← All problems

Szemerédi's problem on infinite generalized Sidon sets

For every integer h≥2h\geq 2 and every infinite BhB_h set A⊆NA\subseteq\mathbb{N}, does one have lim⁡x→∞A(x)x1/h=0\displaystyle\lim_{x\to\infty}\frac{A(x)}{x^{1/h}}=0, where A(x)=∣A∩[1,x]∣A(x)=|A\cap[1,x]|? Equivalently, is lim sup⁡x→∞A(x)x1/h=0\displaystyle\limsup_{x\to\infty}\frac{A(x)}{x^{1/h}}=0 for every infinite BhB_h set?

References

Progress summary

Refreshed
Open · solution reported

An unrefereed preprint claims a counterexample that would settle the problem negatively, but no independent verification has appeared.

The problem concerns how densely an infinite generalized Sidon set can recur after containing a prescribed finite BhB_h set. Wang and Yang claim a construction answering Szemerédi’s question negatively.

Known results

  • For ordinary infinite Sidon sets, the counting exponent satisfies 2−1≤α≤1/2\sqrt{2}-1 \le \alpha \le 1/2; the lower bound is due to Ruzsa.
  • Cilleruelo constructed an explicit infinite Sidon sequence with exponent 2−1\sqrt{2}-1.
  • For h≥3h \ge 3, Cilleruelo gave a nonconstructive infinite BhB_h construction with exponent (h−1)2+1−(h−1)\sqrt{(h-1)^2+1}-(h-1).
  • The general density gap for infinite Bh[g]B_h[g] sets was reported as unresolved.

Claimed negative answer (unrefereed preprint)

Wang and Yang claim that every finite BhB_h set, for h≥2h \ge 2, has an infinite BhB_h superset with lim sup⁡A(x)/x1/h=1/h\limsup A(x)/x^{1/h}=1/h and lim inf⁡A(x)/x1/h=0\liminf A(x)/x^{1/h}=0. This would settle the problem negatively, but the retrieved evidence contains no independent mathematical assessment or verification.

Current status (as of November 2026): A negative resolution is claimed in an unrefereed preprint, but the claim remains unverified; absent that construction, the problem remains open.

Sources

Solutions 0

No solutions have been posted yet.