasoro algebra remains. At \(\mu\ell=-1\), \(c=0\) and \(k=2\ell/G\), so the Virasoro sector becomes trivial and the theory reduces to a pure \(\hat u(1)\) current algebra [2008.06793].

## 3. Linearized spectrum, critical degeneracy, and chirality

Around an AdS\(_3\) background \(\bar g_{\mu\nu}\) of radius \(\ell\), one writes
$$
g_{\mu\nu}=\bar g_{\mu\nu}+h_{\mu\nu},\qquad |h|\ll1.
$$
The linearized equations factorize as
$$
D_L\,D_R\,D_M\,h_{\mu\nu}=0,
$$
with first-order operators
$$
D_{L/R}{}_\mu{}^\nu=\delta_\mu{}^\nu\pm\ell\,\varepsilon_\mu{}^{\alpha\nu}\,\bar\nabla_\alpha,\qquad
D_M{}_\mu{}^\nu=\delta_\mu{}^\nu+\frac1\mu\,\varepsilon_\mu{}^{\alpha\nu}\,\bar\nabla_\alpha.
$$
There are three branches of solutions: left-movers with \((h,\bar h)=(2,0)\), right-movers with \((h,\bar h)=(0,2)\), and massive gravitons with
$$
\bigl(h,\bar h\bigr)=\Bigl(\tfrac32+\tfrac{\mu\ell}2,\,-\tfrac12+\tfrac{\mu\ell}2\Bigr).
$$
For generic \(\mu\ell>1\), the massive branch has negative energy. At \(\mu\ell=1\), the massive weights become \((2,0)\), degenerate with the left-mover, and their energy as well as that of the \((2,0)\) boundary gravitons vanishes. With suitable boundary conditions, these zero-energy modes can be treated as pure gauge or non-normalizable, leaving only right-moving gravitons [0801.4566].

Under CSS boundary conditions, linearized excitations around global AdS\(_3\) split into three finite-energy families: “massive graviton”, “boundary graviton”, and “boundary photon”. At \(\mu\ell=\pm1\), all but a single positive-energy tower acquire zero energy. More precisely, at \(\mu\ell=+1\) the boundary graviton remains with strictly positive energy while the other two families have zero energy; at \(\mu\ell=-1\) the boundary photon carries the positive energy while the graviton-type modes become zero-energy. In this phase space there are therefore no negative-energy linearized excitations at either chiral point [2008.06793].

A different perturbative picture arises when weaker logarithmic boundary conditions are admitted. At \(\mu\ell=1\), the degeneration of \({\cal D}^m\) with \({\cal D}^L\) allows logarithmic modes satisfying
$$
({\cal D}^L)^2\,h^{\rm log}_{\mu\nu}=0,\qquad {\cal D}^L\,h^{\rm log}_{\mu\nu}\neq0.
$$
These log-modes carry strictly negative energy, while the primary massive mode itself has zero energy at the critical point [1703.03630]. The distinction between strict Brown–Henneaux truncation and logarithmic boundary conditions is therefore central to the interpretation of chirality.

## 4. BTZ sector, conserved charges, and entropy

The BTZ family is
$$
ds^2=-N^2dt^2+N^{-2}dr^2+r^2\bigl(d\phi+N^\phi dt\bigr)^2,
$$
with
$$
N^2=-8G M+\frac{r^2}{\ell^2}+\frac{16G^2J^2}{r^2},\qquad
N^\phi=\frac{4GJ}{r^2}\,.
$$
In TMG, the mass and angular momentum are shifted relative to their Einstein values:
$$
\ell M(\mu)=\ell M(\infty)+\frac{J(\infty)}{\mu\ell},\qquad
J(\mu)=J(\infty)+\frac{M(\infty)}{\mu}.
$$
At \(\mu\ell=1\), every BTZ solution obeys
$$
\ell M=J.
$$
Equivalently, all classical geometries carry purely right-moving charge [0801.4566].

The same chiralization appears in the entropy formula. Assuming a unitary dual two-dimensional CFT with the Brown–Henneaux central charges, the Cardy expression reproduces the BTZ entropy including the Chern–Simons correction,
$$
S_{\rm BH}=\frac{\pi^2\ell}{3}\,\bigl(c_L\,T_L+c_R\,T_R\bigr),\qquad
T_{L/R}=\frac{r_+\mp r_-}{2\pi\ell}.
$$
At \(\mu\ell=1\), \(c_L=0\), so only the right-moving sector contributes [0801.4566].

In the CSS phase space, BTZ black holes carry shifted conserved quantities
$$
\mathcal M=M-\frac{J}{\mu\ell^2},\qquad \mathcal J=J-\frac{M}{\mu}.
$$
Regularity requires \(\mathcal M\ge|\mathcal J|\). At \(\mu\ell=\pm1\), the extremality condition becomes
$$
\mathcal M=\mp\mathcal J,
$$
producing a purely chiral black-hole spectrum. Their Bekenstein–Hawking–Chern–Simons entropy is
$$
S_{BH}=\frac{\pi r_+}{2G}-\frac{\pi r_-}{2G\,\mu\ell}.
$$
The dual Warped CFT description reproduces this entropy for all \(\mu\), and at the special points the counting collapses to either a purely chiral Virasoro expression at \(\mu\ell=+1\) or a pure \(\hat u(1)\) expression at \(\mu\ell=-1\) [2008.06793].

## 5. Holography: chiral CFT, LCFT, and WCFT interpretations

The original conjecture identifies chiral gravity at \(\mu\ell=1\) with a purely right-moving boundary CFT whose central charges are
$$
(c_L,c_R)=\Bigl(0,\frac{3\ell}{G}\Bigr).
$$
In this formulation, the disappearance of left-moving excitations and the relation \(\ell M=J\) in the BTZ sector are taken as bulk evidence for a holomorphic boundary theory [0801.4566].

