We study a basic private estimation problem: each of $n$ users draws a single
i.i.d. sample from an unknown Gaussian distribution, and the goal is to
estimate the mean of this Gaussian distribution while satisfying local
differential privacy for each user. Informally, local differential priva