Erdős Problem #137: term-level rough-part structure in a very bad interval #
If the block product F k n = n(n+1)⋯(n+k-1) is powerful, then for each individual term n+i,
the part supported on primes p ≥ k is itself powerful. A prime p ≥ k divides at most one of the
k block factors, so powerfulness of the whole product forces p^2 ∣ n+i. This is the local
anatomy behind Tao's "very bad interval" language. No abc, no radical lower bound, no analytic
number theory.
The part of m supported on primes < k.
Equations
- Erdos137.SmoothPartBelow k m = ∏ p ∈ m.primeFactors with p < k, p ^ m.factorization p
Instances For
The part of m supported on primes p ≥ k (written ¬ p < k to pair with
Finset.prod_filter_mul_prod_filter_not).
Equations
- Erdos137.RoughPartAbove k m = ∏ p ∈ m.primeFactors with ¬p < k, p ^ m.factorization p
Instances For
Term-level smooth/rough factorization: SmoothPartBelow k m * RoughPartAbove k m = m.
Factorization of the term-level rough part.
If a prime divides the rough part, it is a prime factor of m and is ≥ k.
Rough part of each term is powerful in a very bad interval.
Squarefree terms have trivial rough part #
A nonzero powerful divisor of a nonzero squarefree number is 1.
A small reusable helper: if d ∣ m, m is squarefree, and d is powerful, then every prime
valuation of d is both ≥ 2 if nonzero (powerfulness) and ≤ 1 (since d ∣ m squarefree), so
all valuations of d vanish.
Squarefree term corollary. In a powerful block, the k-rough part of a squarefree term is
trivial: if F k n is powerful and n+i is squarefree, then n+i has no prime factor p ≥ k, so
all of its prime support lies below k.