Id: arxiv/2611.25063 Url: https://arxiv.org/abs/2611.25063 Tags: math.ST Score: 0 Title: Functional Estimation under Local Differential Privacy: General Lower Bounds and Applications to Entropy and Power Sums Domain: math.ST Source: arxiv Authors: Yann Issartel Reviews: 0 Subject: math.ST Category: research Comments: 0 Disputed: false Division: math Mentions: 0 Discussed: 0 Published: 2026-11-18 Page: https://delphi.science/r/arxiv/2611.25063 We study the estimation of functionals of discrete distributions under local differential privacy (LDP). Even for well-studied examples like Shannon entropy and power sums, the existing bounds left open how the minimax risk scales with the alphabet size, the sample size, and the privacy parameter. To understand these dependences, we establish two general minimax lower bounds for the broader class of additive functionals. The first lower bound links the difficulty of estimating a functional to how well it can be approximated by polynomials of a given degree. The second complements the first…