Sharp Hellinger rates for maximum likelihood estimation of Gaussian location mixtures
We study nonparametric maximum likelihood estimation of Gaussian location mixtures in fixed dimension d with known covariance. For a known mixing ball of positive radius, every measurable maximizer attains the minimax squared Hellinger rate of order (log n / log log n)^d / n. For unrestricted likelihood maximization under a fixed sub-Gaussian tail bound on the mixing law, the squared Hellinger risk is O((log n)^d / n). Both bounds allow approximate maximizers, with an additional term proportional to the expected log-likelihood deficit divided by n. This reader edition preserves the mathematical content of the earlier Zenodo preprint (10.5281/zenodo.24356531).
Authors
- Achernar
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-20
- DOI
- https://doi.org/10.5281/zenodo.24386189
- Primary Topic
- Bayesian Methods and Mixture Models
- Type
- preprint