NP-Hardness Collapsed: Deterministic Resolution of Spin-Glass Ground States via Information-Geometric Manifolds (Scaling from N=8 to N=100)

V3: This version corrects two numerical errors in the N=8 ferromagnetic reference table (Section VII.1B): the spurious energy level E=+8 (14 configurations) has been removed and the E=+4 entry corrected to 70 configurations, consistent with the binomial coefficient C(8,4). Additionally, a textual clarification in Section VII.1E now correctly states that the theoretical minimum equals the negative number of pairwise interactions. None of these corrections affect the spin-glass verification results (Section VII.2), the GCIS ground-state solutions, or any conclusions of the paper. All core results remain unchanged. _________________________ V2: This version includes Appendix A, which was not part of earlier publicly visible versions of the preprint.The appendix provides independent, architecture-agnostic validation of the reported information-geometric collapse mechanism. _________________________ V1: This preprint presents the first large-scale empirical demonstration of a deterministic information-geometric collapse mechanism capable of resolving NP-hard Spin-Glass ground states across the range N=8 to N=100. The GCIS-based I-GCO architecture operates without weights, training, stochasticity, or iterative optimization. Instead, it collapses high-dimensional state spaces onto a one-dimensional manifold while retaining full informational structure across all layers. Exhaustive verification up to N=24 confirms exact ground-state solutions; larger systems show invariant collapse geometry, complete information preservation, and stable correlation symmetries, suggesting scale-free behavior. _________________________ Beyond Spin-Glass systems, the results indicate that informational-geometric manifold collapse may represent a computational modality fundamentally distinct from algorithmic search. The mechanism may generalize to broader optimization, inference, and constraint-satisfaction domains. Ongoing work aims to scale the architecture further and apply it to real-world industrial and scientific problem classes where classical algorithms or quantum approaches are limited. Key Contributions: Deterministic resolution of NP-hard Spin-Glass ground states for N=8 to N=100. Full information retention across 100 layers with non-local correlation symmetry. Evidence for geometric collapse onto a one-dimensional attractor manifold. Empirical behavior incompatible with classical algorithmic or probabilistic methods. Clear pathway toward scaling and applying the mechanism to practical large-scale problems. The findings motivate further exploration of information-geometric computation as a potential foundation for new classes of efficient, deterministic problem-solving systems. Updated nomenclature regarding the Hamiltonian formalism in Section VI to explicitly distinguish between ferromagnetic baselines and frustrated Spin-Glass configurations. This version supersedes the initial preprint previously accessible under DOI 10.5281/zenodo.17782987.