Matrix-Model Lessons Beyond Two Dimensions

Matrix models teach four durable lessons: combinatorial amplitudes can organize topology, loop equations can determine continuum observables, critical tuning can remove a discretization scale, and the resulting asymptotic expansion still needs nonperturbative completion. In tensor models, genus is replaced by Gurau degree, but dominant graphs and geometric universality are different.

Required background. Tensor Models and Random Geometry supplies the higher-rank expansion. From Genus Counting to a Holographic String Regime fixes what genus counting means.

Helpful background. Random Matrices, Spectral Statistics, and Ensemble Questions and Nonperturbative Exponential Effects and Finite-N Sectors supply spectral and completion cautions.

Topological versus degree expansions

For an $N\times N$ matrix model, a ribbon graph of genus $h$ carries

$$\mathcal A_h\propto N^{2-2h}.$$

Its faces triangulate an orientable surface, so the power of $N$ directly tracks topology. At a critical coupling $g_c$, the number of faces diverges and a double-scaling variable keeps all genera:

$$\kappa=N^\alpha(g_c-g)$$

with model-dependent $\alpha$ Brézin and Kazakov 1990.

For rank-$d$ tensors,

$$\mathcal A_{\omega}\propto N^{d-\frac{2}{(d-1)!}\omega},$$

and $\omega$ is a sum of jacket genera, not the topology of one two-dimensional surface. Degree-zero melons dominate and can all have the same leading power while representing branched complexes.

Application: compare critical variables

The planar matrix resolvent

$$W(z)=\frac1N\left\langle\operatorname{Tr}\frac1{z-M}\right\rangle$$

obeys a loop equation whose branch cut defines a spectral curve. Its critical endpoint controls continuum boundary-length amplitudes. The quartic melonic tensor model instead obeys $G=1+qG^2$, with $q_c=1/4$. Both have square-root singularities, yet $W(z)$ resolves loop boundaries on random surfaces whereas $G$ counts melonic insertions.

Thus the transferable mechanism is critical tuning of generating functions. The nontransferable conclusion is that matching a critical exponent produces the same geometry.

Nonperturbative and observable tests

Matrix double scaling can yield differential string equations with multiple nonperturbative solutions sharing one asymptotic series. Tensor double scaling likewise requires a contour or exact integral and a positivity domain. Compare loop observables, graph distance, topology distribution, and matter critical exponents—not only free-energy singularities.

Adversarial analogy test

Match a matrix and tensor model so their two-point susceptibilities have the same exponent. Then compare Hausdorff dimension and boundary observables. If these disagree, the shared exponent is insufficient for universality. Also alter the integration contour: identical perturbation theory with different exponentially small sectors demonstrates completion ambiguity.

Scope

Matrix methods provide powerful design principles, not a shortcut from rank to spacetime dimension. The cross-program geometric diagnostic is developed on Spectral Dimension and Dimensional Flow.

The chapter overview contains the structure diagram and validity and failure diagram. They are embedded there once so that their shared chapter-level context is not repeated on every article.

For the chapter-wide comparison of assumptions, counterevidence, falsifiers, and claim ceilings, see the claim-domain table.

References

• Brézin, E., and V. A. Kazakov. “Exactly Solvable Field Theories of Closed Strings.” Physics Letters B 236 (1990): 144–150. DOI.
• Douglas, M. R., and S. H. Shenker. “Strings in Less Than One Dimension.” Nuclear Physics B 335 (1990): 635–654. DOI.
• Gurau, R. “The 1/N Expansion of Colored Tensor Models.” Annales Henri Poincaré 12 (2011): 829–847. DOI.