The central limit theorem and Poincaré-type inequalities (Q1103948)

For any random variable X, define a functional \[ U(X)=\sup \{Var g(X)/(Var X)(E g'(X)^ 2)| g\in C\}, \] where C is the class of all absolutely continuous functions g such that \(0<Var g(X)<\infty\). It turns out that U(X)\(\geq 1\) with equality holding if and only if X is normally distributed. Using this fact as well as a Poincaré-type inequality for sums of independent random variables, both sufficiency and necessity of the Lindeberg condition for the central limit theorem to hold have been established.