Research paperAcceptedARR-2026-6FDEKPVJ0W8BHBMC · v1 · 2026-08-13

A Finite-Dimensional Nonanalytic Spectral Transition and Exact High-Fidelity Rate–Distortion for Rank-r Born Prediction on Complex Grassmannians

Lluis Eriksson

Abstract

Let P be Haar on the complex Grassmannian Gr(r,d) and set rho_P=P/r. We solve two coupled finite-dimensional problems. First, on the sphere of traceless Hermitian external fields with fixed Frobenius norm, we determine incompatible weak- and strong-field optimizers of the matrix-Bingham free energy log E exp{s tr(AP)}. For 1<r<d/2, an exact cubic Haar moment and a uniform C2 perturbation theorem make the positive one-spike spectrum uniquely optimal at weak field. A parameter-uniform differentiated Grassmann Laplace expansion upgrades the strong-field Ky Fan limit to exact eventual uniqueness of the rank-r two-block spectrum. The globally optimized free energy must therefore fail to be real analytic at a finite field; at its first exit from the one-spike branch there is either nonconjugate coexistence or transverse-Hessian degeneracy. The balanced case r=d/2 is selected by a negative quartic coefficient instead. Second, we determine the unrestricted classical Shannon rate–distortion function for squared-Frobenius reconstruction of rho_P on a nonempty open interval adjacent to zero distortion. Arbitrary standard-Borel classical memories and arbitrary reports reduce by conditional least squares to the full posterior-density body, not merely to the source Grassmannian. Exact eventual two-block optimality and radial localization reduce the complete Shannon dual to one variable. The frontier is attained by a covariant matrix-Bingham channel whose posterior mean is a depolarized rank-r projector. A Jacobi–Selberg calculation gives the exact high-rate constant, and complementation transfers the result to every nontrivial rank. The complete intermediate-phase classification and the full all-distortion rank-r curve remain open.

Deposit recordedFirst repository commit · full Git SHA recorded
PublishedGitHub release · tag ARR-2026-6FDEKPVJ0W8BHBMC-v1
Canonical PDF downloads3GitHub release downloads only; direct ARR PDF reads are not measured
Page viewsNot measuredPage views are not measured until ARR connects a privacy-reviewed, no-cookie analytics source.
Definitions and rankings →
Not yet rated

Verification record

Frontier-model screening
Not assessed
Source integrity
Pass
Bibliographic integrity
Not assessed
Reproducibility
Partial
Lean 4
Not applicable

Recorded under ARR-VERIFY-1.0. ARR verification and screening are not peer review.

Version history

The ARR identifier remains stable. Each version has its own immutable release, timestamp and version identifier.

  • v1 · source snapshot available · viewing

AI assistance statement

OpenAI Codex assisted with symbolic checks, literature triage, and adversarial proof review. The author retains responsibility for every claim.

Frontier-model screening

Status: not_assessed. Any listed reports correspond to this exact version under ARR-SCREEN-1.0; no absent assessment is represented as a pass.

    Longitudinal frontier-model record

    Independent model assessments

    Read the scale and limits
    Not yet rated

    No eligible independent ARR-ASSESS-1.0 report is published for this exact version. Missing evidence is not scored as zero.

    No model reports are published for this version.

    A model assessment is not peer review or a correctness certificate. ARR preserves disagreement, exact-version provenance and later reassessments.

    Editorial disclosure

    Founder-owned pilot record: Lluis Eriksson is both the author and ARR's current founder-editor. No independent editorial review, peer review, or frontier-model screening is claimed for this version. Acceptance here records a technically valid deposit, not a finding that the paper is correct.