also use 0 as a coeff, the original paper uses 0.

for powerseries, my docs were correct, but the implementation
wrong...  partly because I used zip which works on mismatched
lengths..

the last two are scarily short.