Skip to content

Fast matrix and vector maths library for Clojure - as a core.matrix implementation

Notifications You must be signed in to change notification settings

mikera/vectorz-clj

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

vectorz-clj

Join the chat at https://gitter.im/mikera/vectorz-clj

Clojars Project

Build Status Dependency Status

Fast vector and matrix library for Clojure, building on the Vectorz library and designed to work with the core.matrix array programming API.

vectorz-clj is designed so that you don't have to compromise, offering both:

  • An idiomatic high-level Clojure API using core.matrix
  • General purpose multi-dimensional arrays
  • High performance (about as fast as you can get on the JVM). vectorz-clj is currently the fastest pure-JVM vector/matrix library available for Clojure

The library was originally designed for games, simulations and machine learning applications, but should be applicable for any situations where you need numerical double arrays.

Important features:

  • "Pure" functions for an idiomatic functional programming style are provided. These return new vectors without mutating their arguments.
  • Primitive-backed special purpose vectors and matrices for performance, e.g. Vector3 for fast 3D maths.
  • Flexible DSL-style functions for manipulating vectors and matrices, e.g. the ability to create a "view" into a subspace of a large vector.
  • core.matrix fully supported - see: https://github.com/mikera/core.matrix
  • Pure cross-platform JVM code - no native dependencies
  • "Impure" functions that mutate vectors are available for performance when you need it: i.e. you can use a nice functional style most of the time, but switch to mutation when you hit a bottleneck.

Documentation

vectorz-clj is intended to be used primarily as a core.matrix implementation. As such, the main API to understand is core.matrix itself. See the core.matrix wiki for more information:

For more information about the specific details of vectorz-clj itself, see the vectorz-clj Wiki.

Status

vectorz-clj requires Clojure 1.4 or above, Java 1.7 or above, and an up to date version of core.matrix

vectorz-clj is reasonably stable, and implements all of the core.matrix API feature set.

License

Like Vectorz, vectorz-clj is licensed under the LGPL license:

Usage

Follow the instructions to install with Leiningen / Maven from Clojars:

You can then use Vectorz as a standard core.matrix implementation. Example:

    (use 'clojure.core.matrix)
    (use 'clojure.core.matrix.operators)           ;; overrides *, + etc. for matrices
    
    (set-current-implementation :vectorz)  ;; use Vectorz as default matrix implementation
    
    ;; define a 2x2 Matrix
    (def M (matrix [[1 2] [3 4]]))
    M
    => #<Matrix22 [[1.0,2.0][3.0,4.0]]>
    
    ;; define a length 2 vector (a 1D matrix is considered equivalent to a vector in core.matrix)
    (def v (matrix [1 2]))
    v
    => #<Vector2 [1.0,2.0]>
    
    ;; Matrix x Vector elementwise multiply
    (mul M v)
    => #<Matrix22 [[1.0,4.0],[3.0,8.0]]>
    
    ;; Matrix x Vector matrix multiply (inner product)
    (inner-product M v)
    => #<Vector2 [5.0,11.0]>

For more examples see Wiki Examples

About

Fast matrix and vector maths library for Clojure - as a core.matrix implementation

Resources

Stars

Watchers

Forks

Packages

No packages published