Status: complete written disproof; the full counterexample, exact sparse infimum and open-family gap are Lean-checked. The exact infimum for each assignment is also formalized. The standalone source, all-owned axiom audit and stock-kernel replay passed; the merged-package gate is pending.
The Lyapunov positivity conjecture in arXiv:2406.17196v2, Conjecture 1, is false. The proof constructs a rational feasible encoder of power , while every encoder with exactly one nonzero entry per column has infimal power at least . The strict gap is .
All four possible source-to-channel assignments are treated with arbitrary nonzero encoder coefficients. A symbolic calculation also gives an open interval of channel gains with the same failure. The argument refutes the full universal matrix conjecture and does not depend on a numerical optimization, a sampled parameter range, or failure of an infimum to be attained.
The claim concerns the stated matrix optimization. It does not assert a separate resolution of the general communication-code stabilization problem. These notes remain unrefereed.
Files
- PROOF.md: complete disproof and open family.
- SOURCES.md: original statement and bounded prior-art audit.
- ALGEBRA_REVIEW.md: independent source and mathematical review.
- ANALYSIS_REVIEW.md: second independent review.
- verify.py: exact rational and symbolic diagnostics.
Run python3 verify.py with SymPy installed. The shared Lean module checks the full matrix counterexample, all four individual assignment infima and the open-family uniform gap.
Research coordination: issue 197.