Wiki
Wiki

Research notes on every problem, and a library of the papers behind them. Built from the open erdos repository.

Updated


Statement

For integers n≥2kn\geq2k and k≥59k\geq59,

nπ(n)−π(n−k)<exp⁡(nlog⁡n+k+k2log⁡n).(7)n^{\pi(n)-\pi(n-k)} < \exp\left(\frac{n}{\log n}+k+\frac{k}{2\log n}\right). \tag{7}

Ecklund prints the lemma with the single hypothesis k≥59k\geq59 (pp.267--268). The condition n≥2kn\geq2k is the range of his theorem, in which he applies the lemma, and the proof below uses it to place n−kn-k in the domain of (1).

Proof

Use the upper estimate (2) at nn and the lower estimate (1) at n−kn-k from [[factorials_binomials/ecklundjr_1969_prime_divisors_binomial_coefficient/external_inputs|the external-input record]]. Their domains hold because n−k≥k≥59n-k\geq k\geq59. Thus

π(n)−π(n−k)<nlog⁡n(1+32log⁡n)−n−klog⁡(n−k)(1+12log⁡(n−k)).\begin{aligned} \pi(n)-\pi(n-k) <&\frac{n}{\log n}\left(1+\frac{3}{2\log n}\right)\\ &-\frac{n-k}{\log(n-k)} \left(1+\frac{1}{2\log(n-k)}\right). \end{aligned}

Since n−k≤nn-k\leq n, replacing both occurrences of log⁡(n−k)\log(n-k) in the subtracted positive term by the larger log⁡n\log n only increases the right-hand side. Multiplication by log⁡n\log n therefore gives

(π(n)−π(n−k))log⁡n<n(1+32log⁡n)−(n−k)(1+12log⁡n)=nlog⁡n+k+k2log⁡n.\begin{aligned} \bigl(\pi(n)-\pi(n-k)\bigr)\log n &< n\left(1+\frac{3}{2\log n}\right) -(n-k)\left(1+\frac{1}{2\log n}\right)\\ &=\frac{n}{\log n}+k+\frac{k}{2\log n}. \end{aligned}

Exponentiating proves (7).

Verification record

Current review state. Accepted by independent mathematical review (retained as the full-proof review), relative to the two premises named below. Substantive changes to this proof or those premises invalidate the affected scope until rechecked.

Scope and source version. The checked scope is equation (7), both strict inequalities, the threshold k≥59k\geq59, the domains at nn and n−kn-k, and the logarithm replacement. The statement and proof are on printed pp.267--268 / physical pp.2--3 of Ecklund's Pacific Journal of Mathematics 29 (1969), 267--270 publisher PDF, identified on the source card.

Premises and limits. The component assumes exactly Rosser--Schoenfeld estimates (1) and (2) as quoted in Ecklund. Their proofs were not recursively reconstructed or reviewed. No gap remains in the deduction from those premises. No formal verification is recorded.