Circular law
In probability theory, more specifically the study of random matrices, the circular law concerns the distribution of eigenvalues of an n \times n random matrix with independent and identically distributed entries in the limit n \to \infty .
≈ 7 min read
Names in other languages
No names in other languages are held for this entry yet.
Article bodies are English and are not machine-translated. A name in another language is not a translation of the article.
In probability theory, more specifically the study of random matrices, the circular law concerns the distribution of eigenvalues of an n \times n random matrix with independent and identically distributed entries in the limit n \to \infty . It asserts that for any sequence of random matrices whose entries are independent and identically distributed random variables, all with mean zero and variance equal to , the limiting spectral distribution is the uniform distribution over the unit disc.
The probability measure of the Ginibre ensemble satisfies \ln \rho(G) = -\frac\beta 2 \sum_{i j} |G_{ij}|^2 - \ln Z_{\beta, N} = \frac\beta 2 \operatorname{Tr}(GG^*) - \ln Z_{\beta, N} where \beta = 1, 2, 4 respectively for the real, complex, and quaternionic cases; Z_{\beta, N} is a normalization factor; \operatorname{Tr} is the trace; ^* is the matrix adjoint. The most commonly used case is \beta = 2 , and when "Ginibre ensemble" is spoken of, it by default means the \beta = 2 case.
By analogy with the gaussian ensembles, the cases of \beta = 1, 2, 4 are also called the GinOE, GinUE, GinSE, meaning "Ginibre Orthogonal/Unitary/Symplectic Ensemble". For \beta = 2 , the eigenvalues of G_N are distributed according to \rho_N\left(z_1, \ldots, z_N\right)=\frac{1}{Z} \exp \left(-\sum_{k=1}^N\left|z_k\right|^2\right) \prod_{1 \leq j where Z = \pi^N \prod_{k=1}^N k! is a Selberg integral.
Ignoring the term Z , the rest of the formula can be obtained by exploiting the biunitary symmetry of the ensemble. That is, for any unitary U, V , the ensemble UGV has the same distribution. For \beta = 4 , the 2N \times 2N matrix has complex eigenvalues that come in conjugate pairs.
Index the eigenvalues as z_1, \dots, z_n, z_{n+1}, \dots, z_{2n} such that z_{j} = \bar z_{n+j} , then \rho_N\left(z_1, \ldots, z_N\right) \propto \prod_{l=1}^N e^{-2\left|z_l\right|^2}\left|z_l-\bar{z}_l\right|^2 \prod_{1 \leq j 0 For \beta = 1 , the N \times N matrix has N complex eigenvalues, and each eigenvalue has a conjugate that is also an eigenvalue. However, they may no longer come in conjugate pairs, since some eigenvalues may be purely real.
It is not even absolutely continuous, thus does not have a probability density function, but decomposes into sectors depending on the number of real eigenvalues. However, at the N \to \infty limit, the circular law is recovered, since there are only O(\sqrt N) eigenvalues exactly on the real line.
For \beta = 2 , the eigenvalues make up a determinantal point process \rho_{(k), N}\left(z_1, \ldots, z_k\right)=\det\left[K_N\left(z_j, z_l\right)\right]_{j, l=1}^k with correlation kernel K_N(w, z)=\frac{1}{\pi} e^{-\left(|w|^2+|z|^2\right) / 2} \sum_{j=1}^N \frac{(w \bar{z})^{j-1}}{(j-1)!}=\frac{1}{\pi} e^{-\left(|w|^2+|z|^2\right) / 2} e^{w \bar{z}} \frac{\Gamma(N ; w \bar{z})}{\Gamma(N)} where \Gamma(j ; x)=\int_x^{\infty} t^{j-1} e^{-t} d t denotes the upper incomplete gamma function. It has the following asymptotics K_{\infty}^{\mathrm{b}}(w, z):=\lim _{N \rightarrow \infty} K_N(w, z)=\frac{1}{\pi} e^{-\left(|w|^2+|z|^2\right) / 2} e^{w \bar{z}}, K_{\infty}^{\mathrm{e}}\left(z_1, z_2\right):=\lim _{N \rightarrow \infty} K_N\left(-i \sqrt{N}+z_1,-i \sqrt{N}+z_2\right)=e^{-\left(\left|z_1\right|^2+\left|z_2\right|^2\right) / 2} e^{z_1 \bar{z}_2} h\left(\frac{1}{2}\left(-i z_1+i \bar{z}_2\right)\right) where h(z)=\frac{1}{2 \pi}(1+\operatorname{erf}(\sqrt{2} z)) . Plugging in the correlation kernel, the average distribution of all eigenvalues is \frac 1N K_N(z, z) = \frac{\Gamma(N; |z|^2)}{\pi N!} = \frac{\int_ .
