Does marking every node prove the prime-gap result?
The Polymath paper establishes an unconditional bounded-gap result of at most 246. This page offers a finite-prime picture and a scripted checklist; its buttons do not run Lean. The default displayed tuple fails admissibility modulo 5, and its ratio is an authored arithmetic score rather than a computed sieve integral.
The diagram and exported checklist are not a formal certificate. Changing sliders does not prove a new infinite-prime theorem. The bound 246 does not mean every consecutive prime gap is at most 246 and does not establish the twin-prime conjecture.
Take five offsets from the displayed tuple: 0, 6, 12, 28 and 4. Their residues modulo 5 are 0, 1, 2, 3 and 4. Consequently, for any shift n, at least one shifted offset is divisible by 5. This covers all residue classes and disproves admissibility of this supplied tuple. The counterexample concerns the page fixture; it does not refute the published bounded-gap theorem.
The paper defines H1 as the limit inferior of consecutive prime gaps and obtains H1 at most 246 unconditionally. This concerns gaps occurring arbitrarily far along the primes, not a uniform ceiling on every gap. Its separate conditional bound must retain its stated conjectural assumption. Polymath: variants of the Selberg sieve
At k=50 and m=5, the page formula gives 1.9 + 0.184 + 0.05 = 2.134. This is not the static fixture’s 2.084 and is not an evaluation of the variational integrals. Auto-Verify sets a counter to 12, Next increments it and Reset sets it to zero. The green node state records that counter alone.
Lean’s official documentation explains that successful elaboration and kernel acceptance validate a proof relative to its definitions, imports and axioms. The theorem’s intended meaning still requires review. A displayed code block or a JavaScript step counter is not this evidence; this page supplies no verified compiler run. Lean: validating a proof
Published unconditional H1 bound and explicitly conditional improvements.
Explains elaboration, kernel acceptance and interpretation; does not certify this page.