A note on the smooth approximation to |x(1-x)ldots (n-1-x)| using Gaussian error function