Nonperturbative completion of the one-dimensional discretized string (c=1 matrix model / two-dimensional string theory): the non-Borel-summable genus expansion and the barrier boundary-condition ambiguity

In his 1990 solution of the one-dimensional discretized string (the matrix quantum mechanics describing two-dimensional string theory), Giorgio Parisi resummed the genus expansion of the vacuum energy in closed form via a WKB analysis of a local inverted-oscillator Schrödinger problem, and left four issues open: the local Hamiltonian is not essentially self-adjoint, so additive constants depend on the boundary condition at the far side of the potential barrier; the genus expansion is factorially divergent and not Borel summable, with no principled prescription for the resulting nonperturbative ambiguity (fermion tunneling effects of order exp(-const/g_s)); spatially dependent correlation functions and the consistency of the summed continuum theory were unestablished; and non-quadratic (m != 2) critical points were unanalyzed. The correlation-function problem is settled by 1991-92, and the boundary-condition sensitivity of finite (non-exponentially-small) observables is settled (genus-by-genus coefficients are universal; only exponentially small terms depend on the choice); the substantial 2019-2026 ZZ-instanton program proposes a candidate completion for the quadratic case whose consistency remains unproven. The surviving open core is the nonperturbative completion itself at the quadratic (m=2) critical point: constructing a consistent nonperturbative definition of the two-dimensional string that fixes the fermion state on the far side of the barrier by a physical consistency principle, supplies the nonperturbative (tunneling/instanton) contributions resolving the Borel ambiguity, and either establishes unitarity and causality of the emergent spacetime or a rigorous obstruction to any such consistent completion.

Background

FROM DISCRETIZED RANDOM SURFACES TO THE c=1 MATRIX MODEL. A 'non-critical string' is a theory of two-dimensional worldsheets (Euclidean quantum gravity in two dimensions) coupled to matter. A nonperturbative definition is obtained by discretizing: worldsheets are built by gluing triangles on a dynamical lattice (dynamical triangulation), and the sum over geometries becomes the perturbative expansion of a matrix integral or matrix quantum mechanics (MQM), the quantum mechanics of a single $N\times N$ Hermitian matrix $M(t)$. In the large-$N$ expansion the power of $N$ attached to a Feynman diagram counts its genus $g$ (number of handles), so the expansion in $1/N$ reorganizes the discretized sum over surfaces as a sum over topologies - the genus expansion of string theory, with $1/N$ playing the role of the string coupling. This dictionary originates in the planar-diagram analysis of Brezin-Itzykson-Parisi-Zuber. For a string whose worldsheet matter content is a single free boson (central charge $c=1$; the central charge measures the number of degrees of freedom of a two-dimensional conformal field theory, one unit per free boson) the target space is one-dimensional, the relevant model is MQM, and the continuum limit is two-dimensional string theory (one spatial dimension plus a time-like direction carried by the Liouville field, the worldsheet scale factor that remains dynamical after gauge-fixing the worldsheet metric) whose only propagating mode is a single massless scalar, historically called the 'tachyon', coupled to a linearly varying dilaton field (the linear-dilaton background, in which the string coupling grows linearly along the Liouville direction).

THE DOUBLE SCALING LIMIT AND PARISI'S CLOSED FORM. The continuum string is reached by a double scaling limit: $N\to\infty$ while the coupling $\lambda$ is tuned to a critical value $\lambda_c$ at which the eigenvalue 'Fermi sea' (the filled energy levels of the free fermions obtained from the matrix eigenvalues) of the matrix Hamiltonian reaches a stationary point $x_0$ of the eigenvalue potential, keeping the effective string coupling finite so that all genera contribute. Parisi (Phys. Lett. B 238 (1990) 209) analyzed this limit by a semiclassical (WKB / Thomas-Fermi) evaluation (the leading small-Planck-constant approximation, applied here to the eigenvalue density of the many-fermion system) of the eigenvalue density near $x_0$. Writing $\tau=\lambda_c-\lambda$, the singular part of the ground-state energy per eigenvalue obeys the scaling form $E(N,\lambda)/N=\tau^{1+\omega}f(\tau^{-2}N^{-2})$ with $\omega=1/(1/m+1/2)$, where $m$ labels the order of the stationary point ($m=2$: quadratic maximum, the $c=1$ case). For $m=2$, introducing a variable $\tilde\varepsilon$ through $-\tilde\varepsilon\ln\tilde\varepsilon=\tau$ and $z=\tilde\varepsilon/N$, the summed most-singular contributions to the vacuum energy and to the singlet correlators $\langle[\mathrm{tr}\,M^2/N]^k\rangle$ (singlet meaning invariant under the $U(N)$ symmetry that rotates the matrix) are generated by a single scaling function,

$$\nu(z)=\operatorname{Im}\log\Gamma\big(\tfrac12+i/z\big)+c_0+c_1 z,$$

obtained by solving the local Schrödinger problem of an inverted harmonic oscillator (the solutions are parabolic cylinder functions). Three companion papers (Gross-Miljkovic; Brezin-Kazakov-Zamolodchikov; Ginsparg-Zinn-Justin) derived equivalent forms of the double-scaled solution. Expanding $\nu(z)$ at small $z$ reproduces the genus-by-genus expansion with Bernoulli-number coefficients that grow factorially, $A_k\sim k!/(2\pi)^k$; equivalently the genus-$g$ coefficients grow like $(2g)!$. A factorially divergent series is an asymptotic series; when its Borel transform (the standard resummation tool, a Laplace-type transform that divides the $k$-th coefficient by $k!$) has singularities on the integration contour, the series is not Borel summable and the resummed function is ambiguous by terms controlled by the nearest singularity. Here the ambiguity is of 'stringy' instanton strength, $e^{-\mathrm{const}/g_s}$ rather than the field-theory $e^{-\mathrm{const}/g_s^2}$, as emphasized by Shenker, and is physically attributed to tunneling of the eigenvalue fermions through the inverted-oscillator barrier.

THE OPEN ISSUES. The paper left four loose ends. (i) The local inverted-oscillator Hamiltonian on the half-line $x\ge x_0$ is not essentially self-adjoint (essential self-adjointness means a symmetric operator admits a unique self-adjoint extension, so no additional boundary condition is needed for unitary time evolution); the additive constants $c_0,c_1$ in $\nu$ depend on the boundary condition at the far side, or on the shape of the potential for $x>x_0$, and it was not resolved how much of this dependence leaks into finite (non-exponentially-small) observables. (ii) Because the genus expansion is not Borel summable, a principled prescription (a concrete choice of principal part, plus tunneling contributions) is needed to define the theory nonperturbatively; none was given. (iii) Only singlet, zero-momentum observables were computed; spatially dependent correlators and the consistency of the summed continuum theory were left open. (iv) The non-quadratic criticalities $m\neq2$, with different analytic structure and possibly different instanton effects, were left unanalyzed. Later literature resolved (iii) and the finite-part aspect of (i), sharpened (i)+(ii) into the question of the nonperturbative consistency of two-dimensional string theory, and left (iv) essentially untouched. A modern formulation of the central object: the double-scaled $c=1$ matrix model is a system of free fermions in the inverted harmonic oscillator potential; perturbative string amplitudes are computed from the fermion spectral density (equivalently via the collective-field description, the hydrodynamic reformulation of the fermions as a single bosonic field), the dictionary from matrix operators to string tachyon vertex operators involves a momentum-dependent 'leg pole' factor (a nonlocal normalization relating matrix-model density modes to continuum tachyon operators), and the nonperturbative effects are fermion tunneling - in worldsheet language, ZZ instantons (after Zamolodchikov and Zamolodchikov): D-instantons associated with the degenerate representations of Liouville theory.

Problem Statement

PROBLEM. Construct a principled, consistent nonperturbative completion of the double-scaled one-dimensional discretized string at the quadratic critical point - equivalently the $c=1$ matrix model (free fermions in an inverted harmonic oscillator potential) or two-dimensional string theory (Liouville theory coupled to a free boson) - and validate it. Concretely:

1. Fix the barrier/boundary-condition ambiguity. The local Schrödinger problem defining the scaling function $\nu(z)$ is not essentially self-adjoint; equivalently, the double-scaled theory does not specify the eigenvalue potential or the fermion state on the far side ($x>x_0$) of the potential barrier. Determine, by a physical consistency principle rather than convention, which completion is selected (e.g. a wall-modified potential ('type I'), a two-sided filling of the Fermi sea ('type II'), or a no-incoming-flux scattering state), and record which observables are insensitive to the choice (the finite, genus-by-genus parts are universal - an established input to be cited rather than re-derived) and which depend on it (the exponentially small terms of order $e^{-\mathrm{const}/g_s}$).

