--- title: Random Bergman Metrics url: https://www.emergentmind.com/topics/random-bergman-metrics type: topic --- # Random Bergman Metrics Searching arXiv for recent and foundational papers on random Bergman metrics and closely related Bergman-kernel probabilistic geometry. Random Bergman metrics are probability-theoretic models on the finite-dimensional spaces of Bergman metrics that approximate the infinite-dimensional space of Kähler metrics in a fixed Kähler class. On a compact Kähler manifold \((M,\omega_0)\) with a positive holomorphic line bundle \(L\to M\), the basic construction replaces formal path integrals on \[ K([\omega_0])=\{\phi\in C^\infty(M)/\mathbb{R}\mid \omega_\phi:=\omega_0+i\partial\bar\partial \phi>0\} \] by matrix integrals on symmetric spaces of Bergman metrics, typically \[ \mathcal B_k \simeq SL(N_k,\mathbb C)/SU(N_k),\qquad N_k=\dim H^0(M,L^k). \] Randomness is introduced by choosing probability measures on the positive Hermitian matrices that parametrize \(\mathcal B_k\), or by weighting the invariant Haar measure with geometric action functionals. This framework connects Kähler quantization, Bergman kernel asymptotics, random matrix models, large deviations, and geometric stability [1107.4575, 1112.4382, 1404.0659, 2303.11559]. ## 1. Geometric definition and finite-dimensional model Let \((M,\omega_0)\) be a compact Kähler manifold of complex dimension \(n\), with \([\omega_0]\in H^{1,1}(M,\mathbb{Z})\). Then there is a holomorphic line bundle \(L\to M\) with \(c_1(L)=[\omega_0]\), and for each \(k\) one considers the finite-dimensional space of holomorphic sections \[ H^0(M,L^k),\qquad N_k=\dim H^0(M,L^k)\sim k^n+O(k^{n-1}). \] Given an orthonormal basis of sections \(\{s_i(z)\}_{i=1}^{N_k}\), a Bergman metric is defined from a positive Hermitian matrix \(P\) by the Kähler potential \[ \phi_P(z)=\log\!\Big(\sum_{i,j=1}^{N_k}\overline{s_i(z)}\,P_{ij}\,s_j(z)\Big), \] and the associated Kähler form is \[ \omega_P=\omega_0+i\partial\bar\partial \phi_P. \] Since only the projective class of \(P\) matters, one modds out by overall scalars, and the space of Bergman metrics is identified with the symmetric space \(\mathcal B_k\simeq SL(N_k,\mathbb C)/SU(N_k)\). In the survey formulation, the same space is described as \[ B_k\cong \mathrm{GL}(d_k,\mathbb C)/U(d_k)\cong P_{d_k}, \] where \(P_{d_k}\) is the space of positive definite Hermitian \(d_k\times d_k\) matrices with determinant one [1112.4382, 2303.11559]. The parameterization is basis-independent in the precise sense that, for a fixed orthonormal basis \(\{S_1^k,\dots,S_{d_k}^k\}\), any other basis can be written as \[ \sigma_j^A=\sum_l A_{jl}S_l^k, \] and the corresponding metric depends only on \[ P=A^*A. \] Equivalently, Bergman metrics arise as pullbacks of the Fubini–Study form under Kodaira embeddings, for instance \[ \frac{1}{k}\,\iota_\sigma^*\omega_{FS}=\frac{i}{2k}\,\partial\bar\partial \log\sum_{j=1}^{d_k}|\sigma_j|^2. \] The finite-dimensional spaces \(B_k\) or \(\mathcal B_k\) serve as approximants to the full space of Kähler metrics. The key approximation statements are \[ \mathcal K[\omega_0]=\lim_{k\to\infty}\mathcal B_k \] and \[ \lim_{k\to\infty}FS_k\circ \operatorname{Hilb}_k(\phi)=\phi. \] By the Tian–Yau–Zelditch expansion, the canonical Bergman metric corresponding to the identity matrix satisfies \[ \omega_{I_k}=\omega_0+O(k^{-2}), \] so Bergman metrics approximate arbitrary Kähler metrics increasingly well as \(k\to\infty\) [1107.4575, 1112.4382, 2303.11559]. ## 2. Measures on Bergman metric spaces A random Bergman metric is obtained by choosing a probability measure on the positive Hermitian matrix \(P\). One class consists of unitarily invariant or eigenvalue-type measures \[ d\mu(P)=F(\lambda)\,[dP]_{\mathrm{Haar}}, \] where \(P=U^\dagger \Lambda U\), \(\Lambda=\mathrm{diag}(\lambda_1,\dots,\lambda_{N_k})\), and \([dP]_{\mathrm{Haar}}\) is the invariant measure on positive Hermitian matrices. The ensemble is gauge-fixed to the Bergman metric space by enforcing \(\det P=1\), giving a measure \(d\mu_{\mathcal B_k}(P)\) on \(\mathcal B_k\) [1112.4382]. A second construction uses heat kernel measures on the symmetric space of Bergman metrics: \[ d\mu_k^t(P)=p_k(t,I_k,P)\,dP, \] where \(p_k(t,P_1,P_2)\) is the heat kernel on \(P_{d_k}\) and \(dP\) is Haar measure. This measure is invariant under \(U(d_k)\), independent of the choice of orthonormal basis, and interpreted as Brownian motion on \(P_{d_k}\) starting at the identity [2303.11559]. A third construction weights the invariant Haar measure by a geometric action functional. The basic partition function is \[ Z_k(Y_k)=\int_{\mathcal B_k} e^{-Y_k S_k(\omega_0,P)}\,D_k\omega_0, \] where \(D_k\omega_0\) is the invariant Haar measure on \(\mathcal B_k\), \(Y_k\) is a coupling constant, and \(S_k\) is a geometric action functional. Because the Haar volume is infinite and grows rapidly along noncompact geodesic directions, convergence of \(Z_k\) becomes a central issue [1404.0659]. | Construction | Measure | Structural feature | |---|---|---| | Eigenvalue-type ensemble | \(d\mu(P)=F(\lambda)\,[dP]_{\mathrm{Haar}}\) | Unitary invariance | | Wishart ensemble | \(d\mu(P)=C_{g,a}\,e^{-g\,\mathrm{tr}P}(\det P)^a[dP]_{\mathrm{Haar}}\) | Exactly solvable positive-matrix model | | Heat kernel ensemble | \(d\mu_k^t(P)=p_k(t,I_k,P)\,dP\) | Brownian motion on \(P_{d_k}\) | | Action-weighted ensemble | \(Z_k(Y_k)=\int_{\mathcal B_k} e^{-Y_k S_k}\,D_k\omega_0\) | Geometric partition function | These constructions are complementary rather than competing. The matrix-model viewpoint is used both as a regularization of path integrals over Kähler metrics and as a source of explicit probabilistic models on \(\mathcal B_k\) [1107.4575, 1112.4382, 1404.0659]. ## 3. Observables, correlation functions, and solvable models The basic observables are correlators of the Bergman potential \[ \phi_P(z)=\log\!\big(s^\dagger(z)Ps(z)\big), \] and of the metric \[ \omega_P=\omega_0+i\partial\bar\partial \phi_P. \] The \(m\)-point functions are defined schematically by \[ \big\langle \omega(z_1)\cdots \omega(z_m)\big\rangle =\int \omega_P(z_1)\cdots \omega_P(z_m)\, d\mu_{\mathcal B_k}(P). \] Because \(\phi_P\) depends on the directions of the section vectors \(s(z)\), these correlators require nontrivial angular integration over \(U(N_k)\); the Harish-Chandra–Itzykson–Zuber formula is used to perform this integration exactly [1112.4382]. For any eigenvalue-type measure, the one-point function is universal: \[ \langle \omega_P(z)\rangle=\omega_0(z)+O(k^{-2}), \qquad \lim_{k\to\infty}\langle \omega_P(z)\rangle=\omega_0(z). \] Thus the mean random Bergman metric converges to the background metric. The two-point function carries the nontrivial fluctuation data. After angular integration, all dependence on the points \(z_1,z_2\) enters through the normalized off-diagonal Bergman kernel \[ \cos^2\theta_{12} = \frac{\big|\langle s(z_1),s(z_2)\rangle\big|^2}{|s(z_1)|^2\,|s(z_2)|^2} = e^{-kD_0(z_1,z_2)}, \] where \(D_0\) is the Calabi diastatic function of the background Bergman metric. In the large-\(k\) limit, the general structure is \[ \langle \omega(z_1)\omega(z_2)\rangle = \omega_0(z_1)\omega_0(z_2)+\text{(contact term supported on }z_1=z_2\text{)}. \] Away from the diagonal the correlator factorizes, while near the diagonal there is a contact term reflecting short-distance metric fluctuations [1112.4382]. The Wishart model is the main solvable example: \[ d\mu(P)=C_{g,a}\,e^{-g\,\mathrm{tr}P}(\det P)^a[dP]_{\mathrm{Haar}}. \] It allows explicit finite-\(k\) formulas