Why is it better to calculate the midpoint c between a and b in the halving method with
c = a + (b - a) / 2.
instead of simple:
c = (a + b) / 2.
all variables are floating points.
Any potential overflow / loss of precision in intermediate calculations should be avoided.