{"id":"arxiv/2610.04135","url":"https://arxiv.org/abs/2610.04135","tags":["math.ST","stat.ME"],"score":0,"title":"Minimax Rates for Learning Smooth Populations of Parameters","domain":"math.ST","source":"arxiv","authors":["JungHo Lee","Edward H. Kennedy"],"reviews":0,"subject":"math.ST","summary":"We study pointwise estimation of a smooth binomial mixing density and characterize its minimax rates. Assuming that the mixing density is $s$-Hölder smooth, we first consider the homogeneous setting with a common number of binomial trials $t$. We derive matching lower and upper bounds that reveal three regimes depending on the relative sizes of the sample size $n$ and the number of trials $t$. When $t$ is small, the finite number of trials creates an identification barrier that persists regardless of sample size; at intermediate $t$, more data reduce sampling uncertainty, while the number of…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-10-02","reactions":{}}