Conjecture on the variance of the normalized Euler totient function in intervals

Let R0(x):=nxφ(n)nxζ(2)R_0(x):=\sum_{n\leq x}\frac{\varphi(n)}{n}-\frac{x}{\zeta(2)} and, for an interval of size HH, let

R0(x;H):=R0(x+H)R0(x)=x<nx+Hφ(n)nHζ(2).R_0(x;H):=R_0(x+H)-R_0(x)=\sum_{x<n\leq x+H}\frac{\varphi(n)}{n}-\frac{H}{\zeta(2)}.

Let H=Θ(xδ)H=\Theta(x^\delta) for some fixed 0<δ10<\delta\leq 1. Variance conjecture. Then

1XxXR0(x,H)216ζ(2)16ζ(2)2.\frac{1}{X}\sum_{x\leq X}R_0(x,H)^2\sim \frac{1}{6\zeta(2)}-\frac{1}{6\zeta(2)^2}.

This predicts the discrete mean square, and hence the variance, of the normalized Euler totient function over intervals whose length is a fixed power of the endpoint. The cited results establish mean-square information for the unrestricted remainder term, but the stated interval variance remains conjectural in the source.

Sources & referencesView supporting material

Primary source

Tom van Overbeeke, “The variance of the Euler totient function”, arXiv:1706.04028 (2017).

Progress summary

Never refreshed

Nothing recorded yet. Refresh searches the literature and the public web for attempts on this problem, and writes the first summary here.

Solutions 0

No solutions have been posted yet.