Every `t ≥ 2` below `C(22,11) = 705432` whose multiplicity is odd is one of
ProvedSingmaster.odd_mult_lt_of_ltaether-catalogcombinatorics
Every t ≥ 2 below C(22,11) = 705432 whose multiplicity is odd is one of
2, 6, 20, 70, 252, 924, 3432, 12870, 48620, 184756, and its multiplicity is 1
(only for t = 2) or 3.
theorem Singmaster.odd_mult_lt_of_lt{t : ℕ} (ht : 2 ≤ t) (hlt : t < 705432) (hodd : Odd (mult t)) :
mult t = 1 ∨ mult t = 3 := by sorry
Formalization Note Transplanted verbatim from the Aether Catalog source Combinatorics/SingmasterCentralBinomial.lean; the statement is byte-identical to the source declaration, elaborated with autoImplicit disabled in the platform environment.
Preamble
-- Thm stub generated from Combinatorics/SingmasterCentralBinomial.lean import Mathlib import Definitions.Def_Combinatorics_SingmasterCentralBinomial import Definitions.Def_Combinatorics_SingmasterOccurrences import Definitions.Def_Combinatorics_SingmasterParity import Definitions.Def_Combinatorics_SingmasterRefinements /- # Central binomial coefficients occur exactly three times (verified initial segment) Fourth research cycle, building on * `Combinatorics.SingmasterOccurrences` (the occurrence set `occ`, the multiplicity `mult`, monotonicity and growth of binomial coefficients), * `Combinatorics.SingmasterRefinements` (strict unimodality of a row), * `Combinatorics.SingmasterParity` (the parity criterion: `N(t)` is odd iff `t` is a central binomial coefficient). The parity criterion reduced the open question "does any number occur exactly five or exactly seven times?" to the single sequence `C(2m,m) = 2, 6, 20, 70, 252, 924, …`. This file makes that reduction *effective* and then executes it for `m ≤ 10`. ## The effective criterion For `t = C(2m,m)` with `m ≥ 2` we prove a **sandwich theorem** (`Singmaster.choose_lt_centralBinom`): every entry `C(n,k)` with `n ≤ 2m` other than the central entry itself is *strictly smaller* than `t`. Hence any further occurrence of `t` lies in a row `n > 2m`, and being an interior entry it satisfies `C(n,2) ≤ C(n,k) = t`, which caps `n` by an explicit `N` with `t < C(N,2)`. So the multiplicity of `C(2m,m)` is exactly three as soon as a *finite, explicitly bounded* search over `2m < n < N`, `2 ≤ k ≤ n/2` finds no further occurrence (`Singmaster.mult_centralBinom_eq_three`). The search is phrased with `Nat.descFactorial` rather than `Nat.choose`, which is what makes it feasible for the kernel: `C(n,k) = t` is equivalent to `n.descFactorial k = k ! * t`, and the descending factorial costs `k` multiplications instead of `C(n,k)` additions. ## Results * `Singmaster.mult_centralBinom_eq_three` — the effective criterion; * `Singmaster.mult_centralBinom_eq_three_of_le_ten` — `N(C(2m,m)) = 3` for `2 ≤ m ≤ 10`, i.e. for `6, 20, 70, 252, 924, 3432, 12870, 48620, 184756`; * `Singmaster.mult_ne_five_or_seven_of_lt` — **unconditionally, no `t < 705432` occurs exactly five or exactly seven times**. This is the machine-checked version of the empirical observation quoted in the problem statement, and it is obtained from only ten finite searches rather than from a scan of all `t`; * `Singmaster.odd_mult_lt_of_lt` — below `705432` an odd multiplicity is `1` or `3`. -/ open Finset open Singmaster /-! ## Basic size estimates for `C(2m,m)` -/ /-! ## The sandwich theorem -/ /-! ## The effective criterion -/ /-! ## Executing the criterion for `m ≤ 10` Each of the following is a genuine finite search over the explicitly bounded window produced by the criterion; the windows are empty for `m = 2, 3` and grow to `20 < n < 609` for `m = 10`. -/ /-! ## Unconditional consequences below `C(22,11) = 705432` -/ set_option maxHeartbeats 1000000 in set_option maxRecDepth 1000000 in
Formal statement
theorem Singmaster.odd_mult_lt_of_lt{t : ℕ} (ht : 2 ≤ t) (hlt : t < 705432) (hodd : Odd (mult t)) :
mult t = 1 ∨ mult t = 3 := by sorrySource