For general inverse temperature \beta , the OCP has partition function Z_{N}^{\text{OCP}}(\beta) = \int_{\mathbb C^N} e^{-\beta E} d^Nz , and free energy -\beta^{-1} \ln Z_{\beta, N}^{\text{OCP}} . However, it is theoretically more natural to consider the normalized partition function Z_N^{D_R, \mathrm{OCP}}(\beta)=\frac{1}{N!} A_{N, \beta} Z_N^{\mathrm{OCP}}(\beta), \quad A_{N, \beta}=e^{-\beta N^2\left(\frac{1}{4} \log N-\frac{3}{8}\right)} where the N! part accounts for the fact that the particles are indistinguishable from each other, and A_{N,\beta} removes the self-energy of the average plasma, that is, the self-energy of a disk of radius R = \sqrt N and charge density \frac{1}{\pi} .
Thus, Z_N^{D_R, \mathrm{OCP}}(\beta) is the partition function of a "charge neutral" OCP. The log-partition function satisfies -\ln Z_N^{D_R, \mathrm{OCP}}(\beta)\Big|_{\beta=2}=N \beta f(\beta)\Big|_{\beta=2}+\frac{1}{12} \ln N-\zeta^{\prime}(-1)-\frac{1}{720 N^2}+O\left(\frac{1}{N^4}\right) where \beta f(\beta)\Big|_{\beta=2}=\frac{1}{2} \log \left(\frac{1}{2 \pi^3}\right) is the average free energy per particle at the N \to \infty limit.
It is conjectured that for general \beta , -\ln Z_N^{D_R, \mathrm{OCP}}(\beta)=N \beta f(\beta)\Big|_{\beta=2} + \frac{4 \log (\beta / 2)}{3 \pi^{1 / 2}} \sqrt N + \frac{1}{12} \ln N + O(1) Let G_n be sampled from the real or complex ensemble, and let \rho(G_n) be the absolute value of its maximal eigenvalue: \rho(G_n) := \max_j |\lambda_j| We have the following theorem for the edge statistics: {{Math theorem | math_statement = For G_n and \rho\left(G_n\right) as above, with probability one, \lim _{n \rightarrow \infty} \frac{1}{\sqrt{n}} \rho\left(G_n\right)=1 Moreover, let \gamma_n=\log \left(\frac{n}{2 \pi}\right)-2 \log (\log (n)) , then \sqrt{4 n \gamma_n}\left(\frac{1}{\sqrt{n}} \rho\left(G_n\right)-1-\sqrt{\frac{\gamma_n}{4 n}}\right), converges in distribution to the Gumbel law, i.e., the probability measure on \mathbb{R} with cumulative distribution function F_{\mathrm{Gum}}(x)=e^{-e^{-x}} . Furthermore, for any M, \delta > 0 , almost surely \frac{1}{\sqrt n} \rho(G_n) \in 1 + \frac{1}{\sqrt n} [M, (2 + \delta) \sqrt{\ln n}] for all large n . The theorem still holds for quaternionic non-Hermitian matrix ensembles, with e^{-e^{-x}} replaced by e^{-\sqrt 2 e^{-x}} . | name = Edge statistics of the Ginibre ensemble | note = }} This theorem refines the circular law of the Ginibre ensemble.
In words, the circular law says that the spectrum of \frac{1}{\sqrt{n}} G_n almost surely falls uniformly on the unit disc. and the edge statistics theorem states that the radius of the almost-unit-disk is about 1 + \sqrt{\frac{\gamma_n}{4 n}} + \frac{1}{\sqrt{4 n \gamma_n}} z , where z is a random variable sampled from the standard Gumbel distribution.
History
For random matrices with Gaussian distribution of entries (the Ginibre ensembles), the circular law was established in the 1960s by Jean Ginibre. In the 1980s, Vyacheslav Girko introduced an approach which allowed to establish the circular law for more general distributions.
Further progress was made by Zhidong Bai, who established the circular law under certain smoothness assumptions on the distribution. The assumptions were further relaxed in the works of Terence Tao and Van H. Vu, Guangming Pan and Wang Zhou, and Friedrich Götze and Alexander Tikhomirov. Finally, in 2010 Tao and Vu proved the circular law under the minimal assumptions stated above.
The circular law result was extended in 1985 by Girko to an elliptical law for ensembles of matrices with a fixed amount of correlation between the entries above and below the diagonal. The elliptic and circular laws were further generalized by Aceituno, Rogers and Schomerus to the hypotrochoid law which includes higher order correlations.
See also
Wigner semicircle distribution
Text derived from https://en.wikipedia.org/w/index.php?oldid=1376493623 under CC BY-SA 4.0; modifications have been made.