Multiplicity of a subnormal uniform cover is at least the least prime
ProvedSubnormalCover.multiplicity_ge_minFacMain theorem (Sun, Theorem 4.3(i)). Let be a nontrivial uniform cover of an arbitrary group (each lies in exactly of the cosets, and some ) by subnormal subgroups of finite index. Put and , and let and be the smallest and largest prime divisors of . Then among the indices there is a multiple of occurring at least times: there exists with and . In particular the maximal multiplicity satisfies , which is the quantitative form of Theorem 1.1 and answers Erd\H{o}s' open question negatively for subnormal covers.
Formalization note. is coverLcm K (the lcm of the indices), is BFFPyramidal.maxPrimeFac N and is Nat.minFac N; the hypotheses hcov and hnt already force and , so is a genuine prime (not the junk value ) and no separate assumption is needed.
import Mathlib import Definitions.Def_SubnormalCover_uniformCover import Definitions.Def_BFFPyramidal_arithmetic
namespace SubnormalCover
open BFFPyramidal
theorem multiplicity_ge_minFac {G : Type*} [Group G] {k : ℕ}
(K : Fin k → Subgroup G) (a : Fin k → G) (w : ℕ) (hcov : IsUniformCover K a w)
(hfin : ∀ i, (K i).FiniteIndex) (hsn : ∀ i, (K i).IsSubnormal) (hnt : ∃ i, K i ≠ ⊤) :
∃ j : Fin k, maxPrimeFac (coverLcm K) ∣ (K j).index ∧
Nat.minFac (coverLcm K) ≤ indexMultiplicity K ((K j).index) := by
sorry
end SubnormalCoverRead-back
What the Lean code literally says, in plain math · claude-opus-5
Setting. Let be an arbitrary group and an implicit natural number. We are given subgroups indexed by (repetitions permitted), elements , a natural number , and the hypotheses:
- the system of left cosets is a uniform cover of weight : for every , the number of indices with (equivalently ) is exactly ;
- every has finite index: , where is the number of left cosets , equal to by convention when infinite;
- every is subnormal in , i.e. sits in a finite chain of successively normal subgroups;
- there exists an index with ; this forces , hence (with ) , and forces .
Conclusion. Write for the least common multiple of all indices (the Finset lcm, which would be on an empty index family), for the greatest prime factor of taken as the supremum of 's prime-factor set, with since that set is empty, and for the number of indices taking the value . Then there exists (existence only, no uniqueness, and need not be one of the indices witnessing ) such that simultaneously:
- (the greatest prime factor of the lcm of the indices divides the -th index), and
where is Mathlib's least prime factor of , with the conventions and ; the right-hand side counts all indices (including ) with , so it is at least .
Edge behaviour. The inequality is non-strict () and points from the least prime factor of the lcm up to a multiplicity. If were , clause 1 would demand , i.e. , contradicting ; but already rules out , so is a genuine prime . The case cannot occur, since is unsatisfiable over an empty index type. The weight , the elements and the subnormality hypothesis appear only among the hypotheses, not in the conclusion.
Confirmed by the mission captain (proposal self-audit).