IIB Matrix Model Regularization

• The paper introduces the IIB matrix model that replaces worldsheet fields with large-N matrices, offering a UV-finite, nonperturbative regularization framework.
• It shows how emergent spacetime arises dynamically from the eigenvalue spectrum of the matrices, with spontaneous symmetry breaking selecting a (3+1)-dimensional universe.
• The study applies techniques like Gaussian expansion, Monte Carlo simulations, and the Complex Langevin Method to handle the complex-action problem and enforce quantum causality.

Matrix model regularization of type IIB superstrings is a framework in which the path integral of type IIB superstring theory is given a nonperturbative, background-independent definition by replacing the worldsheet fields with large $N$ matrices, leading to the so-called IIB (IKKT) matrix model. This approach provides a concrete ultraviolet (UV) regularization of the sum over worldsheet geometries, encodes spacetime as an emergent phenomenon from matrix dynamics, and implements quantum features such as stringy causality in a manifestly gauge-invariant manner. Both Euclidean and Lorentzian (Minkowskian) versions exist, with recent developments clarifying their interrelation, regularization subtleties, and physical consequences for spacetime emergence and string interactions.

1. Formulation: From Schild and Polyakov Actions to Matrix Regularization

The critical (D=10) type IIB Green–Schwarz superstring admits several equivalent continuum worldsheet actions: Polyakov-type, Schild-type, and Nambu–Goto-type. In the Schild formulation (after suitable gauge and $\kappa$-symmetry fixing), the action reads

$$S_{\rm Schild} = \frac{1}{2\pi} \int d^2\sigma\, e(\sigma) \left[ \frac{1}{4} \{X^\mu, X^\nu\}^2 + i\, \psi^T \Gamma_\mu \{X^\mu, \psi\} \right],$$

where $\{f,g\}$ denotes the Poisson bracket on the worldsheet. Gauge fixing to area-preserving diffeomorphisms enables a direct mapping to matrices: $X^\mu(\sigma) \mapsto X^\mu_{ij}, \qquad \psi(\sigma) \mapsto \Psi_{ij}, \qquad \{f,g\} \mapsto -iN[F,G], \qquad \int d^2\sigma \mapsto \frac{1}{N} \Tr.$ The resulting action is, up to normalization, the IKKT matrix model action in ten dimensions: $S_{\rm IKKT}^{(E)} = -\frac{1}{4g^2} \Tr \big([X^\mu, X^\nu][X_\mu, X_\nu]\big) - \frac{1}{2g^2} \Tr (\bar{\Psi} \Gamma_\mu [X^\mu, \Psi]),$ with $\bar{\Psi} = \Psi^T C_{10}$ and $1/g^2 = N/2$ (Asano, 2024). In the Lorentzian signature, the sign of the commutator-squared term involving $X^0$ is reversed, and the integration contour is rotated $\exp(-S_E)\to\exp(iS_M)$.

2. Nonperturbative Definition and Emergent Spacetime

The IIB matrix model is a zero-dimensional large-$\kappa$0 reduction of ten-dimensional $\kappa$1 super-Yang–Mills theory. Its partition function is

$\kappa$2

where $\kappa$3 are traceless Hermitian $\kappa$4 matrices and $\kappa$5 are their spinor superpartners. As $\kappa$6, the eigenvalue distributions of $\kappa$7 and the structure of the master field $\kappa$8 encode an emergent spacetime geometry (Klinkhamer, 2022). The band-diagonal structure and spectrum of $\kappa$9 define the effective spacetime dimension and metric, realizing “background independence” as spacetime itself arises dynamically from the matrix degrees of freedom (Nishimura et al., 2011).

