narray

A simple library to give (very minimal) functions of multidimensional array compatible with C/Fortran

Installation

1. Add the dependency to your shard.yml:

   dependencies:
     narray:
       github: jmakino/narray
   

2. Run shards install

Getting started

At your project directory (where you ran shards install), try

crystal lib/narray/examples/narray_test.cr

should show some output.

Usage

  require "narray"

provides the Narray class.

  Narray(T).new(nx : Int32,ny : Int32 = 1, nz : Int32 = 1)

creates the Narray of type T of up to three dimensions. For example,

  x = Narray(Float64).new(10)

makes one-dimensional array of size 10 of Float 64, and

  xy = Narray(Float64).new(10,20)

makes two-dimensional array of size 10 times 20 of Float 64.

The elements of x can be accessed and assigned using usual [], like

  pp! x[1]
  x[2]=5.0

and the elements of xy as

  pp! xy[1,2]
  x[2,3]=5.0

The storage order is in the C convention, not in the Fortran convention.

If you supply -Drange_check (or --define range_check) compiler option, the code to check the range of indices is generated. This checking is done through the use of Slice struct, and thus checks only if the address is outside the range of the array. Thus, in the above example, x[10,0] would be an error but x[0,20] would not

You can use Narray.range_check? class function to test if range check is enabled or not.

You can use all member functions supported by Slice. For example, you can sort the array by x.sort!

A special each_index method is available.

   xyz = Narray(Float64).new(10,10,10)
   xyz.each_index{|i,j,k| xyz[i,j,k]= ...}

loops over all indices of xyz. For 1 and 2-dimentional arrays, you can write

   x = Narray(Float64).new(10)
   x.each_index{|i| x[i]= ...}
   xy = Narray(Float64).new(10,10)
   xy.each_index{|i,j| xy[i,j]= ...}

The each_index(idim) function is also avaiable, which loops over the idim-th dimension.