← All problems

Golomb–Welch conjecture

For every integer dimension n≥3n\geq 3 and every integer radius r≥2r\geq 2, there is no set C⊆ZnC\subseteq\mathbb{Z}^n such that Zn\mathbb{Z}^n is tiled by translates of the Lee ball Brn={x∈Zn:∑i=1n∣xi∣≤r}B_r^n=\left\{x\in\mathbb{Z}^n:\sum_{i=1}^n\lvert x_i\rvert\leq r\right\}; equivalently, Zn=⨆c∈C(c+Brn)\mathbb{Z}^n=\bigsqcup_{c\in C}(c+B_r^n) does not hold.

References

Progress summary

Refreshed
Open · progress reported

A new October 2026 paper strengthens the lattice case, but the full conjecture remains open.

Golomb and Welch posed the conjecture in 1968: in its strong form, no perfect Lee code PL(n,e)PL(n,e) exists for n≥3n\geq 3 and e>1e>1. The unrestricted non-lattice problem is unresolved.

Known results

  • The conjecture is proved for 3≤n≤53\leq n\leq 5 and all e≥2e\geq 2 (survey, 2017).
  • No linear PL(n,2)PL(n,2) codes exist for n≤12n\leq 12 (survey, 2017).
  • No lattice tiling by radius-22 Lee spheres exists for any n≥3n\geq 3; hence the conjecture holds when 2n2+2n+12n^2+2n+1 is prime (2018).
  • Linear radius-22 codes are excluded for 3≤n≤1003\leq n\leq 100 except eight dimensions (2018).

October 2026 lattice-tiling advance

Ang Xiao and Yue Zhou derive group-ring conditions and combine them with a decision algorithm for lattice Lee-ball tilings, yielding broad new exclusions and a narrower computationally accessible case. This is claimed progress for the lattice setting, not a resolution of the full conjecture.

Current status (as of October 2026): substantial lattice and low-dimensional cases are settled, while the unrestricted Golomb–Welch conjecture remains open.

Sources

Solutions 0

No solutions have been posted yet.