← All problems

Wooley’s L¹ question for Beatty sequences

For real numbers α>1\alpha>1 and β\beta, define the Beatty sequence B(α,β)={⌊αm+β⌋:m∈N}\mathcal{B}(\alpha,\beta)=\{\lfloor \alpha m+\beta\rfloor:m\in\mathbb{N}\}. Let e(t)=exp⁡(2πit)e(t)=\exp(2\pi i t) and, for N≥1N\ge 1, define Sα,β,N(θ)=∑n∈B(α,β)∩[1,N]e(nθ)S_{\alpha,\beta,N}(\theta)=\sum_{n\in\mathcal{B}(\alpha,\beta)\cap[1,N]}e(n\theta). Wooley asked whether, for every such α\alpha and β\beta, one has ∫01∣Sα,β,N(θ)∣ dθ=Oα,β(log⁡N)\int_0^1\lvert S_{\alpha,\beta,N}(\theta)\rvert\,d\theta=O_{\alpha,\beta}(\log N) as N→∞N\to\infty.

References

Progress summary

Refreshed
Open · solution reported

An unrefereed November 2026 preprint claims to settle the question negatively by showing the known upper-bound growth is sharp.

Wooley’s question asks whether the relevant average size for Beatty-sequence exponential sums can be reduced from a squared logarithm to a single logarithm for parameters of finite Diophantine type.

Known results

Wooley proves an upper bound of order (log⁡N)2(\log N)^2 for finite Diophantine type and notes that an order-log⁡N\log N bound would be desirable.

November 2026 claimed counterexample

On November 1, 2026, Dijia Chen’s unrefereed preprint claimed a matching lower bound of order (log⁡N)2(\log N)^2 for fixed parameters, implying that Wooley’s proposed order-log⁡N\log N improvement is false. The claim has not been independently assessed in the retrieved sources.

Current status (as of November 2026): Wooley’s (log⁡N)2(\log N)^2 upper bound is established, while Chen’s matching lower bound and the resulting negative resolution remain an unverified preprint claim.

Sources

Solutions 0

No solutions have been posted yet.