Nonperturbative completion of two-dimensional quantum gravity: string-equation integration constants, Stokes data, and the matrix-model-continuum correspondence

Witten's 1990 paper on the topological phase of two-dimensional gravity left open five problems connecting the continuum intersection-theoretic formulation with the matrix-model formulation. The perturbative components — all-genus computability of correlators, the undetermined genus-one constant, constructive algorithms for the intersection numbers, and the classification and matter-coupling identification of the multicritical critical points — were resolved within a few years by the Witten conjecture, Kontsevich's proof, the Virasoro constraints, and the minimal-model correspondence, with effective all-genus computation later supplied by Mirzakhani's Weil-Petersson recursion. The nonperturbative components survive as an active research frontier: the string equations of the double-scaled multicritical models determine only formal perturbation series, while their general solutions are resurgent transseries whose instanton sectors, Stokes data, and nonperturbative integration constants are only partially computed, whose physical-solution selection rule is not derived from first principles, and whose precise correspondence with the continuum topological formulation — including exponentially enhanced sectors that until recently had no known matrix-model interpretation — is not established.

Background

Two-dimensional (2D) quantum gravity — the path integral over all metrics on a two-dimensional surface, equivalently the statistical theory of random surfaces — is the simplest arena in which the nonperturbative definition of a gravitational or string theory can be posed. Around 1990 two a priori unrelated formulations existed. The first is the matrix-model formulation: the Feynman-diagram expansion of an integral over an $N\times N$ hermitian matrix enumerates discretized random surfaces (triangulations), with the topology (genus) of the surface appearing as the expansion parameter $1/N^2$. Sending $N\to\infty$ while tuning the matrix potential to a critical point — the double scaling limit — produces a continuum theory of 2D gravity, and each tuning, called a multicritical point, yields gravity coupled to a different conformal matter theory. The second is the continuum formulation. In the 'topological phase' studied by Witten (1990), the observables are intersection numbers on moduli spaces: let $\overline{M}_{g,n}$ be the Deligne-Mumford compactified moduli space of stable genus-$g$ Riemann surfaces with $n$ marked points; each marked point $i$ carries a natural complex line bundle $L_i$ (the cotangent line to the surface at that point) whose first Chern class (the characteristic cohomology class of that line bundle) is $\psi_i$, and the gravitational descendant observable $\tau_d$ has correlator $\langle\tau_{d_1}\cdots\tau_{d_n}\rangle_g=\int_{\overline{M}_{g,n}}\psi_1^{d_1}\cdots\psi_n^{d_n}$. Witten (1990) showed that genus-zero intersection numbers reproduce the genus-zero matrix-model amplitudes, computed genus one up to a single undetermined constant, and posed the problem of establishing the full correspondence. Witten (1991) then conjectured that the generating function $F(t_0,t_1,\dots)$ of all intersection numbers is the unique solution of two conditions: $U=\partial^2 F/\partial t_0^2$ obeys the KdV hierarchy (an infinite tower of nonlinear evolution equations from soliton theory), and the string equation, an additional algebraic constraint reflecting the puncture operator; for pure gravity the string equation reduces to the Painleve I ordinary differential equation (the first of Painlevé's list of second-order nonlinear ODEs). Kontsevich (1992) proved the conjecture by an explicit combinatorial construction (the matrix Airy function, a sum over ribbon graphs, the fat Feynman diagrams of the matrix model). Equivalent reformulations include the Virasoro constraints of Dijkgraaf-Verlinde-Verlinde (1991) — the partition function is annihilated by a half-branch of the Virasoro algebra, the symmetry algebra of two-dimensional conformal field theory — and Mirzakhani's (2007) recursion for Weil-Petersson volumes (volumes of moduli space for its natural Weil-Petersson symplectic form), which gives effective all-genus computation. In parallel, the 1990 matrix-model literature (Douglas and Shenker; Brezin and Kazakov) established that the universality classes of the double-scaled one-matrix model are precisely its multicritical points, completing the perturbative classification: as documented in the TASI 1992 lectures of Ginsparg and Moore, the m-th multicritical point describes the (2, 2m-1) minimal model coupled to gravity — a member of the discrete series of two-dimensional conformal field theories with central charge $c<1$ that is non-unitary in general, pure gravity being the (2,3) member — and its string susceptibility exponent $\gamma_{\mathrm{str}}=-1/m$ (the critical exponent governing the vanishing of the continuum free energy at the critical point) happens to coincide with that of the m-th member of the unitary discrete series. These results settled the perturbative part of the 1990 program. What they do not settle is the nonperturbative part, which is the subject of this problem. The matrix integral is a nonperturbative definition of 2D gravity (it exists at finite $N$), but the continuum equations it yields in the double-scaling limit — the string equation and the KdV hierarchy — determine only a formal power series in the string coupling $g$. Their general solutions are transseries: asymptotic series supplemented by exponentially small instanton sectors, schematically $\sum_{n\ge0} e^{-nA/g}\Phi_n(g)$ for an instanton action $A$. The coefficients and Stokes constants of these sectors — the nonperturbative integration constants of the string equation, where the Stokes constants fix the jumps in the transseries coefficients as Stokes lines are crossed — are not fixed by the perturbative equations: different choices define different nonperturbative theories with the same perturbation expansion, an ambiguity that must be removed by nonperturbative input. Resurgence is the modern framework organizing this structure: it relates the factorial large-order growth of the perturbative coefficients to the instanton sectors and encodes how sectors mix across Stokes lines in the Borel plane (the complex plane in which divergent series are resummed). In the matrix model, the instanton sectors are attributed to eigenvalue tunneling — individual matrix eigenvalues escaping from the cut of the spectral curve (the algebraic curve relating the eigenvalue density to the matrix potential) to distant turning points, including on the non-physical sheet of that curve. A complete nonperturbative definition of 2D gravity in this framework must compute the full transseries and Stokes data of the multicritical string equations, identify the boundary condition or selection rule that picks the physical solution, and derive the correspondence of this nonperturbative data with the continuum intersection-theoretic formulation — tasks that remain only partially accomplished, mostly for the simplest (pure gravity, Painleve I) case.

