--- title: Complex Elliptic Ginibre Ensemble url: https://www.emergentmind.com/topics/complex-elliptic-ginibre-ensemble type: topic --- # Complex Elliptic Ginibre Ensemble The complex elliptic Ginibre ensemble is a one-parameter Gaussian family of non-Hermitian random matrices that interpolates between the complex Ginibre ensemble and the Gaussian unitary ensemble (GUE). It is simultaneously a matrix model with correlated entries, a determinantal point process in the complex plane, and a two-dimensional one-component Coulomb gas at inverse temperature \(\beta=2\) in a quadrupolar field. Its interpolation parameter controls the passage from a genuinely two-dimensional complex spectrum to an almost Hermitian regime, and the resulting model supports a detailed theory of global elliptic laws, weak- and strong-non-Hermiticity limits, eigenvector non-orthogonality, characteristic polynomial asymptotics, and time-domain spectral diagnostics [2203.00287, 2306.16720]. ## 1. Definition and exact determinantal structure A standard formulation defines the ensemble on complex \(n\times n\) matrices \(M_n\) by \[ \mathbb{P}_{n,\tau}(M_n)=Z_{n,\tau}^{-1}\exp\left\{-\frac{n}{1-\tau^2}\Big(M_n^\dagger M_n-\frac{\tau}{2}\big(M_n^2+(M_n^\dagger)^2\big)\Big)\right\}\,dM_n,\qquad \tau\in[0,1), \] so that \(\tau=0\) gives the complex Ginibre ensemble and \(\tau\uparrow1\) approaches the Hermitian regime [2212.00525]. Equivalent Gaussian realizations write the matrix as a linear combination of two independent GUE matrices; one form is \[ A_{n,t}=\sqrt{\frac{1+t}{2}}\,X_n+i\sqrt{\frac{1-t}{2}}\,Y_n, \] with \(X_n\) and \(Y_n\) independent GUE matrices, while another uses the Hermitian and anti-Hermitian parts of a complex Ginibre matrix [2306.16720, 2602.20469]. Different papers use \(\tau\), \(t\), or \(T\) for the same Hermiticity parameter and either fix the entry variance to \(1/N\) or divide the matrix by \(\sqrt n\); these are normalization conventions rather than distinct models. For \(d=1\), the ensemble is a determinantal point process with kernel \[ \mathbb K_n(Z,Z') = \frac{n}{\pi\sqrt{1-\tau^2}} \frac{1}{\sqrt{\omega(\sqrt n Z)\,\omega(\sqrt n Z')}}\, \sum_{j=0}^{n-1}\frac{1}{j!}\left(\frac{\tau}{2}\right)^j H_j\!\left(\sqrt{\frac{n}{2\tau}}\,Z\right) H_j\!\left(\sqrt{\frac{n}{2\tau}}\,\overline{Z'}\right), \] with elliptic Gaussian weight \[ \omega(Z)=\exp\!\left(-\frac{1}{1-\tau^2}\left(|Z|^2-\frac{\tau}{2}(Z^2+\overline Z^{\,2})\right)\right) =\exp\!\left(-\frac{(\Re Z)^2}{1+\tau}-\frac{(\Im Z)^2}{1-\tau}\right). \] The correlation functions are determinants of this kernel, and the orthogonal-polynomial input is supplied by planar Hermite polynomials [2203.00287]. This exact solvability underlies most asymptotic results for the model. ## 2. Global spectral geometry and the numerical range At macroscopic scale, the empirical eigenvalue distribution converges to the uniform measure on an ellipse. In one standard normalization the limiting support is \[ E_\tau=\left\{x+iy\in\mathbb C:\left(\frac{x}{1+\tau}\right)^2+\left(\frac{y}{1-\tau}\right)^2\le 1\right\}, \] with semiaxes \(1+\tau\) and \(1-\tau\); at \(\tau=0\) this is the circular law, while \(\tau=1\) degenerates to the interval \([-2,2]\) in the Hermitian limit [2306.16720]. In the kernel formalism, the limiting density is \[ \lim_{n\to\infty}\frac1n\mathbb K_n(z,z)=\frac{\mathbf 1_{z\in\mathcal E_\tau}}{\pi(1-\tau^2)}, \] which is the elliptic law in projection-kernel form [2203.00287]. The eigenvalue ellipse does not exhaust the geometry relevant to non-normality. For the elliptic Ginibre ensemble, the numerical range \(W(X)\) has a distinct large-\(N\) limit. If \(X^{\rm e}\) is the elliptic Ginibre matrix with Ginibre entry variance \(1/N\), then \[ \lim_{N\to\infty} d_H\!\big(W(X^{\rm e}),E_{a,b}\big)=0 \quad\text{a.s.