Vacua with spontaneously broken $$S_{\rm Schild} = \frac{1}{2\pi} \int d^2\sigma\, e(\sigma) \left[ \frac{1}{4} \{X^\mu, X^\nu\}^2 + i\, \psi^T \Gamma_\mu \{X^\mu, \psi\} \right],$$0 symmetry (down to $$S_{\rm Schild} = \frac{1}{2\pi} \int d^2\sigma\, e(\sigma) \left[ \frac{1}{4} \{X^\mu, X^\nu\}^2 + i\, \psi^T \Gamma_\mu \{X^\mu, \psi\} \right],$$1) correspond to $$S_{\rm Schild} = \frac{1}{2\pi} \int d^2\sigma\, e(\sigma) \left[ \frac{1}{4} \{X^\mu, X^\nu\}^2 + i\, \psi^T \Gamma_\mu \{X^\mu, \psi\} \right],$$2-dimensional extended spacetimes. The constant-volume property ($$S_{\rm Schild} = \frac{1}{2\pi} \int d^2\sigma\, e(\sigma) \left[ \frac{1}{4} \{X^\mu, X^\nu\}^2 + i\, \psi^T \Gamma_\mu \{X^\mu, \psi\} \right],$$3 is $$S_{\rm Schild} = \frac{1}{2\pi} \int d^2\sigma\, e(\sigma) \left[ \frac{1}{4} \{X^\mu, X^\nu\}^2 + i\, \psi^T \Gamma_\mu \{X^\mu, \psi\} \right],$$4-independent for $$S_{\rm Schild} = \frac{1}{2\pi} \int d^2\sigma\, e(\sigma) \left[ \frac{1}{4} \{X^\mu, X^\nu\}^2 + i\, \psi^T \Gamma_\mu \{X^\mu, \psi\} \right],$$5) and minimization of the free energy at $$S_{\rm Schild} = \frac{1}{2\pi} \int d^2\sigma\, e(\sigma) \left[ \frac{1}{4} \{X^\mu, X^\nu\}^2 + i\, \psi^T \Gamma_\mu \{X^\mu, \psi\} \right],$$6 point to a dynamical selection of a $$S_{\rm Schild} = \frac{1}{2\pi} \int d^2\sigma\, e(\sigma) \left[ \frac{1}{4} \{X^\mu, X^\nu\}^2 + i\, \psi^T \Gamma_\mu \{X^\mu, \psi\} \right],$$7-dimensional universe (Nishimura et al., 2011, Anagnostopoulos et al., 2012, Anagnostopoulos et al., 2015).

3. Causality, Path Integral Structure, and Minkowskian Matrix Models

Matrix model regularization preserves quantum-causal structure (“stringy causality”) in the Lorentzian path integral. In the continuum Schild or Nambu–Goto action, careful treatment of the measure and $$S_{\rm Schild} = \frac{1}{2\pi} \int d^2\sigma\, e(\sigma) \left[ \frac{1}{4} \{X^\mu, X^\nu\}^2 + i\, \psi^T \Gamma_\mu \{X^\mu, \psi\} \right],$$8 prescription leads to cancellation of contributions from world-sheets with space-like separation. In the regularized (matrix) picture, only configurations with positive-definite fuzzy-worldsheet area (i.e., all eigenvalues of $$S_{\rm Schild} = \frac{1}{2\pi} \int d^2\sigma\, e(\sigma) \left[ \frac{1}{4} \{X^\mu, X^\nu\}^2 + i\, \psi^T \Gamma_\mu \{X^\mu, \psi\} \right],$$9 positive) contribute. This enforces that the string does not propagate between points at space-like separation (Asano, 2024, Asano, 27 Apr 2026).

The Minkowskian NBI-type IKKT matrix model arises when the auxiliary matrix $\{f,g\}$0 replaces the worldsheet volume element, and integrating over $\{f,g\}$1 ensures causal suppression of space-like contributions: $\{f,g\}$2 (Asano, 27 Apr 2026). This structure extends directly to the large-$\{f,g\}$3 limit, ensuring a well-defined contour and quantum-causal path integral.

4. Regularization Schemes: Euclidean, Lorentzian, and Mass-Deformed Models

The issue of path integral convergence is acute in the Lorentzian case due to the non-compactness of $\{f,g\}$4 and flat directions. Conventional infrared (IR) regularization (e.g., adding a mass term $\{f,g\}$5) breaks Lorentz symmetry and can “classicalize” Lorentz-invariant observables, which is an artifact rather than a genuine quantum effect (Asano et al., 2024). Recent proposals introduce explicit nonperturbative gauge-fixing for Lorentz symmetry using Faddeev–Popov determinants and gauge slices such as $\{f,g\}$6 for all $\{f,g\}$7, which removes the divergent gauge-orbit volume without breaking Lorentz invariance, enabling time evolution to be reconstructed from the eigenvalues of $\{f,g\}$8 (Asano et al., 2024).

Alternative regularizations include mass-deformed (polarized) IKKT models with fuzzy-sphere saddle points. In the regime of large mass parameter, the path integral localizes onto fuzzy $\{f,g\}$9 configurations, giving a lattice-like regularization for D1-branes in a flux-supported cavity, whereas in the massless limit one expects the full IIB supergravity to emerge. Supersymmetric localization techniques allow the exact computation of partition functions over moduli of fuzzy-sphere configurations (Hartnoll et al., 2024).

Covariant-derivative interpretations and finite-$X^\mu(\sigma) \mapsto X^\mu_{ij}, \qquad \psi(\sigma) \mapsto \Psi_{ij}, \qquad \{f,g\} \mapsto -iN[F,G], \qquad \int d^2\sigma \mapsto \frac{1}{N} \Tr.$0 regularizations using Berezin–Toeplitz quantization have also been developed to describe curved backgrounds as finite matrices faithful to the structure of covariant differential operators (Hattori et al., 2024). This provides a bridge to numerical and analytical studies in curved noncommutative geometry.

