if statement returning false when the condition is true

Question

I am having trouble with an if statement that doesn't always return true when it meets the condition (see code below).

This if statement is in a loop that is embedded in another loop that runs for 7 (6, actually because the last one is empty) iterations and I have noticed that it works correctly for the first 3 iterations. The thing that drew my attention was that

in the first iteration of loop 1 the condition is met at the 3rd iteration of loop 2 and it returns true
in the second iteration of loop 1 the condition is met at the 2nd iteration of loop 2 and it returns true
in the third iteration of loop 2 the condition is met at the 1st iteration of loop 2 and it returns true
in the 4th iteration the condition is true in all cases, but it always returns false
in the 5th iteration the condition is false in all cases and it returns false
same in the 6th iteration
the 7 iteration of the loop is empty


void compute_neighbours(vector<vector<double>> &positions,double &GPmin, double &GPmax)
{
 vector<int> neighbours_i;
 neighbours.assign(positions.size(),neighbours_i);

 for (unsigned i=0; i<positions.size();i++)
   {
     cout << "Grid point " << i << " looking at :" << endl;
     for (unsigned k=i+1; k<positions.size();k++)
     {
      cout << " " << k << " distance: ";
      double r=delta(positions[k],positions[i]).modulo();
      cout << r << ", GPmin: " << GPmin << ", GPmax: " << GPmax;
      if ((r>=GPmin) and (r<=GPmax))
      {
        cout << "--adding point ";
        neighbours[i].push_back(k);
        neighbours[k].push_back(i);
      }

      cout << endl;
     }
   }
}

This is the output I obtain:

Grid point 0 looking at :
 1 distance: 0.212132, GPmin: 0, GPmax: 0.15
 2 distance: 0.212132, GPmin: 0, GPmax: 0.15
 3 distance: 0.15, GPmin: 0, GPmax: 0.15--adding point 
 4 distance: 0.212132, GPmin: 0, GPmax: 0.15
 5 distance: 0.212132, GPmin: 0, GPmax: 0.15
 6 distance: 0.3, GPmin: 0, GPmax: 0.15
Grid point 1 looking at :
 2 distance: 0.212132, GPmin: 0, GPmax: 0.15
 3 distance: 0.15, GPmin: 0, GPmax: 0.15--adding point 
 4 distance: 0.212132, GPmin: 0, GPmax: 0.15
 5 distance: 0.3, GPmin: 0, GPmax: 0.15
 6 distance: 0.212132, GPmin: 0, GPmax: 0.15
Grid point 2 looking at :
 3 distance: 0.15, GPmin: 0, GPmax: 0.15--adding point 
 4 distance: 0.3, GPmin: 0, GPmax: 0.15
 5 distance: 0.212132, GPmin: 0, GPmax: 0.15
 6 distance: 0.212132, GPmin: 0, GPmax: 0.15
Grid point 3 looking at :
 4 distance: 0.15, GPmin: 0, GPmax: 0.15
 5 distance: 0.15, GPmin: 0, GPmax: 0.15
 6 distance: 0.15, GPmin: 0, GPmax: 0.15
Grid point 4 looking at :
 5 distance: 0.212132, GPmin: 0, GPmax: 0.15
 6 distance: 0.212132, GPmin: 0, GPmax: 0.15
Grid point 5 looking at :
 6 distance: 0.212132, GPmin: 0, GPmax: 0.15
Grid point 6 looking at :
Grid Point 0's neighbours are: 3,
Grid Point 1's neighbours are: 3,
Grid Point 2's neighbours are: 3,
Grid Point 3's neighbours are: 0,1,2,
Grid Point 4's neighbours are: 
Grid Point 5's neighbours are: 
Grid Point 6's neighbours are: 
Maximum number of neighbours: 3

Grid point 3 should also be neighbours with 4, 5 and 6, but for some reason the if statement returns false even if the distance falls into the specified range (which doesn't happen in the other iterations of the loop).

I have tried replacing the delta().modulo() function by a manually coded euclidean distance, and obtained the same results.

Any ideas about what is going wrong will be greatly appreciated!

Maybe it's slightly outside of the range? Try printing it with more precision: `std::setprecision(1000) << k`. — HolyBlackCat, Sep 01 '19 at 19:12
If statements don't return false when the condition is true. The condition is false. That's your proof. So, you made a mistake. Narrow it down to a [mcve] and you'll find the mistake. — Lightness Races in Orbit, Sep 01 '19 at 19:12
@KorelK [Equivalent](https://en.wikipedia.org/wiki/Operators_in_C_and_C%2B%2B#C.2B.2B_operator_synonyms) and therefore irrelevant. — Lightness Races in Orbit, Sep 01 '19 at 19:13
See [Is floating point math broken?](https://stackoverflow.com/questions/588004/is-floating-point-math-broken) — 1201ProgramAlarm, Sep 01 '19 at 19:17

score 4 · Answer 1 · answered Sep 01 '19 at 19:32

You are running into a round off error. It prints 0.15, but in reality is something slightly smaller or larger, so that the comparison fails.

For example,

#include <iostream>

int main() {
  double x = 0.15, y = 0.14999999999;
  if (x <= y)
    std::cout << "Equation returns true." << std::endl;
  else
    std::cout << "x = " << x << "; y = " << y << "; but x <= y is false." << std::endl;
}

which prints

x = 0.15; y = 0.15; but x <= y is false.

See https://ideone.com/GoRkqD

if statement returning false when the condition is true

This is the output I obtain:

1 Answers1