CODEWITHSUNDEEP
c++3dfloating-point
Alan
Alan

Reputation: 303

Calculate the 3d distance between point and plane, C++

I'm using: D = |ax + by + cz + d| / |n| where n is the normal to plane; a, b, c, d are the coefficients of the equation of the plane; x, y, z are the coordinates of the point from the plane. To calculate the distance from a 3d point to a 3d plane. The issue that I'm having is that the distances in question are extremely small and this is causing the result( a double ) to be represented in scientific notation, which is not handled correctly in if statements. For example:

if( dist == 0 )
{
    //Execute this
}

If dist is any scientific number the code inside the if statement is executed, even though dist is not 0. My question is, is there anyway the scientific number can be converted back into fixed notation to make it usable in if statements similar to these?? Im using VisualStudio 2010, C++.

Upvotes: 0

Views: 1385

Answers (2)

Matt
Matt

Reputation: 7160

dist is not represented in scientific notation (unless you are storing it as a string) that's just how it is printed. As another minor point, it's usually a good idea to compare to a value or the same type. 0 is an integer, 0.0 is a double.

From what I can see from some quick tests, in order for you to be seeing dist == 0 as true, it would actually have to be zero. That means you have all the numbers down to DBL_MIN, which is 2.2250738585072014e-308 for a 64 bit IEEE754 fpu. More likely your maths is wrong, and it is actually zero. Check your numerator before you do the division.

What on earth is physically that size? Well if you are specifying the diameter of an electron in units of "the diameter of the universe", then that's only 3.2×10^-42. I'm not sure there is an easy way to visualize just how small doubles can be. I tried 1 / number of atoms in the universe and it still wasn't small enough.

Upvotes: 0

Archie
Archie

Reputation: 2672

Normally you would use some tolerance value to compare floating-point numbers:

#define EPSILON (1e-6)

// dist == 0.0?
if (dist < EPSILON) {
    // ...
}

Or to compare to any other floating point v:

// dist == v?
if (fabs(dist - v) < EPSILON) {
    // ...
}

Sure, you have to choose EPSILON according to your problem.

Upvotes: 3

Related Questions