---
title: Superluminal Liouville Walls in 2D String Theory
url: https://www.emergentmind.com/topics/superluminal-liouville-walls
type: topic
---

# Superluminal Liouville Walls in 2D String Theory

Searching arXiv for the provided papers and closely related work to ground the article.
arXiv search query: "Superluminal Liouville walls 2d String Theory and space-like singularities"
arXiv search query: "On Time-Dependent Backgrounds In 1+1 Dimensional String Theory Liouville wall"
arXiv search query: "SL(2,R) Chern-Simons Liouville Gauge Theory Duality Walls"
Superluminal Liouville walls are time-dependent Liouville or tachyon profiles in \(1+1\)-dimensional string theory for which the target-space wall defined by the Liouville potential moves faster than light once a control parameter crosses a critical value. In the \(c=1\) noncritical bosonic string, this occurs when the parameter \(r\) or \(p\) exceeds \(1\), so that the wall trajectory becomes space-like; in that regime the usual S-matrix ceases to exist, and in the dual \(c=1\) matrix model the emergent spacetime perceived by collective-field fluctuations terminates at a space-like boundary where couplings diverge [2509.12778]. The term requires terminological care: in a distinct literature on duality/domain walls, “Liouville walls” refer instead to interfaces implementing mapping-class-group actions and encoded by Liouville or quantum-Teichmüller structures, with no notion of superluminal propagation [1103.5748].

## 1. Definition and kinematic criterion

In the worldsheet description of the \(c=1\) bosonic string, the relevant fields are the target time \(t\equiv X^0\) and the Liouville mode \(\phi\), with a linear dilaton background and a tachyon/Liouville potential. A convenient form is
\[
S_{\rm ws}=\frac{1}{4\pi}\int d^2\sigma \left[(\partial t)^2+(\partial \phi)^2+Q R^{(2)} \phi \right]+\int d^2\sigma\,V_{\rm WS}(\phi,t),
\]
with \(Q=2\) in the standard \(c=1\) theory, and
\[
V_{\rm WS}=- \mu \phi e^{2\phi} + \frac{\Gamma(r)}{\Gamma(1-r)} \int d^2\sigma\, e^{(2-r)\phi} (\lambda_+ e^{rt} + \lambda_- e^{-rt}) .
\]
A parallel parametrization writes the Lorentzian deformation as
\[
\delta S_L=\lambda_+T_p+\lambda_-T_{-p}=\frac{\Gamma(p)}{\Gamma(1-p)}\int d^2z\,e^{(2-p)\phi}\left(\lambda_+e^{pt}+\lambda_-e^{-pt}\right),
\]
with \(0<p<2\) so that the perturbation is relevant near the boundary [2509.12778].

In the weak-coupling asymptotic region \(\phi\to+\infty\), the moving wall is defined by the level set
\[
f(t,\phi)\equiv (2-r)\phi+rt=\text{const.}
\]
Equivalently, in the spacetime zero-mode picture one defines the wall by \(V_{\rm st}(t,\phi)=1\), with
\[
V_{\rm st}(t,\phi)=\mu e^{2\phi}+\lambda_+e^{(2-p)\phi+pt}
\]
when \(\lambda_-=0\). Solving asymptotically gives
\[
\phi_{\rm wall}(t)=\phi_0-\frac{r}{2-r}t,
\qquad
\frac{d\phi_w}{dt}=-\frac{p}{2-p},
\]
so the wall speed is
\[
v_{\rm wall}=\left|\frac{d\phi_{\rm wall}}{dt}\right|=\frac{r}{2-r}
\quad\text{or}\quad
\left|\frac{d\phi_w}{dt}\right|=\frac{p}{2-p}.
\]
Using the flat target metric \(ds^2=-dt^2+d\phi^2\), the normal \(n_\mu=\partial_\mu f=(r,2-r)\) has norm
\[
n^2=g^{\mu\nu}\partial_\mu f \partial_\nu f=-r^2+(2-r)^2=4-4r.
\]
Hence the wall is time-like for \(r<1\), null at \(r=1\), and space-like for \(r>1\); equivalently, \(v_{\rm wall}<1\), \(=1\), or \(>1\). The same threshold appears in the \(p\)-parametrization: \(p<1\) gives a timelike wall, \(p=1\) a null wall, and \(1<p<2\) a spacelike or superluminal wall [2311.17992].

