Research notes on every problem, and a library of the papers behind them. Built from the open erdos repository.
Updated
This is the complete author-recorded proof of
L220.
Fix y,M,ε as in its statement. Let H consist of the
positive integers all of whose prime factors are at most y, including
1, and put P=∏p≤yp. Every positive integer has a unique
factorization n=sk with s∈H and gcd(k,P)=1.
Choose distinct primes r0,t0>y and put Q=r0t0P. The progression
e≡r0t0(modQ) consists of integers coprime to P and divisible
by both r0,t0. In particular none is 1 or a prime power. Write
g(e)=⌊41log2e⌋. Choose e1 in this progression
sufficiently large that, for every e≥e1,
g(e)≥2Q,23Me1/4≤εe/2.(1)
Set ej+1=ej+Q⌊g(ej)/Q⌋ and E={e1,e2,…}.
All these integers remain in the progression, and their gaps satisfy
g(ej)/2≤ej+1−ej≤g(ej). Thus
E(X):=#(E∩[1,X])=O(X/log(X+2)).(2)
For completeness, elements up to X contribute at most X;
above X their gaps are at least a fixed positive multiple of
logX, once X is sufficiently large. Enlarging the constant covers
the bounded range.
The entire binary weight of the orbit with rough part e is
Ae=s∈H∑φ(s)2−se.
For e≥2, positivity, the term s=1 and φ(s)≤s give
2−e≤Ae≤(1−2−e)22−e<21−e.(3)
Put d=MAe1/2. For any real r with ∣r−S∣<d, start with
η0=r−S, and define, with nearest-integer rounding,
δj=M⌊MAejηj−1⌉,ηj=ηj−1−δjAej.(4)
Then δ1=0, every δj is divisible by M, and
∣ηj∣≤MAej/2. For j≥2, (3) and the gap bound give
The last elementary bound follows by taking the product over pa∥n
in φ(n)2/n: the factor pa−2(p−1)2 is at least one for odd
p, and the factor for p=2, when present, is at least 1/2.
Define h(ej)=φ(ej)+δj and h(k)=φ(k) at all other
integers coprime to P. For the unique decomposition n=sk above, set
f(n)=φ(s)h(k).(6)
Equation (5), including δ1=0, proves positivity, integrality and
the asserted relative bounds. It also gives f(n)≡φ(n)(modM).
Every prime power has rough part either 1 or a prime power, hence outside
E, so its value is unchanged. Multiplication by a prime p≤y changes
only the smooth part s. The two required laws for f(pn) therefore
follow from the corresponding exact laws for φ(s).
The relative bound implies absolute convergence. The orbits with distinct
rough parts are disjoint, so absolute convergence permits regrouping:
n≥1∑(f(n)−φ(n))2−n=j≥1∑δjAej=r−S.(7)
The last equality follows by telescoping (4), since (3) implies
ηj→0. All orbit terms were included in Ae; there is no omitted
contribution from multiples of the changed integers.
Finally the changed set is contained in {se:s∈H,e∈E}.
For s≤X, (2) bounds its count by
O(X∑s∈Hs−1/logX). For s>X, the crude
bound E(X/s)≤X/s gives at most
Xs∈Hs>X∑s1≤X3/4s∈H∑s−1/2=O(X3/4).
Both sums over H converge, being finite Euler products
∏p≤y(1−p−α)−1 for α=1,1/2. This proves the
claimed O(X/logX) bound and completes the proof. □