numerl

Numeric Erlang is the fundamental package needed for scientific computing with Erlang.

An OTP library

It provides:

• A powerful N-dimensional array module
• Useful linear algebra (In development)

Installation

Simply run:

$ rebar3 compile

Use it in your project adding it as a dependency in your rebar.conf.

{deps, 
    [
        {enn, {git, "https://github.com/BorjaEst/numerl.git", {tag, "<version>"}}}
    ]}.

You can find more information about dependencies in rebar3 - dependencies.

Usage

This library is based on the popular NUmPy. The idea is to share a similar syntax (always it is possible) see quickstart and the following comments:

• Corrdinates references are shared: numpy[1 ,2, 1] means numerl[1, 2, 1]. See 'ndarray:get/2' to know how to get the ndarray elements.
• You can use the atom ':' to reference a full dimension when using 'ndarray:get/2' to obtain sub-ndarrays.
• You can index with lists when using 'ndarray:get/2' to get more than one element in a dimension.

Note the following differences:

• Functions are called by Erlang syntax, so instead of running 'a.shape' you should run numerl:shape(a).
• To access the indexes you must use 'ndarray:get/2'. For example to get a sub-ndarray of a 3D ndarray you can use numerl:get([':',':',[0,1]],NdArray).
• Because Erlang is dynamically typed, you should not specify the type of the elements inside the array.

Examples

First thing you need is an ndarray. You can create it this way:

1> Shape   = [4,3,2].
2> Buffer  = lists:seq(1,24).
3> NdArray = ndarray:new(Shape, Buffer).
{ndarray,[4,3,2],
         [1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24]}

Where 'Shape' are the dimensions of the ndarray and 'Buffer' the values of the array. This would represent the following ndarray in numpy:

import numpy as np
ndarray = np.array([[[ 1, 5, 9],
                     [ 2, 6,10],
                     [ 3, 7,11],
                     [ 4, 8,12]],

                    [[13,17,21],
                     [14,18,22],
                     [15,19,23],
                     [16,20,24]]])

Now you can use the functions inside numerl to operate this ndarray. For example, you can calculate the mean over the axis 0 by running:

4> Mean = numerl:mean(NdArray, Axis=0).
{ndarray,[1,3,2],[2.5,6.5,10.5,14.5,18.5,22.5]}

As you can see, after applying an operation over a dimension, it reduces the size of that dimension to '1'. You can compress all simple dimensions by running 'ndarray:reduce/1': 4>numerl:mean(NdArray, 0). {ndarray,[1,3,2],[2.5,6.5,10.5,14.5,18.5,22.5]}

5> ndarray:reduce(Mean).
{ndarray,[3,2],[2.5,6.5,10.5,14.5,18.5,22.5]}

And that is all, you can find all functions inside the module together with the documentation.

Call for Contributions

numerl appreciates help from a wide range of different backgrounds. Small improvements or fixes are always appreciated. If you are considering larger contributions outside the traditional coding work, please contact me.

TODO:

• Create a transpose function for n-dimensions array in ndarray module.
• Measure speed against other libraries:
  • When interactive mode
  • When compiled (HiPe) mode
• Add your needed functions into the numerl module.
• Tools to speed up integrating C/C++ code.

License

This software is under Apache-2.0 license.