Plectis
Erdős problem #68

The factorial-denominator series

Is the series sum_{n >= 2} 1/(n! - 1) irrational?

Open 12 Lean modules 2 main results Source directory

What is checked in Lean

  • The exact integral normal form: the series is irrational if and only if the strict factorial successor Z_m misses divisibility by m cofinally.
  • The equivalent carry form: rationality is equivalent to eventual unit carry, so cofinally many non-unit carries imply irrationality.
  • Canonical factorial digits, exact remainder recurrences, zero-tail propagation, and the divisor-factorial centre recurrence.
  • Integral channel congruences, the lcm tax, the prime unit translator, and factorial-grid Cramer residual identities.
  • Weighted projection and factor-split rigidity consumers with their exact modular and Archimedean hypotheses.
  • The zero-plateau and first-exit carry dichotomy, prime-power prefix obstructions, and exact doubled-prime branch characterisations.
  • A Lean-checked implication from one exact non-unit carry to a rational-denominator lower bound; an independently regenerated exact GMP certificate at m = 300000 supplies the finite input.

Main results

Statements and boundaries are copied from the corpus record. Registry status: not_registered; the claim registry does not carry these declarations and kernel checking them does not promote them into reviewed public claims

What is not checked

  • Irrationality of the factorial-denominator series.
  • A theorem producing cofinally many divisibility misses or non-unit carries.
  • Cofinal strict nonintegrality of the factorial-grid Cramer residual.
  • The weighted collision-product and complementary-residue estimates needed by the factor-split projection consumer.
  • The GMP computation is an exact finite external certificate, not a Lean evaluation and not an unbounded argument.

Open obligations

Prove any one of these and the corresponding reduction above becomes unconditional.

  • weighted_collision_and_complementary_residue On an unbounded family of tailored prime blocks, prove both the weighted collision-core product bound and the independent complementary-residue inequality consumed by the exact unit-factor scale split.
  • cofinal_prime_power_amplification At infinitely many genuinely nonterminal indices, prove quantitative repeated-support valuation amplification strong enough to meet the checked endpoint inequality.
  • cofinal_lower_endpoint_escape For infinitely many primes, force the reduced predecessor gap outside the exact lower unit-carry cylinder.
  • cofinal_doubled_prime_branch_failure For infinitely many odd primes, rule out both exact p^2-divisibility branches at index 2p by coupling the carry value to the predecessor residue.
  • cramer_residual_nonintegrality Construct an unbounded factorial-grid family whose exact Cramer residual is nonintegral, using a determinant, valuation, cancellation, or gcd-of-minors certificate that controls the finite sign-changing block.

Lean modules

Comparator review

The Comparator replays selected exact interfaces against a pinned toolchain. It never widens what Lean proved.

Replayed by the Comparator

Statement and type parity replayed against the pinned toolchain.

  • Exact equivalence between irrationality, cofinally many non-unit carries, and cofinal strict-successor divisibility failures.

    An exact reformulation does not supply the required cofinal failures.

    factorial_carry_characterisation · targeted

Represented indirectly

Covered through a neighbouring selected interface, not replayed one-to-one.

  • Finite channel congruences, a two-term prime-channel corrector, and endpoint-weighted projection rigidity.

    These finite and structural results do not produce a cofinal obstruction.

    factorial_channel_and_projection_rigidity · represented_by_isolated_headline

Outside the Comparator selection

Stated in the corpus; not part of the replayed interface set.

  • Cofinal zero-branch or large private-factor hypotheses imply the required irrationality conclusion.

    The producer hypotheses are unproved.

    factorial_conditional_producers · not_selected_deep_project_predicate_stack

  • Lean checks small exact misses; a hash-bound GMP scan reaches index 300000 and yields the corresponding finite denominator exclusion.

    Finite computation does not change the cofinal quantifier.

    factorial_finite_certificates · not_applicable_to_external_execution

  • A paper deduction gives a lower bound for the lcm of factorial-gap denominators using an external multiplicity theorem.

    Comparator cannot certify the cited input or authored deduction.

    factorial_lcm_growth · not_applicable_not_a_lean_declaration

The paper

Problem note 20 pp

Factorial Carries and Finite Channel Obstructions

For the factorial-denominator series, Lean gives an exact factorial-successor normal form and the equivalent eventual-unit-carry criterion, together with integral-channel, prime-translator, projection-rigidity, and Cramer-residual consumers. An independently regenerated exact finite certificate reaches index 300000, and the checked consumer excludes every smaller positive rational denominator. Five quantified producer problems isolate the missing cofinal step; none is proved, and Erdős #68 remains open.