The variance asymptotic conjecture for sums of two squares

At least 7 years old · documented by

Let b(n)b(n) be the indicator of integers representable as sums of two squares. Let B(x)=∑n≤xb(n)B(x)=\sum_{n\leq x}b(n), let I(x;H)I(x;H) be the paper's smooth approximation to B(x+H)−B(x)B(x+H)-B(x), and define

Vb(X;H):=1X∫X2X(B(x+H)−B(x)−I(x;H))2 dx.V_b(X;H):=\frac{1}{X}\int_X^{2X}\big(B(x+H)-B(x)-I(x;H)\big)^2\,dx.

Let KK be the Landau–Ramanujan constant and let GG denote the function defined in the paper. The variance asymptotic conjecture. For fixed δ∈(0,1)\delta\in(0,1), as X→∞X\to\infty with H=XδH=X^\delta,

Vb(X;H)=(KG(1−δ)+o(1))Hlog⁡X.V_b(X;H)=\left(KG(1-\delta)+o(1)\right)\frac{H}{\sqrt{\log X}}.

The theorem motivating this conjecture establishes the corresponding function-field limiting structure, while the integer asymptotic remains open.

References

Primary source

Ofir Gorodetsky and Brad Rodgers, “The variance of the number of sums of two squares in F_q[T] in short intervals”, arXiv:1810.06002 (2020).

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.