An upper-bound for variance of a random variable satisfying a linear inequality

35 Views Asked by At

Let $\mathbf{x}$ be a vector of $m$ correlated non-negative random variables satisfying the following inequality $$\mathbf{A}\mathbf{x}\leq\mathbf{1},$$ where $\mathbf{A}$ is a $n\times m$ random matrix. Is there any upper-bound for the variance of elements of $\mathbf{x}$?