Papers
Topics
Authors
Recent
2000 character limit reached

Toeplitz Determinants with FH Singularities

Updated 16 December 2025
  • Toeplitz determinants with FH singularities are matrix determinants defined by symbols with jump and root-type singularities on the unit circle, capturing complex scaling behavior.
  • They are analyzed via asymptotic expansions that include power-law factors, oscillatory terms, and crossover regimes governed by Painlevé equations.
  • Applications span statistical mechanics, random matrix theory, and quantum physics, where these determinants encode correlation functions and gap probabilities.

A Toeplitz determinant with Fisher-Hartwig (FH) singularities is a determinant of a Toeplitz matrix whose underlying symbol exhibits root-type and/or jump-type singularities at a finite set of points on the unit circle. These structures arise naturally in statistical mechanics, random matrix theory, fermionic physics, and integrable probability, where the determinants encode correlation functions, gap probabilities, or multiplicative statistics. The Fisher-Hartwig conjecture (and subsequent theorem) provides their precise asymptotic expansion as the matrix size grows, capturing leading powers, amplitudes, oscillatory terms, and the subtle effects of merging or interacting singularities.

1. Fisher-Hartwig Symbols and Toeplitz Determinants

A Toeplitz matrix of dimension nn is specified by a function (“symbol”) f(z)f(z) defined on the unit circle z=1|z|=1, with entries [fjk]0j,kn1[f_{j-k}]_{0\leq j,k\leq n-1}, where fmf_{m} is the mm-th Fourier coefficient of ff. The Fisher-Hartwig symbol is a product of a smooth (nonvanishing, analytic) factor and finitely many singular terms located at points zj=eiφjz_j=e^{i\varphi_j}: f(eiθ)=eV(eiθ)j=1meiθzj2αjeiβjsgn(θφj)f(e^{i\theta})=e^{V(e^{i\theta})} \prod_{j=1}^m |e^{i\theta}-z_j|^{2\alpha_j} e^{i\beta_j \,\mathrm{sgn}(\theta-\varphi_j)} where V(z)V(z) is real-analytic near z=1|z|=1, and (αj,βj)(\alpha_j, \beta_j) are the root and jump exponents at zjz_j. The Toeplitz determinant is defined as Dn[f]=det(fjk)0j,kn1D_n[f] = \det(f_{j-k})_{0\leq j,k\leq n-1} or equivalently via the Heine integral as a partition function with log-gas interactions on the circle (Bourgade et al., 9 Dec 2025, Abanov et al., 2011).

2. Asymptotic Formulas: The Fisher-Hartwig Expansion

For large nn, the Fisher-Hartwig formula provides the leading behavior of Dn[f]D_n[f]: Dn[f]=E[f]nnj=1m(αj2βj2)j=1mG(1+αj+βj)G(1+αjβj)G(1+2αj)1j<kmzjzk2(βjβkαjαk)(1+o(1))D_n[f] = E[f]^n\, n^{\sum_{j=1}^m (\alpha_j^2 - \beta_j^2)}\, \prod_{j=1}^m \frac{G(1+\alpha_j+\beta_j)G(1+\alpha_j-\beta_j)}{G(1+2\alpha_j)} \prod_{1\leq j<k\leq m} |z_j-z_k|^{2(\beta_j\beta_k - \alpha_j\alpha_k)} (1+o(1)) where E[f]=exp(V0)E[f]=\exp(V_{0}) with V0V_0 the zero Fourier mode of VV and GG is the Barnes G-function (Bourgade et al., 9 Dec 2025, Abanov et al., 2011, Fahs, 2019). The powers of nn encode local singular behavior at each zjz_j. Further corrections involve oscillatory and polynomial-in-$1/n$ terms arising from “branches” in the FH representations, corresponding to shifts in the integer part of the jumps: Dn[f]{kj}exp[nV0+injkjφjj(βjkj)2lnn]A({αj,βj;kj})exp(=1c({αj,βj;kj})n)D_n[f] \sim \sum_{\{k_j\}} \exp[n V_0 + i n\sum_j k_j \varphi_j - \sum_j (\beta_j - k_j)^2 \ln n]\, \mathcal{A}(\{\alpha_j, \beta_j; k_j\}) \exp\left(\sum_{\ell=1}^{\infty} \frac{c_\ell(\{\alpha_j,\beta_j; k_j\})}{n^{\ell}}\right) The sum is over all relevant “branches” (integer shifts kjk_j such that jkj=0\sum_j k_j = 0) (Ivanov et al., 2011, Ivanov et al., 2013). At special values of the exponents, competing branches contribute equally, producing so-called “switching” and crossover behavior (Ivanov et al., 2011). When singularities merge, the asymptotics are given by a uniform expression involving Painlevé V transcendent (Claeys et al., 2010, Claeys et al., 2014, Forkel, 2023).

