If you use struct to hold the number as fraction $\frac{a}{b}$,it will be easy to only output a number like $\frac{a}{b}$ at the end,so that the precision requirements is not that harsh.