Cusick’s conjecture

Let s2(n)s_2(n) denote the number of ones in the binary expansion of nn, and for each integer t≥1t\ge 1 define ctc_t to be the natural density ct=lim⁡N→∞1N#{0≤n<N:s2(n+t)≥s2(n)}c_t=\lim_{N\to\infty}\frac{1}{N}\#\{0\le n<N:s_2(n+t)\ge s_2(n)\}. Determine the asymptotic behavior, as k→∞k\to\infty, of inf⁡t≥1, s2(t)=k(ct−12)\inf_{t\ge 1,\,s_2(t)=k}\left(c_t-\frac12\right). The claimed sharp asymptotic is inf⁡s2(t)=k(ct−12)∼12π(log⁡2kk)3/2\inf_{s_2(t)=k}\left(c_t-\frac12\right)\sim\frac{1}{2\sqrt{\pi}}\left(\frac{\log_2 k}{k}\right)^{3/2} as k→∞k\to\infty.

References

Progress summary

Refreshed
Claimed solved

A June 2026 preprint claims the original conjecture is proved, while a newer report claims only a sharper estimate and neither claim has independent verification.

Cusick’s conjecture asks whether ct>12c_t>\frac12 for every positive integer tt, where ctc_t measures how often adding tt does not decrease the binary digit sum. The original conjecture is now the subject of a claimed proof, while the listed problem concerns a sharper quantitative asymptotic.

Known results

  • Drmota et al. (2015): ct>12c_t>\frac12 for a set of tt of asymptotic density one.
  • Drmota et al. (2015): positivity for integers whose binary expansions contain sufficiently many blocks of ones.
  • Drmota et al. (2015): for tj=(4j−1)/3t_j=(4^j-1)/3, ctj=12+342πj+O(j−3/2)c_{t_j}=\frac12+\frac{\sqrt{3}}{4\sqrt{2\pi j}}+O(j^{-3/2}), showing the bound is asymptotically tight.

June–August 2026 claimed resolution and refinement

On June 22, 2026, Kaimin Cheng’s preprint A first-exit proof of Cusick’s sum-of-digits conjecture claimed ct≥12+2−2s2(t)−1c_t\ge\frac12+2^{-2s_2(t)-1} for every t≥1t\ge1, hence a complete proof; this remains unverified. A report dated August 26, 2026 described a further claimed fixed-weight asymptotic, 12π(log⁡2kk)3/2\frac{1}{2\sqrt{\pi}}\left(\frac{\log_2 k}{k}\right)^{3/2}, but supplied no author or independent confirmation.

Current status (as of August 2026): The original inequality ct>12c_t>\frac12 is claimed proved by Cheng but remains unverified, and the sharper fixed-weight asymptotic is likewise only claimed; no independent verification or disproof is recorded.

Sources

Solutions 0

No solutions have been posted yet.