A Remark on the Convolution with Box Splines