The float domain includes approximate representations of real numbers hav-
ing precision at least p. Also, real and double precision have implementation-
dependent precision, where the precision of double precision is never smaller
than the one of real.