For \(\lambda_+>0\) and \(p<1\), both past and future null infinities remain part of the boundary and asymptotic in/out tachyons are well-defined. For \(1<p<2\), the wall becomes spacelike at late times and overtakes outgoing signals; future \(\mathcal{I}^+\) is shielded, so standard S-matrix out-states do not exist. At the threshold \(p=1\), the wall asymptotically travels at the speed of light and future \(\mathcal{I}^+\) pinches off [2311.17992].

## 2. Worldsheet background, asymptotic modes, and acceleration

The Euclidean worldsheet action for the static background is
\[
S_E=\frac{1}{4\pi}\int d^2z\,\sqrt{\hat g}\,\left[(\hat\nabla X)^2+(\hat\nabla\phi)^2+2\phi R(\hat g)-4\pi\mu\,\phi\,e^{2\phi}\right],
\]
and Wick rotation \(X\to-it\) gives the Lorentzian action
\[
S_L=\frac{1}{4\pi}\int d^2z\,\left[-(\nabla t)^2+(\nabla\phi)^2-4\pi\mu\,\phi\,e^{2\phi}\right].
\]
Near the boundary \(\phi\to-\infty\), the massless closed-string tachyon operators behave as
\[
T_p\simeq \frac{\Gamma(|p|)}{\Gamma(1-|p|)}\int d^2z\,e^{(2-|p|)\phi}e^{ipX}
\quad\Rightarrow\quad
\mathcal{T}^\pm_\omega\simeq \frac{\Gamma(-i\omega)}{\Gamma(1+i\omega)}\int d^2z\,e^{(2+i\omega)\phi}e^{\mp i\omega t},
\]
with wavefunctions asymptotically \(\sim e^{i\omega(\phi\mp t)}\) for incoming and outgoing massless modes [2311.17992].

The time-dependent deformation interpolates between an early-time static regime and a late-time constant-velocity regime. The transition is controlled by the characteristic point
\[
(\phi^*,t^*)=\Big(-\tfrac{1}{2}\ln\mu,\,-\tfrac{1}{p}\ln\lambda_+ + \tfrac{2-p}{2p}\ln\mu\Big),
\]
which controls energy-violating processes. This is the acceleration region in which time dependence is dynamically important, while the asymptotic wall velocity is set by \(p/(2-p)\) [2311.17992].

In lightlike coordinates \(u=t-\phi\) and \(v=t+\phi\), the spacelike regime admits a maximal accessible \(v\) along the wall. For \(\lambda_-=0\),
\[
e^{pv_{\rm max}}=\frac{(p-1)^{p-1}}{p^p}\,\frac{1}{\lambda_+\mu^{p-1}},
\]
so outgoing observables that rely on arbitrarily large positive \(v\) fail once the wall is spacelike. When both \(\lambda_\pm\) are nonzero, the potential becomes
\[
V_{\rm st}(t,\phi)=\mu e^{2\phi}+\lambda_+e^{(2-p)\phi+pt}+\lambda_-e^{(2-p)\phi-pt},
\]
and the ceiling generalizes to
\[
e^{pv_{\rm max}}=\frac{(p-1)^{p-1}}{p^p}\,\frac{F(s)}{\lambda_+\mu^{p-1}},
\qquad
F(s)\left(1+\frac{p^ps}{(p-1)^p(1-s)^{2-p}F(s)}\right)^{p-1}=1.
\]
The same two-sided background admits a Hartle–Hawking construction, and the parameter
\[
\frac{s}{(1-s)^{2-p}}=(p-1)\lambda_+\lambda_-\,\mu^{p-2}
\]
controls the acceleration in the far past and future [2311.17992].

