5
In this figure, roundoff errors are enhanced by using
float
variables instead of
double
.