}, \] where \[ E_{a,b}=\left\{(x,y)\in\mathbb R^2:\frac{x^2}{a^2}+\frac{y^2}{b^2}\le 1\right\}, \qquad a=\sqrt{2(1+\tau)},\quad b=\sqrt{2(1-\tau)}. \] Thus the limiting numerical range is again an ellipse, but with semiaxes enlarged by a factor \(\sqrt2\) relative to the spectral support [2602.20469]. This distinction is structurally important. The numerical range is always convex; for normal matrices it equals the convex hull of the spectrum, whereas for non-normal matrices it contains the spectrum and captures instability, pseudospectral effects, and eigenvector non-orthogonality. In the elliptic Ginibre case, the spectrum lies in the ellipse with axes \(1+\tau\) and \(1-\tau\), while the numerical range converges to the “fatter” ellipse with axes \(\sqrt{2(1+\tau)}\) and \(\sqrt{2(1-\tau)}\). In the Hermitian limit \(\tau\to1\), the numerical-range ellipse collapses to \([-2,2]\), matching the convex hull of the limiting GUE spectrum [2602.20469]. ## 3. Strong and weak non-Hermiticity regimes At fixed \(\tau\in(-1,1)\), the ensemble is in the strong non-Hermiticity regime. The global support remains elliptic, and local bulk correlations are Ginibre. For both the standard elliptic ensemble and certain non-Gaussian fixed-trace or trace-squared extensions, the microscopic bulk kernel converges to \[ K_{\mathrm{strong}}(z,w)=\frac1\pi \exp\!\left(-\frac{z^2+\bar w^2}{2}+z\bar w\right), \] so the strong-non-Hermitian bulk universality class is the same as Ginibre [1610.06517]. A complementary contour-integral analysis shows that, in the bulk, the residue at \(s=\tau\) dominates and produces the Ginibre kernel, whereas boundary saddle points control macroscopic edge contributions [2203.00287]. Weak non-Hermiticity is obtained by letting \(\tau\) approach \(1\) with \(N\). In the bulk, one standard scaling is \[ \tau_N=1-\frac{\kappa}{N}, \] or equivalently \(\tau_N=1-\frac{\kappa^2}{2N\nu(X)^2}\) at a bulk point \(X\). In this regime the global support collapses to the real axis with semicircular density, but the local point process remains genuinely complex and converges to the Fyodorov–Khoruzhenko–Sommers kernel \[ K_{\mathrm{weak}}(z_1,z_2) = \frac{\sqrt2}{\sqrt\pi}\, \exp\!\left(-\frac{y_1^2+y_2^2}{\kappa^2}\right) \frac{1}{2\pi}\int_{-\pi}^{\pi} \exp\!\left(-\frac{\kappa^2u^2}{2}+iu(z_1-\bar z_2)\right)\,du. \] As \(\kappa\to0\), this collapses to the Hermitian sine-kernel limit after integrating out the imaginary directions; as \(\kappa\to\infty\), it recovers Ginibre statistics after the appropriate rescaling [1610.06517]. Two observable-specific weak-non-Hermiticity crossover laws have been derived in detail. For bulk spacings of ordered real parts, the spacing distribution is a generalized Gaudin–Mehta law \(\wp(s,\sigma)\), defined through a Fredholm determinant of a two-dimensional weakly non-Hermitian sine kernel. It interpolates between the GUE Gaudin–Mehta distribution \(F_1(s)\) as \(\sigma\downarrow0\) and the Poisson law \(1-e^{-s}\) as \(\sigma\to\infty\) [2212.00525]. At the right spectral edge, with \[ \sigma_n=n^{1/6}\sqrt{1-\tau_n}, \] the limiting distribution \(F_\sigma(t)\) of the rightmost eigenvalue is represented by a Fredholm determinant and by an integro-differential Painlevé-II-type equation. It interpolates between the Tracy–Widom GUE law as \(\sigma\downarrow0\) and a Gumbel limit after the appropriate centering and scaling as \(\sigma\to\infty\) [2208.04684]. ## 4. Eigenvectors, overlaps, and non-normal dynamics Unlike the GUE, the complex elliptic Ginibre ensemble is intrinsically non-normal, so left and right eigenvectors are distinct and their overlaps are nontrivial. In the stochastic-process formulation, one considers a matrix-valued diffusion \(J(t)\) with Hermiticity parameter \(T\in(-1,1)\). Its complex eigenvalues satisfy Dyson-like stochastic differential equations \[ d\lambda_i(t)=dM_i(t)+T\sum_{j\ne i}\frac{1}{\lambda_i(t)-\lambda_j(t)}\,dt, \] while the quadratic covariations are governed exactly by eigenvector overlaps: \[ d(\lambda_i,\lambda_j)_t=O_{ij}(t)\,dt. \] Thus the overlap processes are not auxiliary observables; they are the local covariance structure of eigenvalue motion. For \(T=1\) one recovers Dyson Brownian motion, for \(T=0\) the drift vanishes and the eigenvalues are complex martingales, and for all \(-1