A strange thing about for-loop

[timney]

Hi, everyone,I have a question about for-loop .Here is simplified question.

Output result for clip1 is 0.9 ,and for clip2 is 0.7 ?
Does it means 0.8<0.8 but 0.7!<0.7 for C++ ?
How strange it is!

  //clip1
void main(){
	int i;
	double x(0),h(0.1);
	for(i=1;x<0.8;i++){
		x=x+h;
	}

	cout<<x<<endl;
}
//clip2
void main(){
	int i;
	double x(0),h(0.1);
	for(i=1;x<0.7;i++){
		x=x+h;
	}

	cout<<x<<endl;
}

[crimsonzero2]

This is very strange (I tried it myself also), but I would personally use while loops for this since the int i isn't really necessary for the program.

[TheIdeasMan]

You shouldn't use FP values (floats & doubles ) in loop conditions. They are stored as binary fractions and cannot represent every real number.

Always use ints & cast them to FP inside the loop.

Read up about data types in the tutorial - top left of this page.

Hope all goes well.

[timney]

Thanks for all your explanations.
However, problem remains when i use while-loop.
And it is not easy to just use int value in loop condition.
How should i modify if i have to use the condition "x<0.8"?

Thanks a lot!

[TheIdeasMan]

And it is not easy to just use int value in loop condition.

How should i modify if i have to use the condition "x<0.8"?

It is easy to use an int in the loop condition - but what is slightly trickier is how to work out how many times it will run.

If you have a start value, a step value & an end value - you can calculate how many times the loop will run as an int.

However, problem remains when i use while-loop.

What I said in my last post applies to all types of conditional statements including all types of loops and if statements.

[timney]

yep, your explanations are the same as what my professors said.
I understand now that floating point is an approximate representation of real numbers and never use floating point to make comparison.
But my program is about computational physics, ........you know,......... i use fractional number and real number very often .....╮(╯▽╰)╭

[Chervil]

If you were testing for two floating-point numbers being equal, then rather than doing a direct comparison, subtract one from the other and test if the absolute value of the result is smaller than some limit.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

#include <iostream> #include <cmath> using namespace std; int main() { double x(0), h(0.1); double diff = fabs(x - 0.8); while (diff > 1e-10) { x += h; diff = fabs(x - 0.8); cout << "diff: " << diff << endl; } cout << x << endl; }

diff: 0.7 diff: 0.6 diff: 0.5 diff: 0.4 diff: 0.3 diff: 0.2 diff: 0.1 diff: 1.11022e-016 0.8