Kimura and Navand (arXiv:2408.06888; Lett. Math. Phys. 116 (2026)) evaluated Fredholm determinants of kernels K(ξ,ζ) = ∫{ℝ₊} φ_ξ(x)φ_ζ(x)dx whose defining wave functions φ_ξ solve a Schrödinger-type equation (∂²_x − v(x;ξ))φ_ξ = ξφ_ξ, expressing the determinant for the half-line I = [τ,∞) through an auxiliary function q that satisfies a closed second-order integro-differential equation — a continuous deformation of the Tracy–Widom distribution, recovered at the Airy point. Their theorem, however, restricts the wave function to the composed form φ_ξ(x) = φ(u(x)+γξ) (Condition 1.4) and to a single interval, and the isomonodromic interpretation of the underlying infinite Lax matrices remains formal. The authors explicitly list seven yet-uncharted directions beyond the ones they undertook, which could be interesting to explore and all of which remain open: extend the functional formula to wave functions not of composed form; extend to defining differential operators beyond the second-order self-adjoint case; generalize the formula to arbitrary unions of intervals; compute the large-gap asymptotics, which would notably allow a check of compatibility with the Forrester–Chen–Eriksen–Tracy results; determine whether the auxiliary system closes to a finite-size Schlesinger system in χ{0,0} and χ_{0,1}; prove the Schlesinger equations componentwise and derive the Jimbo–Miwa–Ueno trace properties; and, as a more technical comment, examine whether Condition 3.6 can be lifted.

Background

A Fredholm determinant F(I) = det(ℐ − K₀Π_I) of a trace-class integral operator K₀ with kernel K(ξ,ζ), restricted to an interval I by the projector Π_I, gives the probability that a determinantal point process (a random point configuration whose correlations are given by determinants of K) has no point in I — the so-called gap probability. The classical example is the Tracy–Widom distribution for the largest eigenvalue of large random Hermitian matrices: there K is the Airy kernel, K(ξ,ζ) = ∫_{ℝ₊} Ai(ξ+x)Ai(ζ+x)dx, and Tracy and Widom (1994) expressed F([τ,∞)) as a functional exp(−∫_τ^∞ (σ−τ)q(σ)²dσ) of the Hastings–McLeod solution q of the Painlevé II equation q″(σ) = σq(σ) + 2q(σ)³, where σ is the independent variable. This type of result — a closed scalar (integro-)differential boundary value problem completely determining the Fredholm determinant — is the gold standard for such evaluations, with analogues for higher-order and multicritical edge kernels (Claeys–Its–Krasovsky; Le Doussal–Majumdar–Schehr; Kimura–Zahabi) that all still assume the wave function enters through an additive composition φ_ξ(x) = φ(x+ξ).

Kimura and Navand start instead from a generic real potential v(x;ξ) and a wave function φ_ξ solving the Schrödinger-type equation (∂²_x − v(x;ξ))φ_ξ(x) = ξφ_ξ(x), and build the kernel K(ξ,ζ) = ∫{ℝ₊} φ_ξ(x)φ_ζ(x)dx. Their motivation comes from double-scaled random matrix models of two-dimensional quantum gravity (e.g. the Virasoro minimal string), where the double scaling limit of orthogonal polynomials produces exactly such wave functions and the gap probability probes non-perturbative discreteness of the spectrum; the class also contains the finite-temperature kernels of the Kardar–Parisi–Zhang equation, the stochastic evolution equation describing growing interfaces. Their main theorem (Theorem 4.1) proves, under a set of regularity assumptions (Conditions 1.2, 1.3, 1.5, 1.6) plus two structural hypotheses, that for I = [τ,∞) the Fredholm determinant is given by an explicit functional of an auxiliary function q alone, where q satisfies a closed second-order integro-differential equation and is fixed by the boundary condition q(σ) → ψ(σ) = φ_σ(0) as σ → +∞; at the Airy point the formulas collapse to the classical Tracy–Widom ones. The two structural hypotheses are: Condition 1.4, requiring the wave function to be of composed form φ_ξ(x) = φ(u(x)+γξ) for twice-differentiable u, φ and γ ≠ 0 — the composed form is what makes the chain rule convert x-derivatives into ξ-derivatives and drives every closure step of the proof — and Condition 3.6, requiring the ratio χ{0,0}(τ_j)/ψ(τ_j) to be well-defined at every endpoint, where χ_{0,a} are auxiliary wave functions appearing in the proof. Along the way, the dynamics of the full infinite family of auxiliary wave functions χ_{n,a}, indexed by n ∈ ℤ≥0 and a ∈ {0, 1, 2}, is shown to admit a Lax formulation — a representation of the dynamics as the compatibility condition of a pair of linear systems, carried by infinite-size matrices A_j and B_ξ — and these matrices formally satisfy the Schlesinger equations, the deformation equations of monodromy-preserving (isomonodromic) systems in the sense of Jimbo–Miwa–Ueno; the componentwise proof, however, is completed only for matrix components with column index p ≥ n, the remaining components being verifiable in principle but combinatorially unmanageable. An alternative Riemann–Hilbert approach (reconstructing an unknown function from prescribed jumps across contours in the complex plane) due to Bothner handles integrable kernels with additive or multiplicative composition without any differential equation, but it does not produce a closed scalar boundary value problem for the determinant. The authors close their paper with seven explicitly stated uncharted directions that they suggest as interesting to explore, the seventh of which they describe as more of a technical comment than a future direction.

Problem Statement

Let v(x;ξ) be a generic real potential and φ_ξ(x) a solution of the Schrödinger-type equation (∂²_x − v(x;ξ))φ_ξ(x) = ξφ_ξ(x), satisfying the standard regularity assumptions (square-integrability on ℝ₊ in x, sufficient decay in ξ so that the associated integral operator is bounded). Define the kernel K(ξ,ζ) = ∫{ℝ₊} φ_ξ(x)φ_ζ(x)dx, the integral operator K₀ acting by this kernel, and the Fredholm determinant F(I) = det(ℐ − K₀Π_I) on L²(ℝ), where Π_I is multiplication by the indicator of the interval I. The common objective is to complete, both analytically and structurally, the Tracy–Widom-type theory of these determinants that Kimura–Navand initiated for the restricted class where φ_ξ(x) = φ(u(x)+γξ) and I = [τ,∞): specifically, to resolve the seven directions the authors explicitly list as open. (1) Remove the composed-form restriction (Condition 1.4): establish the functional formula — an expression of F(I) as an explicit functional of an auxiliary function q governed by a closed second-order integro-differential boundary value problem with q asymptotic to ψ(ξ) = φ_ξ(0) — for arbitrary wave functions solving the Schrödinger-type equation, replacing the chain-rule identity that underlies the current proof by a general closure relation ∂xφ_ξ(x) = ch(x,ξ)∂ξφ_ξ(x) and developing the closure tools it requires. (2) Extend the evaluation to kernels defined by general linear differential operators, Sφ_ξ = ξφ_ξ, beyond the second-order self-adjoint Schrödinger form — for instance (∂^{2p}x − v(x;ξ))φ_ξ = ξφ_ξ, or operators containing odd powers of derivatives where self-adjointness is lost — deriving analogous functional and integro-differential reductions. (3) Generalize the functional formula from I = [τ,∞) to an arbitrary union of intervals I = ⊔{j=1}^{m}[τ{2j−1},τ{2j}] (the gap probability for any interval), noting that the intermediate machinery of the source already holds for such I. (4) Compute the large-gap limit lim{τ→−∞} F([τ,∞)) in the general deformed setting; amongst other interests, this would notably allow a check of compatibility with the Forrester–Chen–Eriksen–Tracy large-gap results. (5) Determine whether the dynamics of the auxiliary wave functions close on χ_{0,0} and χ_{0,1} alone into a finite-size system of Schlesinger equations, closing the system in the n-direction in the way the source closed it in the a-direction. (6) Prove the Schlesinger equations for the infinite Lax matrices A_j and B_ξ for arbitrary components without formal arguments, and derive the associated trace properties in the spirit of Jimbo–Miwa–Ueno, thereby establishing the connection with isomonodromic systems. (7) Determine whether Condition 3.6 can be lifted, i.e. analyze the behavior of χ_{0,0} at the endpoints τ_j where ψ vanishes, in particular whether χ_{0,0} vanishes there more rapidly than ψ, so that the ratio χ_{0,0}/ψ remains meaningful. In the authors' own grouping, items 1–2 are directed towards a universal evaluation of the determinant (arbitrary admissible wave function, arbitrary defining operator), items 3–4 would characterize further the Fredholm determinants considered (arbitrary interval, asymptotics), items 5–6 are related to the Lax formalism and would deliver the rigorous isomonodromic interpretation of the framework, and item 7 is a specific and technical question removing a hypothesis of the main theorem. Answers to items 5 and 7 may legitimately be negative, if supported by explicit counterexamples within the admissible class. The seven directions are posed jointly as the objective above: it is met only when each of them is settled, so progress on a proper subset — while genuine — does not by itself complete it.

References

  1. Taro Kimura and Xavier Navand, "Fredholm determinants from Schrödinger-type equations, and deformation of Tracy–Widom distribution", Letters in Mathematical Physics 116 (2026) 15. DOI: 10.1007/s11005-025-02034-8; arXiv:2408.06888. https://arxiv.org/abs/2408.06888
  2. Craig A. Tracy and Harold Widom, "Level spacing distributions and the Airy kernel", Communications in Mathematical Physics 159 (1994) 151–174. DOI: 10.1007/BF02100489; arXiv:hep-th/9211141. https://arxiv.org/abs/hep-th/9211141
  3. P. J. Forrester, "The spectrum edge of random matrix ensembles", Nuclear Physics B 402 (1993) 709–728. DOI: 10.1016/0550-3213(93)90126-A. https://doi.org/10.1016/0550-3213(93)90126-A
  4. Yang Chen, Kasper J. Eriksen and Craig A. Tracy, "Largest eigenvalue distribution in the double scaling limit of matrix models: A Coulomb fluid approach", Journal of Physics A 28 (1995) L207–L211. DOI: 10.1088/0305-4470/28/7/001; arXiv:hep-th/9502123. https://arxiv.org/abs/hep-th/9502123
  5. Michio Jimbo, Tetsuji Miwa and Kimio Ueno, "Monodromy preserving deformation of linear ordinary differential equations with rational coefficients: I. General theory and τ-function", Physica D 2 (1981) 306–352. DOI: 10.1016/0167-2789(81)90013-0. https://doi.org/10.1016/0167-2789(81)90013-0
  6. Thomas Bothner, "A Riemann-Hilbert approach to Fredholm determinants of Hankel composition operators: scalar-valued kernels", Journal of Functional Analysis 285 (2023) 110160. DOI: 10.1016/j.jfa.2023.110160; arXiv:2205.15007. https://arxiv.org/abs/2205.15007
  7. T. Claeys, A. Its and I. Krasovsky, "Higher-order analogues of the Tracy–Widom distribution and the Painlevé II hierarchy", Communications on Pure and Applied Mathematics 63 (2010) 362–412. DOI: 10.1002/cpa.20284; arXiv:0901.2473. https://arxiv.org/abs/0901.2473
  8. Pierre Le Doussal, Satya N. Majumdar and Grégory Schehr, "Multicritical Edge Statistics for the Momenta of Fermions in Nonharmonic Traps", Physical Review Letters 121 (2018) 030603. DOI: 10.1103/PhysRevLett.121.030603; arXiv:1802.06436. https://arxiv.org/abs/1802.06436
  9. Taro Kimura and Ali Zahabi, "Universal cusp scaling in random partitions", Letters in Mathematical Physics 114 (2024) 27. DOI: 10.1007/s11005-024-01771-6; arXiv:2208.07288. https://arxiv.org/abs/2208.07288
  10. Promit Ghosal and Guilherme L. F. Silva, "Universality for multiplicative statistics of Hermitian random matrices and the integro-differential Painlevé II equation", Communications in Mathematical Physics 397 (2023) 1237–1307. DOI: 10.1007/s00220-022-04518-3; arXiv:2201.12941. https://arxiv.org/abs/2201.12941
  11. Tom Claeys and Guilherme L. F. Silva, "Deformations of biorthogonal ensembles and universality", Electronic Journal of Probability 30 (2025). DOI: 10.1214/25-ejp1409; arXiv:2411.15830. https://arxiv.org/abs/2411.15830
  12. Caio E. Candido, Victor Alves, Thomas Chouteau, Charles F. Santos and Guilherme L. F. Silva, "Deformations of OP ensembles in a bulk critical scaling", Nonlinearity 39 (2025) 015001. DOI: 10.1088/1361-6544/ae29db; arXiv:2506.05622. https://arxiv.org/abs/2506.05622