Optimizing over Multiple Distributions under Generalized Quasar-Convexity Condition

We study a typical optimization model where the optimization variable is composed of multiple probability distributions. Though the model appears frequently in practice, such as for policy problems, it lacks specific analysis in the general setting.