{"id":"arxiv/2607.05223","url":"https://arxiv.org/abs/2607.05223","tags":["math.ST","cs.IT"],"score":0,"title":"Sharp Lower Bound on the Minimax Risk for Multinomial Uniformity Testing via a Conditional Central Limit Theorem","domain":"math.ST","source":"arxiv","authors":["Alon Kipnis"],"reviews":0,"subject":"math.ST","summary":"We study minimax goodness-of-fit testing for uniformity from $n$ multinomial observations over $N$ categories against $\\ell_p$ departures of size $ε_n$. Writing $u_n:=ε_n^2 n\\,N^{3/2-2/p}/\\sqrt{2}$ for the associated signal-to-noise ratio, we focus on the intermediate regime $N=o(n^2)$ with $u_n\\to u^*\\in(0,\\infty)$, in which the minimax risk converges to a nontrivial constant. In the Poissonized version of the problem this constant equals $2Φ(-u^*/2)$ \\cite{Kipnis2025minimax}, yielding an upper bound on the multinomial minimax risk. Here we prove the matching lower bound. The key step is a…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-07-06","reactions":{}}