arXiv:2610.34364
Exact Identification of Quantum Correlations in the Minimal Bell Scenario
Liang-Liang Sun, Jie Wang, Sixia Yu
Abstract
Bell nonlocality places the long-standing philosophical disputes over the nature of quantum theory on a mathematically rigorous, experimentally testable footing, positioning quantum mechanics between local realism and the most general no-signaling theories. Yet the precise quantum boundary has remained elusive even in the minimal Bell scenario, where the extreme points generating the set are known. Here we characterize this boundary through a computable necessary and sufficient criterion: a correlation is quantum if and only if it can be decomposed into an extreme quantum correlation and a classical correlation. The criterion follows from our proof that every Bell functional exhibiting quantum advantage admits a unique quantum maximizer. We further classify the finitely many maximal-face structures containing each nonlocal exposed point according to their local deterministic vertices. By sharply distinguishing physically realizable correlations from nonphysical ones, these results resolve a long-standing open problem and provide a new foundation for identifying the principles underlying quantum theory and determining the ultimate information-theoretic limits of quantum correlations.
AI Summary
The paper addresses exact membership and boundary geometry of the full quantum correlation set $\mathcal Q$ in the $(2,2,2)$ Bell scenario, including nonzero marginals. Its central result is that every quantum behavior has a decomposition $Q=\lambda Q_\theta+(1-\lambda)L$, where $Q_\theta$ comes from a pure real two-qubit state with real projective measurements, $L$ is local, and $0\leq\lambda\leq1$. This also gives a $6\otimes6$ realization bound. The main supporting theorem states that every Bell functional with strict quantum advantage has a unique maximizing behavior. The proof reduces optimization to two measurement overlaps and studies the determinant of the globally positive Bell-gap matrix. This determinant has separate degree at most two, with a negative highest mixed coefficient when the correlator coefficients have frustrated signs. Polynomial and operator arguments exclude multiple interior optima. A minimum-quantum-weight convex-geometric construction then yields the decomposition, extended to boundary points by compactness. For each nonlocal exposed point, supporting faces contain no other nonlocal extreme point; the remaining generators are local deterministic vertices. This gives finitely many possible face configurations, subject to global quantum feasibility tests. One certified example has four maximal triangular faces. Membership is reformulated as fixed-dimensional trigonometric or semialgebraic feasibility. Global certification remains necessary: local-search failure does not certify nonquantumness, and searching all two-local-vertex decompositions is not complete. Reported numerical timings concern tested inputs, not worst-case performance or comparisons with NPA.
Central claim may not hold: The boundary-limit proof establishes a pure-two-qubit component in S, but does not establish that this component is extreme as required by the abstract. The advertised complete weight-free test of eight inequalities in five angles is missing: its referenced quantities are undefined, and the supplied fixed-pair test is explicitly incomplete. The main text calls the example's normalized normal-cone section a pentagon, whereas the appendix certifies a quadrilateral for the same behavior.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.