Liouville Quantum Gravity

• Liouville Quantum Gravity is a rigorous framework defining random two-dimensional surfaces via the exponential of the Gaussian free field and Gaussian multiplicative chaos.
• It unifies probabilistic techniques with conformal field theory, linking discrete planar maps to continuous models through SLE couplings and conformal welding.
• Key methodologies include Liouville first-passage percolation, metric measure analysis, and the KPZ relation, enabling insights into fractal geometry and spectral properties.

Liouville Quantum Gravity (LQG) is a mathematically rigorous theory of random two-dimensional Riemannian geometry built from the exponential of the Gaussian Free Field. LQG serves as the universal scaling limit of large random planar maps and is intimately related to conformal field theory, Schramm–Loewner Evolution (SLE), and conformal welding. It provides a canonical object whose geometry encodes both the conformal structure and the metric-measure geometry of random surfaces. Connections to SLE and Conformal Loop Ensembles (CLE) further enable its analysis via probabilistic and integrable techniques.

1. Construction and Characterization of LQG Surfaces

A $\gamma$-LQG surface (for coupling constant $\gamma\in(0,2)$) is an equivalence class of pairs $(D,h)$, where $D\subset\mathbb{C}$ is a planar domain and $h$ is a Gaussian Free Field (GFF) with possible logarithmic singularities, modulo conformal transformations: $$h' = h \circ \varphi + Q \log |\varphi'|, \qquad Q = 2/\gamma + \gamma/2.$$ The central object is the $\gamma$-Liouville area measure, formally

$$\mu_h(dz) = \lim_{\varepsilon\to 0} \varepsilon^{\gamma^2/2} \exp(\gamma h_\varepsilon(z)) dz,$$

where $h_\varepsilon$ is a mollification (typically circle-average) of $h$. The quantum boundary length measure is similarly defined via

$\gamma\in(0,2)$0

on suitable boundaries. These random measures exhibit Weyl scaling and conformal covariance, and their construction relies fundamentally on the theory of Gaussian multiplicative chaos (Berestycki et al., 2024, Ding et al., 2021, Holden et al., 18 Oct 2025).

2. Metric Structure and Volume Growth

LQG becomes a canonical random metric space by equipping the surface with the $\gamma\in(0,2)$1-LQG metric $\gamma\in(0,2)$2, constructed as the subsequential limit (in topology of local uniform convergence) of Liouville first-passage percolation (LFPP) metrics: $\gamma\in(0,2)$3 with $\gamma\in(0,2)$4 the Hausdorff (and Minkowski) dimension of the LQG metric (satisfying $\gamma\in(0,2)$5). The limiting metric $\gamma\in(0,2)$6 is uniquely characterized by the properties of being a length space, locality, Weyl scaling, and conformal coordinate change (Ding et al., 2021, Gwynne et al., 2022).

Volume growth in the LQG metric is fractal: for $\gamma\in(0,2)$7 the $\gamma\in(0,2)$8-metric ball of radius $\gamma\in(0,2)$9,

$(D,h)$0

and all positive and negative moments of $(D,h)$1 are finite (Ang et al., 2020). The KPZ relation links Hausdorff dimensions in the Euclidean and quantum geometries: $(D,h)$2 where $(D,h)$3 is the Euclidean dimension and $(D,h)$4 is the LQG (quantum) dimension (Holden et al., 18 Oct 2025).

3. Geodesics, Geometric Networks, and Metric Balls

In $(D,h)$5, geodesics exist uniquely for most pairs of points, but multiple geodesics—up to a finite deterministic bound $(D,h)$6—can arise (Gwynne, 2020). Typical geodesic networks from a base point exhibit confluence: two geodesics from the base point coincide for a positive-length initial segment before branching. The sets of points joined by exactly two or three geodesics form dense sets, each with a (conjecturally) computable fractal dimension.

Harmonic balls in LQG, defined via the mean-value property relative to $(D,h)$7, are constructed through the Hele–Shaw obstacle problem and are explicitly not $(D,h)$8-metric balls nor Lipschitz domains; their boundaries are Jordan curves (Bou-Rabee et al., 2022). Metric balls exhibit fractal boundaries, and their geometric properties underpin many scaling limits and stochastic growth models.