3. Merging Singularities and Painlevé Transitions

As two Fisher-Hartwig singularities at z1,z2z_1, z_2 coalesce (z1z2z_1\to z_2), the simple product formula is no longer sufficient. Instead, the determinant exhibits a crossover, governed by a universal Painlevé V equation (“double-scaling” coalescence regime): logDn(ft)=logDn(f0)+int(β2β1)+i02int[σ(s)c]dss+O(1)\log D_n(f_t) = \log D_n(f_0) + i n t (\beta_2 - \beta_1) + i\int_{0}^{-2i n t} [\sigma(s) - c ] \frac{ds}{s} + O(1) Here ftf_t is the symbol with two singularities at separation $2t$, f0f_0 the merged (single singularity) symbol, and σ(s)\sigma(s) is a solution to the Jimbo–Miwa–Okamoto σ\sigma-form of Painlevé V with explicit exponents determined by αj\alpha_j, βj\beta_j (Claeys et al., 2014, Claeys et al., 2010, Forkel, 2023). This formula interpolates between regimes of two isolated singularities (t>0t>0 fixed, nn\to\infty) and the merged-case (t0t\to0), capturing the relevant logarithmic corrections and oscillatory behavior.

4. Riemann–Hilbert Approach and Subleading Expansions

The technical machinery underpinning the FH asymptotics exploits the Riemann–Hilbert (RH) problem for orthogonal polynomials on the unit circle (OPUC) with respect to the symbol ff. The Deift–Zhou nonlinear steepest-descent method is applied to construct global (“outer”) and local (“inner,” singular) parametrices that asymptotically solve the RH problem (Deift et al., 2012, Fahs, 2019, Bourgade et al., 9 Dec 2025). Local parametrices near singularities are built from Bessel, confluent hypergeometric, or Painlevé functions, depending on the nature of the singularity and the scaling regime. Matching and recursions yield not only the leading order but also subleading corrections (in inverse powers of nn), with polynomial or oscillatory structure; periodicity of the amplitudes under shifts of the jump exponents has been established (Ivanov et al., 2011, Ivanov et al., 2013).

5. Physical and Probabilistic Applications

Toeplitz determinants with FH singularities are central in quantum many-body physics and random matrix theory. In non-equilibrium Fermi-edge physics and Luttinger liquid models, multiple-step distribution functions yield Toeplitz symbols with multiple jump-type singularities; the resulting determinants encode the asymptotics of Green functions, tunneling currents, and full counting statistics, with exponents and oscillations dictated by the FH parameters and branch structure (Protopopov et al., 2012, Gutman et al., 2010, Protopopov et al., 2012). In random matrix theory, the moments of characteristic polynomials (and their distributions) are written as Toeplitz determinants with FH singularities; in the scaling limits they interpolate between regimes of Gaussian multiplicative chaos and classical log-correlated fields (Forkel, 2023, Webb, 2015).

In gap probabilities for CUE, symbols with jump/root singularities describe hard-edge or soft-edge universality, with the exact gap probabilities expressed via Fredholm determinants of hypergeometric or sine kernels, and their asymptotics reduced to a Painlevé or RHP analysis (Xu et al., 2019, Claeys et al., 2010). In the Ising model, the diagonal susceptibility and spin-spin correlation functions are governed by sums over Toeplitz determinants with FH symbols deformed by external parameters, exhibiting phase transitions and singularities tracked by the FH structure (Tracy et al., 2015, Claeys et al., 2010).

6. Discrete and Non-Hermitian Generalizations

In discrete settings (e.g., log-gas on the roots of unity, characteristic polynomials of random permutation matrices), discrete Toeplitz determinants with FH singularities exhibit the same asymptotic formulas as in the continuum as long as the system is sufficiently dilute; deviations and additional correction factors arise for dense occupation (Webb, 2015). The Fisher-Hartwig asymptotics have also been generalized to two-dimensional settings, notably for non-Hermitian random matrix ensembles, where the partition function with boundary singularities is governed by analogous, but spatially extended, expressions and universality classes (Bourgade et al., 9 Dec 2025).

7. Summary Table: Fisher-Hartwig Asymptotics

Context Symbol Structure Leading Asymptotic Behavior
Isolated singularities f(z)=eV(z)zzj2αjeiβjarg(zzj)f(z) = e^{V(z)}\prod |z-z_j|^{2\alpha_j} e^{i\beta_j\arg(z-z_j)} E[f]nnΣj(αj2βj2)×E[f]^n\, n^{\Sigma_j(\alpha_j^2-\beta_j^2)} \times GG-factors (Bourgade et al., 9 Dec 2025)
Multiple-step/jump scenario Discontinuities in n(ϵ)n(\epsilon) Sum over FH branches; exponents determined by βj\beta_j (Protopopov et al., 2012)
Merging singularities ftf_t with z1 ⁣ ⁣z2z_1\!\to\! z_2 Uniform expansion involving Painlevé V σ\sigma-transcendent (Claeys et al., 2014)
Discrete log-gas ff sampled at MM-th roots of unity FH formula holds for NMN\ll M, corrections otherwise (Webb, 2015)
Non-Hermitian ensembles Edge singularities on z=1|z|=1 2d analog of FH; partition function asymptotics (Bourgade et al., 9 Dec 2025)

This synthesis of the Fisher-Hartwig theory and its extensions reveals the central role of Toeplitz determinants with singular symbols in capturing universal scaling, transition phenomena, and fine structure across mathematical physics, with methods and results grounded in Riemann–Hilbert analysis, asymptotic expansions including both universal constants and highly nontrivial oscillatory and merging corrections, and rich connections to integrable systems and random matrices.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (15)

Whiteboard

Follow Topic

Get notified by email when new papers are published related to Toeplitz Determinants with Fisher-Hartwig Singularities.