1: \begin{abstract}

2: We derive uniform convergence rates for the maximum likelihood estimator and minimax lower bounds for parameter

3: estimation in two-component

4: location-scale Gaussian mixture models with unequal variances.

5: We assume the mixing proportions of the mixture are known and fixed, but make no separation

6: assumption on the underlying mixture components. 

7: A phase transition is shown to exist in the optimal parameter estimation rate, depending

8: on whether or not the mixture is balanced. Key to our analysis is a careful study of the 

9: dependence between the parameters of location-scale

10: Gaussian mixture models, as captured through systems of polynomial equalities and inequalities

11: whose solution set drives the rates we obtain. 

12: A simulation study illustrates the theoretical findings of this work.

13: \end{abstract}

14: