← All problems

Afkhami-Jeddi–Cohn–Hartman–de Laat–Tajdini conjecture on Delsarte-bound rates

Let RD(δ)R_D(\delta) denote the asymptotic rate supplied by Delsarte's linear-programming bound for binary codes of relative distance δ\delta, and let LPd\mathrm{LP}_d denote the corresponding dd-dimensional Cohn–Elkies linear-programming bound for spherical codes or sphere packing, in the normalization of the cited sources. The conjecture includes the assertions RD ⁣(12−1π)=12R_D\!\left(\frac12-\frac1\pi\right)=\frac12 and lim⁡d→∞LPd1/d=e2π\lim_{d\to\infty}\mathrm{LP}_d^{1/d}=\sqrt{\frac{e}{2\pi}}, equivalently that the associated exponent is α∗=12log⁡2 ⁣(2πe)=0.6044…\alpha^*=\frac12\log_2\!\left(\frac{2\pi}{e}\right)=0.6044\ldots.

References

Progress summary

Refreshed
Open · solution reported

An unrefereed 2026 manuscript claims to settle the conjecture, but no independent verification was found.

The conjecture concerns the exact asymptotic rate of the Delsarte linear-programming bound for binary and spherical codes, including the related Cohn–Elkies sphere-packing program.

October 15, 2026 claimed resolution

Andrew Salmon’s report claims that the obstacle problem gives the lower bound and an interlacing-spectrum construction gives a matching upper bound, with LPd1/d→e/(2π)\mathrm{LP}_d^{1/d}\to\sqrt{e/(2\pi)} and exponent α∗=12log⁡2(2π/e)=0.6044…\alpha^*=\frac12\log_2(2\pi/e)=0.6044\ldots. Afkhami-Jeddi, Cohn, Hartman, de Laat, and Tajdini’s preprint separately records the binary rate RD(12−1π)=12R_D(\frac12-\frac1\pi)=\frac12 and attributes the high-dimensional Cohn–Elkies result to “OpenAI’s Chapter 1.” These are claims, not verified results.

Current status (as of October 2026): A claimed complete resolution gives the conjectured asymptotic rate, but it remains unverified; no independent assessment or correction was retrieved.

Sources

Solutions 0

No solutions have been posted yet.