5. Analysis Tools: Gaussian Expansion and Monte Carlo Methods

The absence of a quadratic term in the IKKT action precludes direct perturbative treatment. The Gaussian Expansion Method (GEM) introduces a trial Gaussian action with variational mass parameters, expanding observables in powers of a control parameter and seeking plateaus as indicators of physical solutions (Nishimura et al., 2011). This approach has demonstrated spontaneous breaking of $X^\mu(\sigma) \mapsto X^\mu_{ij}, \qquad \psi(\sigma) \mapsto \Psi_{ij}, \qquad \{f,g\} \mapsto -iN[F,G], \qquad \int d^2\sigma \mapsto \frac{1}{N} \Tr.$1 down to $X^\mu(\sigma) \mapsto X^\mu_{ij}, \qquad \psi(\sigma) \mapsto \Psi_{ij}, \qquad \{f,g\} \mapsto -iN[F,G], \qquad \int d^2\sigma \mapsto \frac{1}{N} \Tr.$2, constancy of the compactification scale, and stability of vacuum energies, with extensive cross-validation from Monte Carlo simulations.

Monte Carlo methods face a severe “complex-action problem” due to an oscillatory Pfaffian phase after integrating out fermions. The factorization method addresses this by jointly sampling observables correlated with the phase and reconstructing true expectation values via constrained partition functions and saddle-point equations. This enables robust calculation of spacetime extents and the pattern of dynamical compactification (Anagnostopoulos et al., 2012, Anagnostopoulos et al., 2015).

To treat the Lorentzian IIB model, the Complex Langevin Method (CLM) samples the complexified matrix configuration space, allowing stochastic evolution with carefully-chosen contour stabilization to evade the sign problem (Hirasawa et al., 2023, Hatakeyama et al., 2021). Such simulations have confirmed the emergence of expanding universes and even dynamically realize Euclidean–Lorentzian signature change (Hatakeyama et al., 2021).

6. Matrix Regularization Beyond the IKKT Model and Dualities

Matrix model regularization has been extended to capture additional stringy effects. Three-algebra (Nambu-bracket) generalizations of the IIB matrix model, introducing triple brackets and auxiliary fields, preserve the full supersymmetry algebra and permit new phases where the action reduces to “one-commutator” models or more exotic potentials. In particular, certain 3-algebra phases can simplify the dynamics or regulate previously uncontrolled IR directions while maintaining maximal supersymmetry (Sato, 2013).

Matrix regularization also provides the framework for nonperturbative realization of dualities. Lightcone-wrapped supermembrane models on $X^\mu(\sigma) \mapsto X^\mu_{ij}, \qquad \psi(\sigma) \mapsto \Psi_{ij}, \qquad \{f,g\} \mapsto -iN[F,G], \qquad \int d^2\sigma \mapsto \frac{1}{N} \Tr.$3, when matrix-regularized, can describe (p,q)-strings and encode both S- and T-duality at the level of matrix representations and boundary conditions. Matrix double-dimensional reduction yields effective matrix-string actions with correct tension dependence, and curved background couplings follow from the matrix analogues of membrane background fields (0708.3484).

7. Physical Interpretation and Implications for String Theory

Matrix regularization of type IIB superstrings via the IKKT model and its extensions provides a manifestly gauge-invariant, UV-finite, nonperturbative formulation of the entire sum over worldsheets and, by duality, all D-brane sectors. Spacetime geometry, dimensionality, and even its signature emerge dynamically from the eigenvalues and structure of large matrices, rather than being put in by hand. The dynamically selected $X^\mu(\sigma) \mapsto X^\mu_{ij}, \qquad \psi(\sigma) \mapsto \Psi_{ij}, \qquad \{f,g\} \mapsto -iN[F,G], \qquad \int d^2\sigma \mapsto \frac{1}{N} \Tr.$4-dimensional extended spacetime, constant-volume relations for compactified directions, and the enforcement of quantum causality at the nonperturbative level are central outputs of this approach (Nishimura et al., 2011, Anagnostopoulos et al., 2012, Asano, 2024, Asano, 27 Apr 2026).

Current research directions focus on achieving complete control over numerical and analytical regularization in both Euclidean and Lorentzian signatures (Hattori et al., 2024, Asano et al., 2024), leveraging supersymmetric localization (Hartnoll et al., 2024), and formulating extensions that systematically incorporate string/brane dualities and the full landscape of nonperturbative string vacua (0708.3484, Sato, 2013).

References: (Asano, 2024, Nishimura et al., 2011, Hartnoll et al., 2024, Klinkhamer, 2022, Hirasawa et al., 2023, Anagnostopoulos et al., 2012, Hattori et al., 2024, Asano et al., 2024, Anagnostopoulos et al., 2015, Asano, 27 Apr 2026, Hatakeyama et al., 2021, 0708.3484, Sato, 2013).