morton-span

This repository implements a morton transform for mdspan

Motivation

In Monte Carlo simulations is oftem nore efficent a morton transform of the particle positions to a morton index. This is because the morton index is a contiguous index and thus can be used to access the particles in a cache friendly way.

Usage

It is reccomended to use cpm to include this library in your project. To do so add the following to your CMakeLists.txt

CPMAddPackage(
  NAME morton-span
  GITHUB_REPOSITORY diamondinoia/morton-span
  GIT_TAG main
)

Then in your code you can use the morton transform as follows for 2d and 3d. c++23 example:

#include <morton_span.h>
    // 2D CASE
    std::mdspan<int, std::dextents<int, 2>
           morton::MortonExtents> morton_span_2d;
    // 3D CASE
    std::mdspan<int, std::dextents<int, 3>,
              morton::MortonExtents> morton_span_3d;

c++ < 23 example:

#include <morton_span.h>
    // 2D CASE
    std::experimental::mdspan<int, std::experimental::extents<std::size_t, std::experimental::dynamic_extent,
        std::experimental::dynamic_extent>, morton::MortonExtents> morton_span_2d;
    // 3D CASE
    std::experimental::mdspan<int, std::experimental::extents<std::size_t, std::experimental::dynamic_extent,
        std::experimental::dynamic_extent, std::experimental::dynamic_extent>, morton::MortonExtents> morton_span_3d;

limitations

• The morton transform is only implemented for 2d and 3d
• The morton transform is only implemented for geometries of euqal side lengths (i.e. a cube or a square)
• The morton transform is only implemented for dimensions that are a power of 2

Acknowledgements

• mdspan
• libmorton