Research notes on every problem, and a library of the papers behind them. Built from the open erdos repository.
Updated
Source. Liu--Sawhney, On further questions regarding unit fractions, arXiv:2404.07113v1, Lemma 2.4, p. 8; see the source digest.
Statement. Fix a sufficiently large (the print leaves this largeness implicit) and an integer interval of length . Suppose is a set of primes none of which exceeds
Put and . Then, with absolute comparison constants,
External input. The fundamental lemma of sieve theory in Koukoulopoulos, The Distribution of Prime Numbers, Graduate Studies in Mathematics 203, AMS (2019), Theorem 18.11(b), is the cited input. The author-hosted preliminary version has Theorem 18.11(b) on printed p. 190/PDF p. 201. The paper invokes its dimension-one case with sieve level and parameter . The external theorem itself is not proved here.
Proof. Write . For each , counting multiples in an interval gives
This is the first sieve axiom in the cited theorem, with density . The second, dimension-one sieve axiom holds with an absolute constant: products of over subsets of primes are bounded by the corresponding products over all primes, which have the usual dimension-one prime-product bound. The third axiom, controlling the weighted remainder sum at level , follows from
Here is the divisor function; the inequality follows, for example, by counting pairs of positive integers with product at most . The ratio of the logarithm of the sieve level to the logarithm of the largest allowed sieving prime is
The fundamental lemma therefore gives the first comparison in the statement. These are exactly the parameters and three axiom checks used by the source.
Finally, uniformly for ,
Since , summing over and exponentiating gives
which proves the second comparison.
Dependencies. The external fundamental lemma cited above, elementary interval counting, and standard dimension-one prime-product estimates (the paper groups the latter with its number-theoretic preliminaries, Theorem 2.1).
Bears on. #298 and #299, through the sieve estimates in the quantitative reciprocal-sum argument.