VULCAIN (v5.1): Hybrid Architecture for O(1)-Memory Exact Algebra and Certified Polynomial Factorization
This paper presents the unified software and mathematical architecture VULCAIN (v5.1), coupling three complementary engines:1. The Discrete Anvil (La Machine v16.0): Constant-memory O(1) deterministic isolation (< 2 KB in L1 cache) of certified prime moduli via primorial mirror wheels (V4 = 2,310), division-free Montgomery Number Theoretic Transforms (NTT), Chinese Remainder Theorem (CRT), and formal verification in Lean 4. 2. Continuous Quadratic Annealing (MBC-Annealing v5): Simultaneous factorization of real polynomials into quadratic blocks via spectral regularization on the positive semi-definite cone and Error-Free Transformations (EFT, 106-bit simulated double-double precision). 3. Vectorized Global Refinement: A posteriori Smale-Kantorovich certification (explicit curvature bound and α < 3 − 2√2 criterion) followed by an overflow-safe vectorized Aberth-Ehrlich refinement pass achieving IEEE 754 machine precision (< 10⁻¹⁵). The appendix includes the complete, self-contained Python implementation (vulcain_v5_complete.py).
Authors
- Mohamed Ghali
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-17
- DOI
- https://doi.org/10.5281/zenodo.24311056
- Primary Topic
- Polynomial and algebraic computation
- Type
- preprint