4. Mating-of-Trees, SLE Couplings, and Conformal Welding

LQG's deep structure is revealed through mating-of-trees: a bijective correspondence exists between certain LQG surfaces decorated by space-filling SLE of parameter $(D,h)$9 and pairs of correlated Brownian motions representing the left/right boundary processes, a construction that integrates discrete random planar map bijections (e.g., Mullin's bijection) with the continuum (Duplantier et al., 2014, Miller, 2017). The Brownian pair a.s. determines the LQG surface and the SLE curve.

Conformal welding of LQG surfaces along boundary arcs with matching quantum lengths yields interfaces that are Schramm–Loewner Evolution (SLE) curves (of $D\subset\mathbb{C}$0 linked to $D\subset\mathbb{C}$1 via $D\subset\mathbb{C}$2 or $D\subset\mathbb{C}$3), with precise welding theorems for both chordal and radial setups (Ang et al., 2024, Holden et al., 2018). The quantum zipper theorem and its critical extension guarantee the uniqueness and conformal invariance of welded surfaces, cementing the SLE-LQG coupling as a powerful tool for encoding random geometry.

CLEs (Conformal Loop Ensembles) coupled to LQG via SLEs provide a Poissonian description of loop-cut quantum disks, with interfaces and the field jointly characterized by stable Lévy processes and associated PPPs. Disk welding recovers the CLE-decorated LQG surface (Miller et al., 2020).

5. Spectral Geometry, Heat Kernel, and Diffusions

Liouville Brownian motion (LBM) is the canonical diffusion associated to the LQG geometry, defined as a time-change of Euclidean Brownian motion by the quantum clock induced by $D\subset\mathbb{C}$4 (Berestycki, 2 Dec 2025). The spectral geometry of LQG, governed by the infinitesimal generator in divergence form, exhibits a Weyl law: $D\subset\mathbb{C}$5 for the eigenvalue counting function $D\subset\mathbb{C}$6, with $D\subset\mathbb{C}$7, showing that the spectral dimension remains $D\subset\mathbb{C}$8. Heat kernel estimates and associated trace asymptotics further probe the random geometry and are under active investigation.

Numerical studies confirm the prediction $D\subset\mathbb{C}$9 for spectral dimension and reveal superdiffusive scaling with respect to the Euclidean metric, showing that the canonical diffusion on LQG is not governed by the underlying Euclidean structure but by the random metric (Bonik et al., 2014).

6. Scaling Limits, Universality, and Planar Map Convergence

LQG is the universal scaling limit of planar maps decorated by critical statistical physics models. Under discrete conformal embeddings such as the Cardy–Smirnov map or Tutte embedding, random triangulations or quadrangulations converge in the Gromov–Hausdorff–Prokhorov sense to the LQG surface $h$0 (Holden et al., 18 Oct 2025, Miller, 2017). The Brownian map, accessed via QLE(8/3,0) construction for $h$1, is a concrete realization of the $h$2-LQG metric space, and allows convergence results for discrete interfaces (self-avoiding walks, percolation, etc.) to SLE processes on the Brownian map.

Extensions to supercritical regimes $h$3 (i.e., $h$4), reveal a dramatically different geometric behavior: random surfaces lose conformal geometry and become dominated by "infinite spikes", with scaling limits corresponding to branched polymer trees (Continuum Random Tree, CRT) upon rare-event conditioning (Bhatia et al., 2024).

7. Open Problems and Future Directions

Open questions include explicit determination of the Hausdorff dimension $h$5 for all $h$6, rigorous analysis of scaling limits for models with central charge $h$7, identification of heat kernel behavior with respect to the LQG metric, and the spectral and geometric properties of LQG on higher-genus surfaces. Furthermore, the universality of KPZ scaling, new connections to quantum chaos, and the development of canonical quantum metrics in more general settings remain central challenges (Berestycki, 2 Dec 2025, Gwynne, 2019, Holden et al., 18 Oct 2025).

The LQG framework integrates probabilistic, analytic, and integrable techniques, unifying paradigms in random geometry and quantum field theory, and continues to drive the mathematical understanding of two-dimensional quantum gravity.