A Useful Inequality on Ratios of Integrals, With Application to Maximum Likelihood Estimation
证明了一个已知协方差不等式的推广形式，并用于给出可观测随机变量Y的对数似然函数关于参数θ严格凹的充分条件，其中θ通过不可观测变量X影响Y的分布。
Abstract A generalization of a well-known covariance inequality is proved and is used to furnish sufficient conditions for the strict concavity of the log-likelihood of an observable random variable Y as a function of a real parameter θ that governs the distribution of an unobservable random variable X on which the distribution of Y depends in a known way.