## 3. Dual matrix-model formulation

The exact dual description is the singlet sector of the \(c=1\) matrix model, i.e. \(N\) free fermions in an inverted harmonic potential with single-particle Hamiltonian
\[
h=\frac{1}{2}(p^2-x^2).
\]
The ground-state Fermi surface is
\[
x^2-p^2=2\mu.
\]
Time-dependent deformations that realize moving Liouville walls are canonical transformations
\[
(x \pm p) \to (x \pm p) + \lambda_\pm e^{\pm r t} (x \mp p)^{r-1},
\]
yielding
\[
x^2-p^2 + \lambda_- e^{- r t} (x+p)^r + \lambda_+ e^{ r t} (x-p)^r + \lambda_+ \lambda_- (x^2-p^2)^{r-1} = 2 \mu .
\]
For \(\lambda_-=0\), defining \(x_\pm=x\pm p\), the Fermi surface simplifies to
\[
x_+x_-+\lambda_+e^{rt}x_-^r=2\mu,
\]
and the paper adopts units \(2\mu=1\). A useful parametrization is
\[
x(\alpha,t)=\cosh \alpha - \frac{\lambda_+}{2} e^{rt} e^{-(r-1)\alpha},
\qquad
p(\alpha,t)=\sinh \alpha - \frac{\lambda_+}{2} e^{rt} e^{-(r-1)\alpha}.
\]
In the alternative notation of the free-fermion formulation, the exact deformation with \(\lambda_-=0\) is
\[
\lambda=a_0\cosh w+a_+e^{(1-p)w+pt},
\qquad
p_\lambda=a_0\sinh w+a_+e^{(1-p)w+pt},
\]
with
\[
a_0=-\sqrt{2\mu},
\qquad
a_+=-\frac{1}{\sqrt{2}}\lambda_+\mu^{\frac{p-1}{2}},
\]
implying
\[
\lambda^2-p_\lambda^2+\frac{2a_+}{(-a_0)^{p-1}e^{pt}(p_\lambda-\lambda)^p}=a_0^2.
\]
The Liouville wall coincides with the intersection \(p_\lambda=0\) of the Fermi surface with the \(\lambda\)-axis and agrees with \(V_{\rm st}=1\) in the bulk description [2509.12778].

Small ripples, identified with tachyon excitations, move along classical trajectories
\[
\lambda(t)=-\rho \cosh(t-\sigma),
\qquad
p_\lambda(t)=-\rho \sinh(t-\sigma),
\qquad
\sigma(\rho)=\frac{1}{p}\ln\frac{(\rho^2-a_0^2)(-a_0)^p}{2a_0a_+\rho^p}.
\]
For \(p<1\), these ripples are reflected and reach \(\mathcal{I}^+\), defining an S-matrix. For \(p>1\), there is a critical
\[
\rho_c=\sqrt{\frac{p}{p-1}}\,|a_0|,
\]
such that if \(\rho>\rho_c\) the ripple never reaches the tip and stays trapped. The maximum accessible \(v\) occurs at \(\rho=\rho_c\) and equals \(v_{\rm max}\) from the worldsheet analysis [2311.17992].

## 4. Collective field, emergent metric, and space-like termination

The collective field \(\partial_x\phi(x,t)\) and its conjugate momentum \(\Pi_\phi\) are
\[
\partial_x \phi = \frac{1}{2\pi}\int dp\, u(x,p,t),
\qquad
\Pi_\phi\,\partial_x\phi = - \frac{1}{2\pi g_s^2}\int dp\,p\,u(x,p,t).
\]
For quadratic profiles, if \(P_\pm(x,t)\) are the two intersections with the Fermi surface, then
\[
\partial_x\phi=\frac{1}{2\pi}(P_+-P_-),
\qquad
\Pi_\phi\,\partial_x\phi=-\frac{1}{4\pi g_s^2}(P_+^2-P_-^2).
\]
The classical action is
\[
S=\frac{1}{g_s^2}\int dt\,dx \left[ \frac{(\partial_t\phi)^2}{\partial_x\phi} - \frac{\pi^2}{6}(\partial_x\phi)^3 + \frac{(x^2-1)(\partial_x\phi)}{2} \right].
\]
Using Hamilton’s equations,
\[
\partial_x\phi=\frac{1}{2\pi}(P_+-P_-),
\qquad
\frac{\partial_t\phi}{\partial_x\phi}=-\frac{1}{2}(P_++P_-).
\]
Expanding
\[
\phi=\phi_0+\frac{g_s}{\sqrt{\pi}}\eta,
\]
one obtains the quadratic action
\[
S^{(2)}=\frac{1}{2\pi}\int dx\,dt \left[ \frac{(\partial_t\eta)^2}{\partial_x \phi_0}
- 2 \frac{\partial_t \phi_0}{(\partial_x \phi_0)^2} (\partial_t\eta)(\partial_x\eta)
+ \left( \frac{(\partial_t \phi_0)^2}{(\partial_x \phi_0)^3} - \pi^2 \partial_x \phi_0 \right)(\partial_x\eta)^2 \right] .
\]
This is a free massless scalar in \(1+1\) dimensions propagating on an acoustic metric conformal to
\[
ds^2 = - dt^2 + \frac{\left(dx + \frac{\partial_t \phi_0}{\partial_x \phi_0} dt\right)^2}{(\pi \partial_x \phi_0)^2},
\]
with Lorentzian signature and
\[
\det g = - \frac{1}{\pi^2 (\partial_x \phi_0)^2}.
\]
For quadratic profiles there exist global coordinates \((\tau,q)\),
\[
\tau = t - \frac{1}{2}(\alpha_+ + \alpha_-),
\qquad
q = \frac{1}{2}(\alpha_+ - \alpha_-),
\]
that make the metric manifestly conformal to Minkowski space [2509.12778].

