Newton Raphson for Cube Root (Precision Error in Double) - c++

Question

I am trying to implement Newton Raphson in C++.
Approach: In my root() function my while(temp-e>0) exists even when temp==e. I know it is due to using doubles and comparing double might give precision errors sometimes. But still I want to know how can I make sure that the loop exists only when temp==0. Error occurring in the test case 64. My code is returning 4.000001, while it should return 4.000000.
Code:
#include<stdio.h>
double root(int n)
{
double x=n,a,b,e=0.000001,temp;
a=2*x;
b=n/(x*x);
temp=1;
while(temp-e>0)
{
x=(a+b)/3;
a=2*x;
b=n/(x*x);
temp=(n/(x*x)-x)/3;
temp=temp<0?-temp:temp;
}
return x;
}
int main()
{
printf("%lf\n",root(64));
return 0;
}
While this code is giving correct answer, where I am doing two more iterations from where temp starts failing:
#include<stdio.h>
double root(int n)
{
double x=n,a,b,e=0.000001,temp;
a=2*x;
b=n/(x*x);
int i=0;
temp=1;
while(i<=2)
{
x=(a+b)/3;
a=2*x;
b=n/(x*x);
temp=(n/(x*x)-x)/3;
temp=temp<0?-temp:temp;
if(temp<e)
i++;
}
return x;
}
int main()
{
int t,n;
scanf("%d",&t);
while(t--)
{
scanf("%d",&n);
printf("%lf\n",root(n));
}
return 0;
}