for correlators. In particular, the two-point function can be written in terms of a hypergeometric function involving \[ {}_2F_1(t_1,t_2,N_k+a;\cos^2\theta_{12}), \] and the final finite-\(k\) expression depends on the geometry only through \(\cos^2\theta_{12}\). In the off-diagonal regime \(z_1\neq z_2\), \[ \langle \omega(z_1)\omega(z_2)\rangle\to \omega_0(z_1)\omega_0(z_2), \] with corrections suppressed as \(e^{-kD_0(z_1,z_2)}\). For the standard scaling of the Wishart parameter \(g\), the contact terms vanish in the large-\(k\) limit, and the ensemble becomes sharply concentrated around the background metric [1112.4382]. A recurrent limitation is that, because the random variable is \(P\) itself, the resulting measures may become too sharply concentrated at large \(k\). The suggestion that a smoother and more physically interesting geometry may arise if one uses \(-\log P\) as the random variable identifies a technical issue in simple matrix-model ensembles rather than a contradiction in the framework [1112.4382]. ## 4. Stability functionals and the critical coupling constant A distinct line of development studies partition functions on \(\mathcal B_k\) with actions drawn from geometric functionals called stability functions. In the GIT setting, a stability function is a function on \(G^\mathbb C\) whose gradient is the moment map, written abstractly as \[ \psi_v(g)=\log |g\cdot v|^2. \] When restricted to a geodesic \(P_t\) in \(\mathcal B_k\), a stability function is asymptotically linear: \[ S_k(P_t)\sim A\,t-B,\qquad t\to\infty, \] where \(A\) is the asymptotic slope and \(B\) is the \(y\)-intercept. In the stable case, the slope is positive along all geodesic rays, so the action gives a “V-shaped” confining potential. The worst geodesic rays are those with the smallest asymptotic slope, and these least stable directions determine the threshold for convergence of the partition function [1404.0659]. The corresponding invariant is the critical coupling constant \[ Y_k^{\rm crit}=\inf\{Y_k:\ Z_k(Y_k)<\infty\}. \] If \(Y_kY_k^{\rm crit}\), the Boltzmann damping dominates the Haar growth and the integral converges. This invariant measures the minimal degree of stability of geodesic rays in \(\mathcal B_k\) relative to the chosen action. A rescaled version is also defined when the symmetric-space metric is dilated by a factor \(E_k\): \[ Y_{k,\mathrm{crit}}^{E_k}=\inf\{Y_k:\ Z_k^{E_k}(Y_k)<\infty\}. \] The main explicit computation is for the \(\nu\)-balancing energy \(S_{\nu,k}\), the normalized restriction to \(\mathcal B_k\) of \[ S_\nu(\omega_0,\phi)=\mathcal L_{\omega_0,\nu}(\phi), \] with matrix form \[ S_{\nu,k}(\omega_0,\phi(P)) = kN_k V\int_M \phi(P)\,\nu-\log\det P. \] This is a convex stability functional on \(\mathcal B_k\), bounded below by \(0\), and its critical point is the \(\nu\)-balanced metric characterized by \[ \int_M \frac{s_i \bar s_j}{\sum_{l,m}\bar s_l P_{lm} s_m}\,\nu = (P^{-1})_{ji}. \] Along a geodesic \[ P_t=Ue^{tA}U^*,\qquad A=\operatorname{diag}(a_1,\dots,a_{N_k}), \] the asymptotics are \[ S_{\nu,k}(\omega_0,\phi(P_t))=N_k a_{N_k} t+O(1),\qquad t\to\infty, \] with uniformly bounded \(y\)-intercept and positive slope \[ A=N_k a_{\max}. \] The exact threshold for the standard Haar measure is \[ Y_k^{\rm crit}=N_k-1, \] equivalently, the partition function converges iff \[ Y_k>N_k-1. \] For the scaled measure one obtains \[ Y_{k,\mathrm{crit}}^{E_k}=E_k(N_k-1). \] On a compact Riemann surface of genus \(h\), Riemann–Roch gives \[ N_k=k+1-h, \] hence \[ Y_k^{\rm crit}=k-h. \] The proof identifies a precise competition between the linear growth of \(S_{\nu,k}\) and the exponential growth of the Haar volume form, whose eigenvalue-coordinate asymptotics involve the Vandermonde-type factor \[ \prod_{i