Holographic renormalization at the chiral point yields a more intricate boundary structure. In Fefferman–Graham gauge, solving the full nonlinear TMG equations at \(\mu=1\) gives the generalized asymptotic expansion
$$
g_{ij}(\rho,x)
= b_{(0)ij}(x)\,\ln\rho
+ g_{(0)ij}(x)
+ \rho\Bigl[b_{(2)ij}(x)\,\ln\rho + g_{(2)ij}(x)\Bigr]
+ O(\rho^2\ln\rho).
$$
The leading logarithmic source \(b_{(0)ij}\) spoils the usual AlAdS fall-off and is dual to a logarithmic partner \(t_{ij}\) of the stress tensor. The corresponding two-point functions at the chiral point are
$$
\langle T(z)T(0)\rangle=0,\qquad
\langle T(z)t(0)\rangle=\frac{c_R}{2z^4},\qquad
\langle t(z)t(0)\rangle=-\,\frac{b}{z^4}\ln(m^2|z|^2),
$$
with
$$
c_L=0,\qquad c_R=\frac{3\ell}{G},\qquad b=-\,\frac{3\ell}{G}.
$$
This identifies the boundary theory as a \(c=0\) logarithmic CFT with a right-moving Virasoro algebra and a rank-2 Jordan block in the left sector [0906.4926].

The one-loop Euclidean partition function reinforces that interpretation. At \(\mu\ell=1\), the graviton determinant does not factorize holomorphically, and the resulting partition function has the structure expected from a logarithmic CFT. The explicit heat-kernel computation therefore supports an LCFT dual rather than a purely holomorphic one [1007.5189]. By contrast, in the strict Brown–Henneaux treatment it is consistent to set the logarithmic branch to zero so that no terms of order \(\log r\) appear, preserving the chiral truncation of the phase space [1006.3099].

With CSS boundary conditions the holographic dual is instead a Warped CFT with Virasoro\(\times u(1)\) symmetry. At \(\mu\ell=+1\) the theory reduces to a chiral Virasoro sector; at \(\mu\ell=-1\) it reduces to a pure affine \(\hat u(1)\) theory [2008.06793].

## 6. Consistency questions, non-Einstein sectors, and generalizations

The internal consistency of chiral TMG has remained tied to the status of perturbation theory about AdS\(_3\). A canonical analysis in Cartan variables at the chiral point found that the dimension of the physical phase space is two per point, corresponding to one local physical degree of freedom, the topologically massive graviton [0806.4185]. A later constraint and symplectic analysis reached a different conclusion: at \(\mu\ell=1\) about AdS\(_3\), TMG exhibits linearization instability, the linearized constraints degenerate, the symplectic form becomes non-invertible on all known perturbative modes, and naive perturbation theory fails because the linearized field equations are necessary but not sufficient [1804.05602].

Strict Brown–Henneaux boundary conditions do not eliminate all non-Einstein solutions. At the chiral point, the second Fefferman–Graham coefficient can be written as
$$
g^{(2)}_{ij}
=
\begin{pmatrix}1&1\\1&1\end{pmatrix}_{ij}\,F(t,\phi)
+
\begin{pmatrix}1&-1\\-1&1\end{pmatrix}_{ij}\,\bar L(t-\phi),
$$
where \(\bar L\) carries the usual right-moving BTZ charges while \(F(t,\phi)\) is unconstrained by the Brown–Henneaux charges. If \(F\) is not a purely left-moving profile, the resulting spacetime is genuinely non-Einstein. Linearized examples are singular at \(r=0\) and have infinite bulk energy, while an exact nonlinear solution exhibits a curvature singularity shielded by an event horizon together with geodesic repulsion and closed timelike curves [1006.3099]. A related boundary stress-tensor analysis found that such non-locally AdS\(_3\) solutions persist, can survive after adding New Massive Gravity terms, and have vanishing conserved charges \(Q[\partial_t]=Q[\partial_\phi]=0\); after Wick rotation they correspond to complex saddles of the Euclidean path integral [1108.0400].

Several generalizations preserve the chiral mechanism while modifying its field content. In spin-3 topological massive gravity, the extra local massive spin-3 mode degenerates with the left-moving one at \(\mu\ell=1\), leaving only the right-moving boundary graviton and spin-3 field; the asymptotic symmetry becomes a classical \(W_3\) algebra with \(c_R=3\ell/G\) [1106.5141]. For \({\mathcal N}=4\) extended supersymmetry, the superspace construction fixes the theory to the chiral point \(\mu\ell=1\) [1605.00103]. In AdS\(_3\) gravity with torsion, a singular point with \(m=\mu^{-1}\) and \(\Lambda=-m^2\) gives
$$
c_L=0,\qquad c_R=\frac{3\ell}{G},
$$
and reduces the bulk theory to a single \(SL(2,\mathbb R)\) Chern–Simons sector [1102.4649].

Other deformations sharpen the constraints on the chiral theory. Chiral Gravity cannot be unitarily deformed by a Fierz–Pauli mass, since the chiral-point tuning produces tachyonic roots unless the Fierz–Pauli mass is set to zero [1311.4736]. Likewise, \(f(R)\) extensions of TMG do not remove the bulk–boundary unitarity clash, because the spin-2 log-modes reappear at the corresponding chiral point with negative energy [1703.03630].

Source: https://www.emergentmind.com/topics/chiral-topologically-massive-gravity-tmg