arXiv:2609.38815
Relative entropy of entanglement and tripartite minimal surface
Takato Mori, Beni Yoshida
Abstract
We propose a holographic dual of the relative entropy of entanglement for a boundary tripartition $A:B:C$. Our proposal is that, at the leading order, \begin{align} E_R(ρ_{AB}) = \frac{1}{4G_N}\Big[\mathrm{Area}(Γ_{\min}) - \mathrm{Area}(γ_{AB})\Big] \notag \end{align} where $Γ_{\min}$ is the minimal bulk tripartition surface and $γ_{AB}$ is the minimal surface homologous to $AB$. For a general random tensor network, we prove the corresponding upper bound by constructing a separable state associated with $Γ_{\min}$ and evaluating its relative entropy. We also establish the matching lower bound rigorously for networks of one, two, and three Haar random tensors by bounding the maximal tripartite product-state overlap. In appropriate geometries, the minimal tripartition surface can form a nontrivial tri-junction resembling the Mercedes logo. The proof proceeds by sequential optimization over candidate product states, which has a natural geometric interpretation as a local search for the minimal tripartition surface.
AI Summary
The paper asks whether the relative entropy of entanglement (REE), defined by minimizing relative entropy over separable states, has a geometric holographic dual. For a pure boundary state partitioned as $A:B:C$, it proposes $E_R(\rho_{A:B})=[\mathrm{Area}(\Gamma_{\min})-\mathrm{Area}(\gamma_{AB})]/(4G_N)+O(1)$, where $\Gamma_{\min}$ separates all three boundary regions and $\gamma_{AB}$ is the RT surface for $AB$. For fixed-graph Haar random tensor networks with exponentially large bond dimensions, it establishes the corresponding general upper bound, measuring cut areas as sums of $\log_2 D_e$. A candidate separable state is constructed by dephasing bonds across the tripartition. Its relative entropy is evaluated through a replica model reduced to binary Ising and OR interactions; an operator inequality controls the replica limit at $q=1$. Matching lower bounds follow from $E_R\geq-S(\rho_{AB})-\log_2\Lambda$, where $\Lambda$ is the maximal tripartite product overlap. Sequential Gaussian concentration and $\epsilon$-net arguments establish $-\log_2\Lambda=\mathrm{Area}(\Gamma_{\min})+O(1)$ for the one- and two-tensor models and a triangular three-tensor model in its internal-cut, or Mercedes, regime, with vanishing failure probability. A continuum AdS$_3$ illustration predicts distinct wedge-connectivity, tripartition, and cross-section transitions at interval angles approximately $1.57$, $2.06$, and $2.29$. The continuum dual and matching lower bound for arbitrary networks remain conjectural. Stated obstacles include organizing optimization orders, defining continuum separability, and identifying a gravitational construction and its backreaction; the analysis also assumes a unique minimal tripartition.
We remark that the key idea of the proof was generated by GPT-5.6 Sol after multiple iterations.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 53 pages, many figures