(Silverman 1982) Recall that the kernel density estimator XXXXXXXXXXis a convolution,                                  f(y) ˆ = 1 n n i=1 k y − Yi h + h. If the observations Yi are rounded and/or...

(Silverman 1982) Recall that the kernel density estimator (12.2.4) is a convolution,

                                 f(y) ˆ = 1 n n i=1 k

If the observations Yi are rounded and/or discretized, show how to use the FFT to compute the kernel density estimate at many points using the FFT.