Re: A question on Newton's Method

In article <424c896c$0$63423$ed2e19e4@xxxxxxxxxxxxxxxxxxxxxxxxxx>, Jon
Harrop <usenet@xxxxxxxxxxxxxx> writes:

> > someone pointed out that using the quadratic equation is not
> > exactly stable when it comes to computers.
>
> Yes, evaluating the expressions for solutions to a quadratic as they are
> taught at school is not numerically stable when you are using floating
> point arithmetic. This is actually pedagogical example used to teach
> numerical robustness.

Actually, the Numerical Recipes in Fortran use this as an example of how
one should program it in Fortran, as opposed to the standard solution
taught in schools, which is susceptible to roundoff error.

.