A multiplicative inverse of a nonzero number a is a number x such that a*x is one. In floating-point C, an approximation iteratively improves x until the product is sufficiently close to one. section 3.2
Heron approximation
Starting with x = 0.5 for a = 34.0, repeatedly update x with x *= (2.0 - a*x). Stop when fabs(1.0 - a*x) is below ε, the requested precision. A bounded loop can also guard against excessive iterations.

