T(p) upper bound for prime groups

Question 8 · arXiv:2006.00534

arXiv Question high confidence— first stated 2020-12-30

Authors: Noga Alon, Noah Kravitz, Matt Larson

Status open medium confidence

Question 8 asks whether T(n) behaves differently when n is prime and how much T(G) depends on group structure beyond just the order. The source paper itself notes that the best known upper bound for T(p) (prime p) is only O(p^{3/4+epsilon}), which may not be tight, since the constructions achieving better bounds for composite n rely on proper subgroups unavailable in Z/pZ. No follow-up paper resolving either part of this question was found in a wide web search across arXiv, Journal of Number Theory, and the Ramanujan Journal.

Reviewer notes. No follow-up paper specifically addressing Question 8 was found. A Ramanujan Journal paper (DOI: 10.1007/s11139-025-01281-6) on minimal additive complements appeared in a 2025 search result but was inaccessible due to authentication redirect. The gap between the O(p^{3/4+epsilon}) bound for prime p and the better O(sqrt(n)) bound achievable when n has divisors of suitable size is the core open issue; the structural question (T(G) vs |G|) is related but broader.

Auto-reviewed 2026-05-15 with claude-sonnet-4-6 (web search enabled).

Question. Is the behavior of $T(n)$ different when $n$ is prime? How much does $T(G)$ depend on the structure of $G$ in addition to the order of $G$?

Context

The best upper bound constructions for $T(n)$ rely on the existence of subgroups of certain orders, so the best known upper bound for $T(p)$ when $p$ is prime is only $O(p^{3/4+\varepsilon})$, which may not be tight.

Source paper

Inverse problems for minimal complements and maximal supplements
Noga Alon, Noah Kravitz, Matt Larson · 2020-12-30
https://arxiv.org/abs/2006.00534 PDF source