Random-point conjecture for sums of square roots modulo one

From papers

For integers r1r\ge 1 and real numbers n1n\ge 1, 0<η1/20<\eta\le 1/2, define

A2r(n,η):=#{(a1,,ar,b1,,br):ai,bjn, i=1raij=1rbjη}.A_{2r}(n,\eta):=\#\left\{(a_1,\ldots,a_r,b_1,\ldots,b_r):a_i,b_j\sim n,\ \left\lVert\sum_{i=1}^r\sqrt{a_i}-\sum_{j=1}^r\sqrt{b_j}\right\rVert\le\eta\right\}.

Here ai,bjna_i,b_j\sim n denotes the same dyadic range as in the source, and \lVert\cdot\rVert is distance to the nearest integer. Random-point conjecture. For every fixed integer r1r\ge 1 and every fixed ε>0\varepsilon>0, uniformly for 0<η1/20<\eta\le 1/2, one has

A2r(n,η)r,εnr+ε+n2r+εη.A_{2r}(n,\eta)\ll_{r,\varepsilon} n^{r+\varepsilon}+n^{2r+\varepsilon}\eta.

The conjecture is motivated by treating the fractional parts of i=1rai\sum_{i=1}^r\sqrt{a_i} as nrn^r random points in [0,1)[0,1). It would provide the higher-moment counting input needed, via the displayed reduction in the source, to estimate the even moments M2r(H,n)M_{2r}(H,n); no resolution is given here.

Progress summary

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

Sources & referencesView supporting material

Primary source

Yixiu Xiao, “Moment Estimates and Discrepancy for Sums of Square Roots Modulo One”, arXiv:2606.28986 (2026).

Solutions 0

No solutions have been posted yet.