2. Resolve the non-Borel-summability. The genus expansion has factorially growing coefficients ($(2g)!$ growth). Specify a principled nonperturbative prescription that removes the ambiguity of the perturbative series and defines the vacuum energy and scattering amplitudes nonperturbatively. The ambiguity is of stringy strength $e^{-\mathrm{const}/g_s}$ (the scale written $\exp(-\pi/z)$ in the source paper's variables).

3. Establish the consistency of the summed continuum theory. Either show that the completed theory defines a physically acceptable quantum theory of the emergent two-dimensional spacetime - a unitary S-matrix satisfying the causality constraints of the emergent spacetime, including the bulk $2\to1$ tachyon scattering test generated by the nonlocal leg-pole transformation, which motivated a 1995 no-go argument against all then-existing completions - or establish a rigorous obstruction (no-go theorem) to any such consistent completion. A rigorous no-go theorem is a negative resolution of the nonperturbative-completion problem, in which case the constructions of items 1-2 are moot; a completed theory, by contrast, must satisfy the unitarity and causality checks.

The common objective uniting these deliverables is to settle whether the one-dimensional discretized string admits a single controlled nonperturbative definition and, if it does, to supply it with validated observable predictions. The following parts of the source question are settled and are not part of what remains: spatially dependent (nonzero momentum and winding) correlation functions and their matching to the continuum Liouville description (computed 1991-92), and the universality of the finite parts of observables under changes of the far-side potential and boundary condition (established: only exponentially small terms are non-universal). The non-quadratic ($m\neq2$) critical points, whose local Schrödinger problems have different analytic structure, are a separate generalization of the present problem and lie outside its acceptance boundary.

