Claude Fable produced a quadratic-query algorithm for a hidden cycle problem
Source check: Source checked. Not independent scientific confirmation.
Finding: Claude Fable 5.1 designed a solver for CycleUncover, a black-box problem that hides an even cycle in a bitstring fitness function. The solver reconstructs a two-coloring from fitness queries. Lean checked the expected O(n²) query bound, while the paper proves an exponential bound for the classical (1+1) evolutionary algorithm on this problem.
Technology: Claude Fable 5.1; staged algorithm design and Lean proof generation
Family Claude: profile · models page
Publication: Preprint · Test context: prospective or independent check. Catalog categories, not an evidence grade.
- AI contribution
- Claude Fable proposed the oracle-query solver and produced the formal expected-query proof under the supplied framework.
- Test and result
- The authors defined the hidden-cycle oracle and supplied a staged formal framework. Their report gives a Lean-checked bound for the generated algorithm and a separate lower-bound argument for the comparator.
- Limitation / still unknown
- This result concerns one constructed oracle family. It neither establishes that the bound is optimal nor demonstrates empirical speed or general maximum independent-set efficiency.
Evidence sources and annotation context
Evidence context (descriptions follow the cited sources; test type is a catalog classification):
Evidence sources: https://arxiv.org/abs/2610.35595v1 · https://arxiv.org/html/2610.35595v1
Source checked · editorial details
The source page was checked against this entry. Independent confirmation is not assessed by this label.
Source type: paper. Paper: linked separately.
Review date: 21 Nov 2026. Review scope: §5.2/Theorem 5.1, Table 2, Appendices E and F.7 and front matter.
No follow-up recorded; this is not a claim that no follow-up exists. Follow-up review scope and date are not recorded. No source-supported relationships recorded.
Editorial assessment: Impact 2/5 · Validation 4/5 · Novelty 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.