These coordinates sharply distinguish the timelike and superluminal regimes. For \(0<r<1\), in the asymptotic region \(\psi\equiv \operatorname{arcosh}x\gg t\gg 1\),
\[
q=\psi,\qquad \tau=t,
\]
so the emergent spacetime is conformal to the full Minkowski strip with a time-like wall at \(q=0\). At the critical point \(r=1\),
\[
q=\operatorname{arcosh}\!\left(x+\frac{\lambda_+}{2}e^t\right),
\qquad
\tau=t,
\]
and the emergent spacetime is the full Minkowski strip with a static mirror at \(q=0\). For \(1<r<2\), as \(t\to\infty\), a constant-\(t\) slice asymptotically approaches
\[
\tau=\frac{2-r}{r}\,q,
\]
while for \(r>2\),
\[
\tau=-\frac{r-2}{r}\,q.
\]
In both cases, the large-\(t\) evolution approaches a straight space-like line from below in the \((\tau,q)\) plane; the matrix-model time ends at this boundary and the emergent spacetime terminates there. Characteristics for \(\eta\) are null lines
\[
\tau\pm q=\text{const.}
\]
This provides the geometrical meaning of the absence of an asymptotic future in the superluminal regime [2509.12778].

## 5. Scattering, particle production, and the loss of asymptotic out-states

For \(p<1\), the theory admits asymptotic observables and one can compute scattering amplitudes by a Wick-rotation strategy in which Euclidean correlators are evaluated with \(X\) compactified on a circle of radius \(R\), choosing \(pR=1\), then Fourier transforming to Euclidean position space, Wick rotating to Lorentzian signature, and Fourier transforming back to energy space. In the static benchmark \(\lambda_+=0\), the \(n\to1\) amplitude at finite temperature is
\[
\Big\langle\prod_{j=1}^n\mathcal{T}^+_{\omega_j}\,\mathcal{T}^-_{\omega'}\Big\rangle
= -(-2\pi)^{-n+1}\partial_\mu^{n-2}\frac{1}{\mu}\,\delta\!\left(\omega'-\sum_j\omega_j\right)\,\prod_{j=1}^n\mu^{-i\omega_j}\,g(\omega_j R),
\]
with
\[
g(\omega R)= -\frac{2\pi i}{1-e^{-2\pi \omega R}}
= -\frac{2\pi i}{1-e^{-\omega/T}},
\qquad
T=\frac{p}{2\pi}.
\]
As \(R\to\infty\),
\[
\Big\langle\prod_{j=1}^n\mathcal{T}^+_{\omega_j}\,\mathcal{T}^-_{\omega'}\Big\rangle
= 2\pi\,i^n\,\delta\!\left(\omega'-\sum_j\omega_j\right)\,\partial_\mu^{n-2}\,\mu^{-i\sum_j\omega_j-1}.
\]
Time dependence breaks energy conservation, and the amplitudes are dominated by the acceleration region near \((\phi^*,t^*)\). For absorption with \(n\) incoming and no outgoing particles,
\[
\Big\langle\prod_{j=1}^n\mathcal{T}^+_{\omega_j}\Big\rangle\sim \mu^a\lambda_+^b,
\qquad
a=2-n-\frac{i}{p}\sum_j\omega_j,
\qquad
b=\frac{i}{p}\sum_j\omega_j,
\]
