What's the most efficient way to test two integer ranges for overlap?

Question

Given two inclusive integer ranges [x1:x2] and [y1:y2], where x1 <= x2 and y1 <= y2, what is the most efficient way to test whether there is any overlap of the two ranges?

A simple implementation is as follows:

bool testOverlap(int x1, int x2, int y1, int y2) {
  return (x1 >= y1 && x1 <= y2) ||
         (x2 >= y1 && x2 <= y2) ||
         (y1 >= x1 && y1 <= x2) ||
         (y2 >= x1 && y2 <= x2);
}

But I expect there are more efficient ways to compute this.

What method would be the most efficient in terms of fewest operations.

Answer 1

return x2 >= y1 && x1 <= y2;

Answer 2

What does it mean for the ranges to overlap? It means there exists some number C which is in both ranges, i.e.

x1 <= C <= x2

and

y1 <= C <= y2

Now, if we are allowed to assume that the ranges are well-formed (so that x1 <= x2 and y1 <= y2) then it is sufficient to test

x1 <= y2 && y1 <= x2

Answer 3

x1 < y1 ? return y1 < x2 : return x2 > y1;

Answer 4

You have the most efficient representation already - it's the bare minimum that needs to be checked unless you know for sure that x1 < x2 etc, then use the solutions others have provided.

You should probably note that some compilers will actually optimise this for you - by returning as soon as any of those 4 expressions return true. If one returns true, so will the end result - so the other checks can just be skipped.

Answer 5

Here's my version:

int xmin = min(x1,x2)
  , xmax = max(x1,x2)
  , ymin = min(y1,y2)
  , ymax = max(y1,y2);

for (int i = xmin; i < xmax; ++i)
    if (ymin <= i && i <= ymax)
        return true;

return false;

Unless you're running some high-performance range-checker on billions of widely-spaced integers, our versions should perform similarly. My point is, this is micro-optimization.

Answer 6

I suppose the question was about the fastest, not the shortest code. The fastest version have to avoid branches, so we can write something like this:

for simple case:

static inline bool check_ov1(int x1, int x2, int y1, int y2){
    // insetead of x1 < y2 && y1 < x2
    return (bool)(((unsigned int)((y1-x2)&(x1-y2))) >> (sizeof(int)*8-1));
};

or, for this case:

static inline bool check_ov2(int x1, int x2, int y1, int y2){
    // insetead of x1 <= y2 && y1 <= x2
    return (bool)((((unsigned int)((x2-y1)|(y2-x1))) >> (sizeof(int)*8-1))^1);
};