Mollified random variables

52 Views Asked by At

Given a random variable $X$ with cdf that is not smooth, I read that we can construct a random variable denoted as $X\cdot M$ with smooth cdf and smooth inverse cdf, where $M$ is a random variable supported in $[0,1]$ having smooth cdf. The reference refers to this as a standard mollifier argument and does not provide further details. My question is how this construction works. Is the $\cdot$ merely a product or denotes something else? How does one show the claim? Thank you for any suggestions or reference provided.