I have an API for reading multi-dimensional arrays, requiring to pass a vector of ranges to read sub-rectangles (or hypercubes) from the backing array. I want to read this array "linearly", all elements in some given order with arbitrary chunk sizes. Thus, the task is with an off
and a len
, to translate the elements covered by this range into the smallest possible set of hyper-cubes, i.e. the smallest number of read commands issued in the API.
For example, we can calculate index vectors for the set of dimensions giving a linear index:
def calcIndices(off: Int, shape: Vector[Int]): Vector[Int] = {
val modsDivs = shape zip shape.scanRight(1)(_ * _).tail
modsDivs.map { case (mod, div) =>
(off / div) % mod
}
}