Using convolutions (S3) vs distributions (S4)

https://freakonometrics.hypotheses.org/51923