Universal Spectral-Curvature Phase Diagrams of Graphs: Predicting Non-Euclidean Embedding Distortion and Optimal GNN Geometry via Gram-Eigenvalue Decomposition
Universal two-dimensional spectral-curvature phase diagram (g3, d_eff+) derived from the positive and negative eigenspaces of double-centered squared-distance Gram matrices. Proves the nuclear-norm range identity g3 in [0, 1), the exact diameter-2 complement-spectrum and random-graph asymptotic laws, and structural failure modes of Gromov delta-hyperbolicity, validated across 20 canonical manifolds and 3,246 real-world GNN benchmark graphs.
Authors
- Shoji Fujita
Publication Details
- Journal
- Figshare
- Published
- 2026-11-20
- DOI
- https://doi.org/10.6084/m9.figshare.35069164
- Primary Topic
- Advanced Graph Neural Networks
- Type
- preprint