Average Winding Number for Determinantal Curves associated with 2-Matrix Models in the Class AIII
Mathieu Yahiaoui, Mario Kieburg
TL;DR
This work studies the winding number of the determinantal curve det(K(p)) for one-dimensional disordered quantum systems in symmetry class AIII, using non-Hermitian random-matrix models on the unit circle. It generalizes the partition-function framework from the Ginibre ensemble to a broad class of additive 2-matrix Polya ensembles on GL_N(C), obtaining an exact finite-N expression Z^{(N)}_{m} = det(˜Q^{(N)}_{m}[ω])/det(Q_m) and enabling large-N analysis. In the Muttalib–Borodin Laguerre case, the authors derive a precise asymptotic expansion for the mean winding number: E[Wind_N] = (1/(2π i)) ∮ ( u_γ(p)†ν_γ'(p))/||ν_γ(p)||^2 dp · N + (γ−1−2δ)/2 · (1/(2π i)) ∮ (κ'(p)/κ(p)) · (|a(p)|^{2γ}−|b(p)|^{2γ})/(|a(p)|^{2γ}+|b(p)|^{2γ}) dp + o(1). The leading term corresponds to a geometric Aharonov–Anandan phase, while the subleading correction encodes ensemble-dependent tail information and reduces to zero in the Gaussian limit γ=1; the results illuminate universal versus model-specific aspects of topological invariants in disordered quantum systems and point to broader future directions, including CLTs for winding-number fluctuations and extensions to other symmetry classes. The methodology combines Pólya ensemble theory, Mellin transform techniques, and determinant-based integrals to connect random-matrix structure with topological winding on S^1.
Abstract
To classify one-dimensional disordered quantum systems with chiral symmetry, we analyse the winding number of the determinant of a parametrized non-Hermitian random matrix field over the unit circle modelling the off-diagonal block of a disordered chiral Hamiltonian. The associated partition function is computed explicitly for a broad class of additive two-matrix models extending beyond the Ginibre Unitary Ensemble. In the large-dimension limit, we derive an asymptotic expansion of the average winding number whose leading term exhibits universal features, up to the tail behaviour of the underlying random matrix ensemble, and identify a new correction term absent in the previously studied Ginibre case.
