--- title: Average Bipartite Entanglement Entropy url: https://www.emergentmind.com/topics/average-bipartite-entanglement-entropy type: topic --- # Average Bipartite Entanglement Entropy Average bipartite entanglement entropy quantifies the mean von Neumann entropy of a subsystem resulting from a bipartition of a pure quantum state, serving as a fundamental measure of quantum correlations under random-state or dynamical generation protocols. It is central to quantum information theory, statistical mechanics, and condensed matter physics, providing a rigorous framework to characterize typicality, thermalization, and the structure of quantum phases in complex quantum systems. The analytics predominantly rely on ensemble averages over the Haar measure, random matrix ensembles, or symmetry-constrained sectors, with universal formulas established for diverse regimes. ## 1. Mathematical Framework for Bipartite Entanglement Entropy For a pure quantum state $|\psi\rangle$ in a composite Hilbert space $\mathcal{H}_A \otimes \mathcal{H}_B$, the reduced density matrix of subsystem $A$ is $\rho_A = \mathrm{Tr}_B |\psi\rangle\langle\psi|$. The bipartite von Neumann entropy is defined as \[ S_{vN}(\rho_A) = -\mathrm{Tr}[\rho_A \ln \rho_A]\,, \] and quantifies the entanglement between $A$ and $B$. The average bipartite entropy is computed as an expectation over the relevant ensemble—most commonly the Haar measure (uniform distribution) over pure states. For $\mathrm{dim}(\mathcal{H}_A)=m$ and $\mathrm{dim}(\mathcal{H}_B)=n$ with $m \leq n$, the average entropy is \[ \langle S_{vN} \rangle = \psi_0(mn+1) - \psi_0(n) - \frac{m+1}{2n} \] where $\psi_0(x) = \Gamma'(x)/\Gamma(x)$ denotes the digamma function [1910.01199, 1707.09755, 1006.4091]. Under the Bures–Hall ensemble, which is relevant for certain random mixed-state protocols, the closed form is [2002.04085] \[ E_f[S_{vN}] = \psi_0(mn - m^2/2 + 1) - \psi_0(n + 1/2) \] with the parameterization specified by the Cauchy–Laguerre biorthogonal ensemble. ## 2. Large-Dimension Asymptotics and Universal Laws In the limit $m,n \to \infty$ with ratio $c=m/n$, the mean entropy admits an asymptotic expansion [1707.09755, 1006.4091] \[ \langle S_{vN} \rangle \approx \ln m - \frac{m}{2n} + o(1) \] For balanced bipartition ($m=n=N$), this specializes to \[ \langle S_{vN} \rangle \approx \ln N - \frac{1}{2} \] Random pure states in high dimensions thus yield an "almost maximally mixed" $A$, with subleading $O(1)$ deficit—the "Page curve"—and variance $O(1/N^2)$ [1006.4091]. For Rényi entropies $S_q = \frac{1}{1-q}\ln\mathrm{Tr}\rho_A^q$, the analogous formula at leading order reads [1906.01269]: \[ \langle S_q \rangle \simeq \ln N_A - \frac{1}{q-1}\ln\left[\frac{2^{2q}\Gamma(q+1/2)}{\sqrt{\pi}\,\Gamma(q+2)}\right] \] These results have been extended and confirmed across random matrix models [1906.01269], Coulomb gas techniques [1006.4091], and free-fermion/harmonic oscillator ensembles [1907.09806]. ## 3. Entropy Scaling Laws in Many-Body Quantum Systems - **Volume Law:** In quantum-chaotic many-body systems and generic eigenstates (random pure states), bipartite entanglement entropy scales as the subsystem size \[ S_A \propto L_A \] where $L_A$ is the number of sites or degrees of freedom in $A$ [1708.08453, 2101.05309, 1907.09806]. Subleading corrections typically scale as $\sqrt{L_A}$ for number-conserving sectors away from half filling [1708.08453], or as $O(1)$ for unconstrained random states. - **Universal average in free quadratic models:** Translationally invariant free quadratic Hamiltonians (including spin chains via Jordan–Wigner) yield a universal average over eigenstates [1907.09806] \[ \langle S_\ell \rangle_{\rm univ} = N\,s(x) + o(N) \] where $x = \ell/N$, and $s(x) = -x\ln x - (1-x)\ln(1-x)$ is the binary entropy. - **Quantum-chaotic quadratic Hamiltonians:** In random quadratic models with RMT-type single-particle statistics, the leading entropy is independent of particle-number conservation [2101.05309]: \[ \overline{S} = \left(1 - \frac{1 + \frac{1-f}{f}\ln(1-f)}{\ln2}\right) L_A\ln2 \] with $f=L_A/L$ the subsystem fraction. ## 4. Entanglement Entropy under Symmetry Constraints Average bipartite entanglement entropy in the presence of non-Abelian constraints (e.g., global SU(2) conservation) exhibits modified scaling [2512.22942]: \[ \langle S_A \rangle_j = V f\,\beta(j) + \tfrac12 \log V + C_0(f, j) + o(1) \] where $j = 2J/V$ is the normalized global spin, $\beta(j) = -\tfrac{1+j}{2}\ln\tfrac{1+j}{2} - \tfrac{1-j}{2}\ln\tfrac{1-j}{2}$, and $C_0(f,j)$ encodes symmetry-dependent constants. The subleading $\frac12\log V$ term reflects the scaling of Clebsch–Gordan coefficients, and the O(1) constant encodes angular-momentum block effects. ## 5. Critical Phenomena and Spectral Phase Diagrams The entanglement spectrum and average entropy admit sharp phase transitions in the thermodynamic limit [1906.01269, 1006.4091]: - **Concentration line ($u_C(q)$):** Onset of eigenvalue density at zero; third-order transition for $q \neq 1$, fourth-order at $q=1$. - **Evaporation line ($u_E(q)$):** Detachment of the largest eigenvalue ("macroscopic occupation"); first-order for $q \neq 1$, second-order for $q=1$. States below $u_C$ reside in the entangled (semicircle-law-like) phase; $u_C