Problem Statement

In Witten's 1990 formulation of the topological phase of two-dimensional gravity, the following issues were left open: (i) the topological (intersection-theoretic) machinery was computationally weaker than the matrix models beyond genus zero and one; (ii) a genus-one constant $p$ (with $\langle 1\rangle_1=p\ln\epsilon$ for the scaling parameter $\epsilon$) was left undetermined by the topological recursion; (iii) no continuum derivation existed of the higher-genus matrix-model amplitudes or of the nonperturbative integration constants of the string equations; (iv) no explicit constructive algorithms existed for the intersection representatives on $\overline{M}_{g,s}$ at arbitrary genus; (v) the classification of critical points, their mapping to matter couplings, and the precise map of nonperturbative data between the matrix-model and continuum descriptions were incomplete. Issues (i), (ii), (iv) and the perturbative parts of (iii) and (v) are settled by the Witten conjecture and its proof by Kontsevich, the Virasoro constraints, the Weil-Petersson recursion, and the multicritical classification. The surviving open problem is the nonperturbative completion of two-dimensional quantum gravity in this framework: determine the nonperturbative content of the double-scaled multicritical string equations — the complete resurgent transseries solutions, their Stokes data, and their nonperturbative integration constants — together with the boundary condition or selection rule that removes the nonperturbative ambiguity and singles out the physical solution; and establish the precise correspondence of this nonperturbative data between the matrix-model definition (eigenvalue tunneling, including exponentially enhanced and non-physical-sheet sectors, and finite-$N$ effects) and the continuum topological-gravity formulation, beyond the lowest multicritical points for which partial results are known. The scope is the multicritical family: pure gravity — whose string equation is the Painleve I equation — together with at least one higher multicritical point, where any multicritical point strictly above pure gravity qualifies as the higher point. The question is what the theory is nonperturbatively: which transseries solution is realized, and why — beyond the resummation of its perturbation series. The nonperturbative references are partial inputs: the family-level construction of 2602.15101 supplies the closed-form transseries solutions but explicitly defers the complete non-linear Stokes data; 2108.11409 computes nonperturbative instanton actions and Stokes data for the multicritical and minimal-string cases and shows the transseries solutions to be resonant, but leaves the nonperturbative integration constants and the physical-solution selection open; and 2210.13479 derives Stokes data for the cubic and quartic matrix models and their Painleve I double-scaling limit, while 2601.03351 extracts continuum nonperturbative transseries data with the (2,3) minimal string as its case study. No cited work fixes the nonperturbative integration constants by a selection rule derived from a nonperturbative definition, and none establishes the matrix-model-to-continuum nonperturbative correspondence for the multicritical family.

