nrange

The Rust standard library has a Range type, which represents a bounds for a scalar value (i.e. 1 ≤ n ≤ 100). A range can be open ended, iterated over, and tested for overlap with other ranges. I designed nrange as an extension of the Range type to vector space. This makes code less nested. Below is an example comparing an operation on a 3D array using standard Ranges and nrange.

let array3 = [[[0; 128]; 128]; 128];
// Using standard Ranges
for x in 0..128 {
  for y in 0..128 {
    for z in 0..128 {
      let element = &array3[x][y][z];
      // Do something with element
    }
  }
} 
// Using nrange
for [x, y, z] in nrange!(0..128, 0..128, 0..128) {
  let element = &array3[x][y][z];
  // Do something with element
}

Performance

My implementation is competitive with similar libraries in the Rust ecosystem. While nrange only works for ranges of contiguous integers, itertools and cartesian-rs work for any iterators. However, by restricting my use case I can extract more performance gains and integrate better with the Rust standard library. Below are benchmark results comparing my solution to comparable libraries:

• itertools ..... 1,821,573 ns
• cartesian-rs... 989,232 ns
• nrange ........ 968,853 ns