This is libnum from the CAML-light contrib directory.
It provides arbitrary-precision arithmetic functions over integers
and rationals to CAML.