when approximating the value of transcendental functions, why are some centers more useful than others?