so each incoming energy contributes the phase
\[
\left(\frac{\lambda_+}{\mu}\right)^{\frac{i\omega}{p}}
=\exp\!\left[\frac{i\omega}{p}\ln\!\left(\frac{\lambda_+}{\mu}\right)\right].
\]
For emission with no incoming particles, defined for \(p<1\),
\[
\Big\langle\prod_{l=1}^{n}\mathcal{T}^-_{\omega_l}\Big\rangle\sim \mu^{\,i\,\frac{1-p}{p}\sum_l\omega_l-n+2}\,\lambda_+^{-\frac{i}{p}\sum_l\omega_l},
\]
with phase per outgoing mode
\[
(\lambda_+\mu^{p-1})^{-\frac{i\omega}{p}}
=\exp\!\left[-\frac{i\omega}{p}\ln\!\big(\lambda_+\mu^{p-1}\big)\right].
\]
The exact Lorentzian \(n\to1\) amplitude for \(p<1\) is
\[
\Big\langle \prod_{j=1}^n\mathcal{T}^+_{\omega_j}\,\mathcal{T}^-_{\omega'}\Big\rangle_{\lambda_+}
= -(-2\pi)^{-n}\frac{\Gamma(\alpha)\Gamma(\beta)}{p\,\Gamma(\alpha+\beta+1)}\,\partial_\mu^{n-1}\,\mu^{-\beta}\,\lambda_+^{-\alpha}\,\prod_{j=1}^n g(\omega_j/p),
\]
with
\[
\alpha=\frac{i}{p}\left(\omega'-\sum_{j=1}^n\omega_j\right),
\qquad
\beta= \frac{i}{p}\sum_{j=1}^n\omega_j+i\frac{p-1}{p}\,\omega'=i\omega'-\alpha.
\]
The energy-conserving limit \(\alpha\to0\) is
\[
\Big\langle \cdots \Big\rangle_{\lambda_+}\to (-2\pi)^{-n}\left(-\frac{\ln\lambda_+}{p}\right)\partial_\mu^{n-2}\,\mu^{-i\omega'-1}\,\prod_{j=1}^n g(\omega_j/p).
\]
The one-point function is
\[
\Big\langle \mathcal{T}^-_{\omega'}\Big\rangle_{\lambda_+}
= \frac{\Gamma\!\big(\frac{i}{p}\omega'\big)\,\Gamma\!\big(i\,\frac{p-1}{p}\omega'-1\big)}{p\,\Gamma(i\omega'+1)}\,\mu^{-i\frac{p-1}{p}\omega'+1}\,\lambda_+^{-\frac{i}{p}\omega'},
\]
and the connected \(n\)-point emission amplitude is
\[
\Big\langle \mathcal{T}^-_{\omega_1}\cdots \mathcal{T}^-_{\omega_n}\Big\rangle_{\lambda_+}
=\left(-\frac{1}{2\pi}\right)^{n-1}\frac{1}{p}\,\partial_\mu^{n-3}\frac{1}{\mu}\prod_{j=1}^n\frac{\Gamma\!\big(\tfrac{i\omega_j}{p}\big)\,\Gamma\!\big(\tfrac{p-1}{p}i\omega_j+1\big)}{\Gamma(i\omega_j+1)}\,(\lambda_+\mu^{p-1})^{-\frac{i\omega_j}{p}}.
\]
Using the \(\omega'\to0\) limit and
\[
\lim_{\omega\to 0}\frac{2\pi\,i\,\omega\,\mathcal{T}^-_\omega}{p}=-\partial_\mu,
\]
the purely absorptive amplitude with no outgoing particles vanishes for all \(p<1\):
\[
\Big\langle\mathcal{T}^+_{\omega_1}\cdots\mathcal{T}^+_{\omega_n}\Big\rangle_{\lambda_+}=0.
\]
For \(p>1\), these out-state amplitudes cease to be the appropriate observables, because future \(\mathcal{I}^+\) is shielded by the spacelike wall. Position-space observables remain well-defined only for \(v<v_{\rm max}\), and they describe radiation emitted before the wall turns spacelike [2311.17992].