References

1. Edward Witten, "On the structure of the topological phase of two-dimensional gravity", Nuclear Physics B 340 (1990) 281-332, DOI 10.1016/0550-3213(90)90449-N, https://doi.org/10.1016/0550-3213(90)90449-N
2. Edward Witten, "Two-dimensional gravity and intersection theory on moduli space", Surveys in Differential Geometry 1 (1991) 243-310, DOI 10.4310/SDG.1990.v1.n1.a5, https://doi.org/10.4310/SDG.1990.v1.n1.a5
3. Maxim Kontsevich, "Intersection theory on the moduli space of curves and the matrix Airy function", Communications in Mathematical Physics 147 (1992) 1-23, DOI 10.1007/BF02099526, https://doi.org/10.1007/BF02099526
4. Michael R. Douglas, Stephen H. Shenker, "Strings in less than one dimension", Nuclear Physics B 335 (1990) 635-654, DOI 10.1016/0550-3213(90)90522-F, https://doi.org/10.1016/0550-3213(90)90522-F
5. E. Brézin, V. A. Kazakov, "Exactly solvable field theories of closed strings", Physics Letters B 236 (1990) 144-150, DOI 10.1016/0370-2693(90)90818-Q, https://doi.org/10.1016/0370-2693(90)90818-Q
6. Robbert Dijkgraaf, Herman L. Verlinde, Erik P. Verlinde, "Loop equations and Virasoro constraints in non-perturbative two-dimensional quantum gravity", Nuclear Physics B 348 (1991) 435-456, DOI 10.1016/0550-3213(91)90199-8, https://doi.org/10.1016/0550-3213(91)90199-8
7. Maryam Mirzakhani, "Weil-Petersson volumes and intersection theory on the moduli space of curves", Journal of the American Mathematical Society 20 (2007) 1-23, DOI 10.1090/S0894-0347-06-00526-1, https://doi.org/10.1090/S0894-0347-06-00526-1
8. Paul Ginsparg, Gregory Moore, "Lectures on 2D gravity and 2D string theory (TASI 1992)", arXiv:hep-th/9304011, https://arxiv.org/abs/hep-th/9304011
9. Sara Pasquetti, Ricardo Schiappa, "Borel and Stokes nonperturbative phenomena in topological string theory and c=1 matrix models", Annales Henri Poincare 11 (2010) 351-431, arXiv:0907.4082, DOI 10.1007/s00023-010-0044-5, https://arxiv.org/abs/0907.4082
10. Marcos Marino, Ricardo Schiappa, Maximilian Schwick, "New instantons for matrix models", arXiv:2210.13479 (2022; v3 August 2026), DOI 10.48550/arXiv.2210.13479, https://arxiv.org/abs/2210.13479
11. Jasper Kager, João Rodrigues, Ricardo Schiappa, Maximilian Schwick, Noam Tamarin, "Exact Solutions to Matrix Models and String Theories: The Local Construction", arXiv:2602.15101 (2026), https://arxiv.org/abs/2602.15101
12. Clifford V. Johnson, João Rodrigues, "Non-perturbative data for Weil-Petersson volumes and intersection numbers using ordinary differential equations", arXiv:2601.03351 (2026; v2 February 2026), https://arxiv.org/abs/2601.03351
13. Paolo Gregori, Ricardo Schiappa, "From Minimal Strings towards Jackiw-Teitelboch Gravity: On their Resurgence, Resonance, and Black Holes", Class. Quantum Grav. 41 (2024) 115001, arXiv:2108.11409, https://arxiv.org/abs/2108.11409