imclab / ndarray-proxy

Lazy ndarrays

ndarray-proxy

Creates lazily initialized ndarrays. These give a logical view of a function as an ndarray.

Introduction

Lazy arrays work just like ndarrays, and can be used in all the same ways. However, unlike dense arrays they do not keep any storage and instead use a function to determine the contents of the array. Here is a simple example:

var proxy = require("ndarray-proxy")

//Create a 10x10 lazily initialized ndarray
var x = proxy([10, 10], function(i, j) {
  console.log("Called:", i, j)
  return 10*i + j
})

x.get(1,2)    //Prints out:   Called: 1 2
x.get(7,8)    //Prints out:   Called: 7 8

It is possible to slice the view of a lazy ndarray, just like a regular ndarray:

x.lo(2,3).get(1,1)            //Prints out:  Called: 3 4
x.transpose(1,0).get(3,7)     //Prints out:  Called: 7 3

You can use lazy ndarrays with cwise or any other library that works with ndarrays.

It is also possible to add a setter to the lazy ndarray to implement custom assignment operations. For example:

var y = proxy([10,10],
    function(i,j) { return 10*i+j; }, 
    function(i,j,v) {
      console.log("SET:", i, j, "=", v)
    })

y.set(2,3,10) //Prints:  SET: 2 3 = 10

If a setter isn't specified, then writing to the array throws an error.

Install

You can install the library using npm:

npm install ndarray-lazy

And like all ndarray modules it should work in a browser that supports typed arrays using browserify.

API

require("ndarray-proxy")(shape, get_func[, set_func])

Creates a wrapper for an ndarray

• shape is the shape of the wrapped ndarray
• get_func(i0,i1,...) implements access to the ndarray
• set_func(i0,i1,...,v) implements writing to the ndarray

Returns A proxy view of the functions.

Credits

(c) 2013 Mikola Lysenko. MIT License