## 6. Singular boundary, strong coupling, and terminological ambiguity

The superluminal regime is characterized not only by the absence of a conventional asymptotic future but also by the appearance of a space-like singularity in perturbative collective field theory. The cubic action for the fluctuation \(\eta\) is
\[
S^{(3)} = -\frac{1}{\pi^{3/2}} \int dx\,dt \left[ \frac{1}{2} \frac{(\partial_t \eta)^2 (\partial_x \eta)}{(\partial_x \phi_0)^2}
- \frac{\partial_t \phi_0}{(\partial_x \phi_0)^3} (\partial_t \eta)(\partial_x \eta)^2
+ \left( \frac{(\partial_t \phi_0)^2}{(\partial_x \phi_0)^4} + \frac{\pi^2}{6} \right)(\partial_x \eta)^3 \right] .
\]
Along the space-like boundary, at least one coupling diverges. For \(r=2\), \(\lambda_+<0\), as \(\tau\to0\),
\[
\frac{(\partial_t \phi_0)^2}{(\partial_x \phi_0)^4}
\sim \frac{e^{4\tau} (\tanh q)^2}{(1-e^{2\tau})^2}\to\infty.
\]
For \(1<r<2\), along
\[
\tau=\frac{2-r}{r}q,
\]
one finds
\[
\frac{(\partial_t \phi_0)^2}{(\partial_x \phi_0)^4}
\sim
\frac{e^{2( t + (2(r-1)/r) q )}}
{\left( \frac{1}{2} e^{ t - (2(r-1)^2/r) q } + \frac{1}{2} e^{ -t + (2/r) q } \right)^2},
\]
which diverges along the boundary. For \(r>2\), near
\[
\tau=-\frac{r-2}{r}q,
\]
one has
\[
\frac{(\partial_t \phi_0)^2}{(\partial_x \phi_0)^4}
\sim
\frac{e^{ 2( t + (2(r-1)/r) q ) }}
{[ \sinh( t - (2/r) q ) ]^2},
\]
which again diverges along the boundary. The fermionic theory itself remains well-defined; the breakdown occurs in the perturbative bosonic, or collective, description. Finite \(N\) effects and finite \(g_s\) effects are therefore needed near the space-like boundary [2509.12778].

The phrase “superluminal Liouville walls” should also be distinguished from an unrelated use of “Liouville walls” in the study of duality walls. In that setting, the three-manifold is
\[
M=\Sigma\times I,
\]
with \(\Sigma=\Sigma_{g,h}\) a punctured Riemann surface and \(I=[0,1]\), and the proposed correspondence equates an \(SL(2,\mathbb{R})\) Chern–Simons partition function with the \(S^3_b\) partition function of a \(3d\) \( \mathcal{N}=2\) superconformal field theory living on a duality wall in a \(4d\) class \(S\) theory:
\[
Z_{\left[T[SU(2);\varphi;m]\right]}\!\left[S^3_b\right](a,a')=Z_{\text{CS}\!\left[M_{\Sigma,(l,l'),\varphi}\right]}(m,k),
\]
with
\[
a=l,\qquad a'=l',\qquad \hbar:=\frac{4\pi}{k+2}=2\pi b^2.
\]
The wall partition function is identified with a mapping-class-group matrix element,
\[
Z_{3\text{d}}(a,a')=\varphi_{l,l'}=\langle l|\varphi|l'\rangle,
\]
and, after gluing, with a trace on the mapping torus. In this literature, the “walls” are interfaces or operator insertions implementing S-duality across the wall, encoded by Liouville conformal blocks and quantum Teichmüller theory; the paper explicitly states that there is no notion of superluminal propagation anywhere in the construction [1103.5748].

This terminological distinction matters because the two subjects share Liouville-theoretic structures while referring to different physical objects. In the \(c=1\) string context, a superluminal Liouville wall is a time-dependent background potential whose spacelike trajectory shields null infinity and ends the emergent spacetime on a space-like boundary. In the duality-wall context, “Liouville walls” are modular interfaces in class \(S\) dynamics whose exact \(S^3_b\) partition functions match \(SL(2,\mathbb{R})/SL(2,\mathbb{C})\) Chern–Simons amplitudes. The shared terminology does not imply shared causal physics.

Source: https://www.emergentmind.com/topics/superluminal-liouville-walls