Tensor, the Vector/Matrix/Tensor library has reached v1.0!

With the recent addition of Numbers and some improvements of the documentation, Tensor now has reached version 1.0.0!

For people that do not yet know about it:

Tensor is a library that allows you to work with sparse Vectors, Matrices and higher-order Tensors, with the following nice features:

  • An implementation of the Access protocol, so you can do mymatrix[42][3].
  • It supports the arithmetic functions you would expect from vectors, matrices and tensors. These are implemented using Numbers, which means that they work on any numeric type.
  • What is even more, Tensor itself implements Numbers’ Numeric behaviour, which means that anything that does number arithmetic can now do (elementwise) vector/matrix/tensor arithmetic! It also means that you can nest matrices ad infinitum!
  • A sparse implementation: Only elements deviating from the default element of a data structure are stored. This means that e.g. a nearly-empty 10000x10000 element matrix takes up only a neglegible amount of memory.
  • While you can work with numbers, you can store anything inside: Using it as a representation of a game board (a matrix for chess, or a 3-dimensional tensor for a Rubik’s Cube), for instance, is something that is very possible.
  • Functions to rotate, transpose, transform, combine, separate, map over and reduce vectors/matrices/tensors.

Here are some examples from the Readme:


iex> vec = Vector.new([1,2,3,4,5])
#Vector-(5)[1, 2, 3, 4, 5]
iex> vec2 = Vector.new(~w{foo bar baz qux})
#Vector-(4)["foo", "bar", "baz", "qux"]
iex> vec2[2]
iex> Vector.add(vec, 3)
#Vector-(5)[4, 5, 6, 7, 8]
iex> Vector.add(vec, vec)
#Vector-(5)[2, 4, 6, 8, 10]


iex> mat = Matrix.new([[1,2,3],[4,5,6],[7,8,9]],3,3)
β”Œ                          ┐
β”‚       1,       2,       3β”‚
β”‚       4,       5,       6β”‚
β”‚       7,       8,       9β”‚
β””                          β”˜
iex> Matrix.rotate_clockwise(mat)
β”Œ                          ┐
β”‚       7,       4,       1β”‚
β”‚       8,       5,       2β”‚
β”‚       9,       6,       3β”‚
β””                          β”˜
iex> mat[0]
#Vector-(3)[1, 2, 3]
iex> mat[2][2]
iex> Matrix.diag([1,2,3])
β”Œ                          ┐
β”‚       1,       0,       0β”‚
β”‚       0,       2,       0β”‚
β”‚       0,       0,       3β”‚
β””                          β”˜

iex> Matrix.add(mat, 2)
β”Œ                          ┐
β”‚       3,       4,       5β”‚
β”‚       6,       7,       8β”‚
β”‚       9,      10,      11β”‚
β””                          β”˜
iex> Matrix.add(mat, mat)
β”Œ                          ┐
β”‚       2,       4,       6β”‚
β”‚       8,      10,      12β”‚
β”‚      14,      16,      18β”‚
β””                          β”˜


  iex> tensor = Tensor.new([[[1,2],[3,4],[5,6]],[[7,8],[9,10],[11,12]]], [3,3,2])
        1,       2
          3,       4
            5,       6
        7,       8
          9,      10
            11,      12
        0,       0
          0,       0
            0,       0
  iex> tensor[1]
  β”Œ                 ┐
  β”‚       7,       8β”‚
  β”‚       9,      10β”‚
  β”‚      11,      12β”‚
  β””                 β”˜

You can find it on Hex.pm and on GitHub.


Awesome! :heart:

1 Like

Great work! I will have to dust off my chess program and use this library to simplify a lot of the functions.



1 Like

This is so cool !

1 Like

Very nice! Reached 1.0.0, been waiting for Numbers. ^.^

1 Like

I’m using Tensor in one of the Advent of Code solutions and got tripped up by β€œheight” vs. β€œwidth”.

It’s been a long time since discrete math, but a search confirms what I thought I remembered: β€œA matrix with m rows and n columns is said to have dimension m Γ— n”.

But the signature for Matrix.new is new(list_of_lists \\ [], width, height, identity \\ 0).

This seems to be backwards from the usual representation-- the width is the number of columns, while the height is the number of rows.

Matrix.new(3, 2) does what I expect – 3 rows and 2 columns – so this is really just a docs issue. (I opened a PR to fix it. First one, it looks like! :slight_smile: )



Wow, thank you so much for your pull request!

It has been merged and added to the new 1.0.1 version of the library. :smiley:

1 Like

You’re welcome! Unfortunately I am now completely stuck on how to update an element in the matrix. :confused:

Details here: https://github.com/Qqwy/tensor/issues/3

OR if you’d rather questions went elsewhere, feel free to close that and let me know where to ask!