References

1. Giorgio Parisi, On the one dimensional discretized string, Phys. Lett. B 238 (1990) 209-212, DOI 10.1016/0370-2693(90)91722-N, https://doi.org/10.1016/0370-2693(90)91722-N
2. David J. Gross and Nikola Miljković, A nonperturbative solution of D=1 string theory, Phys. Lett. B 238 (1990) 217-223, DOI 10.1016/0370-2693(90)91724-P, https://doi.org/10.1016/0370-2693(90)91724-P
3. E. Brézin, V. A. Kazakov and A. B. Zamolodchikov, Scaling violation in a field theory of closed strings in one physical dimension, Nucl. Phys. B 338 (1990) 673-688, DOI 10.1016/0550-3213(90)90647-V, https://doi.org/10.1016/0550-3213(90)90647-V
4. Paul H. Ginsparg and Jean Zinn-Justin, 2D gravity+1D matter, Phys. Lett. B 240 (1990) 333-340, DOI 10.1016/0370-2693(90)91108-N, https://doi.org/10.1016/0370-2693(90)91108-N
5. E. Brézin, C. Itzykson, G. Parisi and J. B. Zuber, Planar diagrams, Commun. Math. Phys. 59 (1978) 35-51, DOI 10.1007/BF01614153, https://doi.org/10.1007/BF01614153
6. Igor R. Klebanov, String theory in two dimensions, lectures at the ICTP Spring School on Particle Physics, Trieste, April 1991 (preprint PUPT-1271), arXiv:hep-th/9108019, https://arxiv.org/abs/hep-th/9108019
7. Paul Ginsparg and Gregory Moore, Lectures on 2D gravity and 2D string theory, lectures at the 1992 TASI summer school, arXiv:hep-th/9304011 (1993), https://arxiv.org/abs/hep-th/9304011
8. Gregory W. Moore, M. Ronen Plesser and Sanjaye Ramgoolam, Exact S-matrix for two-dimensional string theory, Nucl. Phys. B 377 (1992) 143-190, arXiv:hep-th/9111035, DOI 10.1016/0550-3213(92)90020-C, https://arxiv.org/abs/hep-th/9111035
9. David J. Gross, Igor R. Klebanov and M. J. Newman, The two-point correlation function of the one-dimensional matrix model, Nucl. Phys. B 350 (1991) 621-634, DOI 10.1016/0550-3213(91)90158-T, https://doi.org/10.1016/0550-3213(91)90158-T
10. Norisuke Sakai and Yoshiaki Tanii, Correlation functions of c=1 matter coupled to two-dimensional gravity, Prog. Theor. Phys. 86 (1991) 547-553, DOI 10.1143/PTP.86.547, https://doi.org/10.1143/PTP.86.547
11. Stephen H. Shenker, The strength of nonperturbative effects in string theory, in Random Surfaces and Quantum Gravity (Proceedings, NATO Advanced Research Workshop, Cargese, France, May 27 - June 2, 1990), eds. O. Alvarez, E. Marinari and P. Windey, NATO ASI Series B vol. 262, Plenum, New York (1991) 191-200, https://inspirehep.net/literature/298523 (proceedings contribution; no DOI or arXiv identifier)
12. Paul H. Ginsparg and Jean Zinn-Justin, Large order behavior of non-perturbative gravity, Phys. Lett. B 255 (1991) 189-196, DOI 10.1016/0370-2693(91)90234-H, https://doi.org/10.1016/0370-2693(91)90234-H
13. Joseph Polchinski, On the nonperturbative consistency of d=2 string theory, Phys. Rev. Lett. 74 (1995) 638-641, arXiv:hep-th/9409168, DOI 10.1103/PhysRevLett.74.638, https://arxiv.org/abs/hep-th/9409168
14. Martin O'Loughlin, On consistent boundary conditions for c=1 string theory, Nucl. Phys. B 449 (1995) 80-90, arXiv:hep-th/9501039, DOI 10.1016/0550-3213(95)00244-M, https://arxiv.org/abs/hep-th/9501039
15. Steven S. Gubser and Igor R. Klebanov, A modified c=1 matrix model with new critical behavior, Phys. Lett. B 340 (1994) 35-42, arXiv:hep-th/9407014, DOI 10.1016/0370-2693(94)91294-7, https://arxiv.org/abs/hep-th/9407014
16. V. A. Kazakov, External matrix field problem and new multicriticities in (two)-dimensional random surfaces, Nucl. Phys. B 354 (1991) 614-624, DOI 10.1016/0550-3213(91)90368-8, https://doi.org/10.1016/0550-3213(91)90368-8, https://inspirehep.net/literature/298433
17. Bruno Balthazar, Victor A. Rodriguez and Xi Yin, ZZ instantons and the non-perturbative dual of c=1 string theory, JHEP 05 (2023) 048, arXiv:1907.07688, DOI 10.1007/JHEP05(2023)048, https://arxiv.org/abs/1907.07688
18. Bruno Balthazar, Victor A. Rodriguez and Xi Yin, Multi-instanton calculus in c=1 string theory, JHEP 05 (2023) 050, arXiv:1912.07170, DOI 10.1007/JHEP05(2023)050, https://arxiv.org/abs/1912.07170
19. Ashoke Sen, Fixing an ambiguity in two dimensional string theory using string field theory, JHEP 03 (2020) 005, arXiv:1908.02782, DOI 10.1007/JHEP03(2020)005, https://arxiv.org/abs/1908.02782
20. Ashoke Sen, Divergent to complex amplitudes in two dimensional string theory (published as 'Divergent ⇒ complex amplitudes in two dimensional string theory'), JHEP 02 (2021) 086, arXiv:2003.12076, DOI 10.1007/JHEP02(2021)086, https://arxiv.org/abs/2003.12076
21. Rishabh Kaushik, c=1, R=1 and N>>1: ZZ instantons in 2D string theory and matrix integrals, JHEP 08 (2025) 177, arXiv:2501.14023, DOI 10.1007/JHEP08(2025)177, https://arxiv.org/abs/2501.14023
22. Dionysios Anninos and Beatrix Mühlmann, Notes on matrix models (matrix musings), J. Stat. Mech. (2020) 083109, arXiv:2004.01171, DOI 10.1088/1742-5468/aba499, https://arxiv.org/abs/2004.01171
23. Scott Collier, Lorenz Eberhardt and Victor A. Rodriguez, c=1 strings as a matrix integral, arXiv:2604.06301 (2026), https://arxiv.org/abs/2604.06301
24. Sumit R. Das, Shaun D. Hampton, Sinong Liu and Gautam Mandal, Fate of 'space-like singularities' in c=1 matrix model, arXiv:2606.31925 (2026), https://arxiv.org/abs/2606.31925