A new continuum limit of matrix models

J. Ambjørn, R. Loll, Y. Watabiki, W. Westra, Stefan ZohrenPhysics 2008 · DOI · arXiv

Introduction

In the ongoing effort to formulate a consistent theory of quantum gravity, two-dimensional models serve as vital theoretical laboratories. For decades, the dominant framework has been Dynamical Triangulations (DT), which utilizes matrix models to sum over all possible ways to glue triangles into a surface. While mathematically rich, DT possesses a significant physical quirk: the resulting surfaces are "fractal-like" with a Hausdorff dimension of 4, diverging significantly from the smooth 2D surfaces one might expect.

To address this, Causal Dynamical Triangulations (CDT) was introduced, enforcing a strict "causal" structure by distinguishing between space and time directions. This constraint results in smoother, two-dimensional geometries. However, the precise relationship between CDT and the well-established machinery of standard matrix models remained elusive. This paper provides the missing link, demonstrating that CDT emerges as a specific, previously overlooked "limit" of a standard Hermitian matrix model.

The Challenge of Fractal Geometries

In standard DT, the continuum limit is reached by tuning the internal coupling constants of a matrix model to a critical point while taking the lattice spacing $a$ (the side length of the triangles) to zero. This process typically leads to the "Liouville" gravity phase, characterized by the aforementioned fractal surfaces. The problem is that this limit allows for unrestricted "branching" of the geometry—baby universes can sprout from any point, leading to a highly non-local and non-causal structure.

A Novel Scaling Prescription

The authors propose a departure from the standard renormalization group flow. They start with a Hermitian matrix model defined by the potential $V(\phi)$. Their key innovation is the introduction of a new coupling constant, $g_s$, which scales the entire potential:

$$V(\phi) \to \frac{1}{g_s} V(\phi)$$

Instead of the traditional approach, they perform a simultaneous scaling:

  1. The lattice spacing $a$ is taken to zero.
  2. The new coupling $g_s$ is taken to zero such that $g_s = G_s a^3$, where $G_s$ remains a finite continuum coupling.

This specific scaling behavior is designed to "squeeze" the eigenvalue distribution of the matrix. By analyzing the "disc amplitude"—a measure of the boundary of the universe—the authors show that this limit effectively suppresses the unphysical branching that plagues standard DT.

Figure 3: Geometric decomposition of the discrete disk amplitude in a matrix model, illustrating how the boundary of a surface is built up from individual matrix interactions.

Geometric Insights: Penalizing the Split

In the geometric interpretation of matrix models, certain interaction terms represent the "splitting" of a spatial universe into two. In the old scaling limit, these splitting events occur with a frequency that dominates the geometry, leading to the fractal dimension of 4.

By introducing $g_s$ and taking it to zero as $a^3$, the authors essentially impose a mathematical "penalty" on these splitting events. This penalty enforces the causality required by CDT. One of the most striking results of this work is that the continuum limit of the original matrix model is itself another matrix model. Specifically, the CDT theory can be represented by a matrix model with a simple cubic potential: $V(\Phi) \propto \Lambda \Phi - \frac{1}{6}\Phi^3$.

Figure 6: Illustration of the splitting and branching processes in the continuum limit. The new scaling limit provides a mechanism to control these branching events, recovering the smooth causal structures of CDT.

This limit also explains why CDT surfaces have a Hausdorff dimension of 2 (like a normal sheet of paper). Furthermore, in this new CDT limit, "non-universal" terms—mathematical artifacts that usually must be removed by hand in matrix models—vanish naturally, resulting in a much cleaner theoretical framework.

Results and Theoretical Unification

The paper achieves several major milestones:

  • Derivation of the CDT Disk Amplitude: The authors successfully derived the Hartle-Hawking wave function (the probability amplitude for a universe of a certain size) for CDT directly from a matrix model.
  • Universal Scaling Law: They established the precise relation $g_s \propto a^3$ as the necessary condition to reach the CDT continuum limit.
  • Generalization: The authors demonstrated that this limit is not restricted to simple cubic potentials but can be generalized to "multi-critical" models with higher-order interactions.
  • Unification: This work explains why recently developed CDT string field theories have matrix-model representations, effectively unifying two different mathematical descriptions of the same underlying physics.

Conclusion and Future Horizons

While this work primarily addresses "pure" gravity (gravity in a vacuum), it opens several exciting avenues for future research. The most immediate prospect is coupling matter, such as the Ising model, to this new matrix model limit. Because CDT geometries are more regular than those in DT, this might provide a simpler path to calculating exact results for matter on random lattices, potentially recovering the famous "Onsager exponents."

Furthermore, while this study is conducted in two dimensions, CDT is a leading candidate for 4D quantum gravity. Understanding the matrix model connection in 2D provides a rigorous analytical roadmap for exploring higher-dimensional cases and hints at deeper connections to supersymmetric gauge theories via the Dijkgraaf-Vafa correspondence.

Summary updated

BibTeX
@inproceedings{ambjrn2008new,
  author = {J. Ambjørn and R. Loll and Y. Watabiki and W. Westra and Stefan Zohren},
  title = {{A new continuum limit of matrix models}},
  booktitle = {Physics},
  year = {2008},
  doi = {10.1016/j.physletb.2008.11.003},
}
Figure