The performance of kernel density functions in kernel distribution function estimation
We note that the uniform is the optimal kernel density for kernel estimation of the distribution function, or its inverse, and that several other popular kernel densities perform virtually as well. Parallels with the kernel density estimation case are made.