If we implement the arctan(x) using the embedded 9 × 9 bit multipliers we have to take into account that our values are in the range − 1 ≤ x < 1. We therefore use a fractional integer representation in a 1.8 format.