node package manager
Stop wasting time. Easily manage code sharing in your team. Create a free org »



Build Status npm version

BLAS Level 1 operations for complex-valued ndarrays


This library implements the basic vector operations of the Level 1 Basic Linear Algebra Subprograms (BLAS). Many of these functions are also implemented in ndarray-ops—which also has functions that are not included in BLAS. So the right answer is probably some blend of the two. This library exists mainly to frame things in a relatively standard, coherent framework.

NB: This library performs no checks to ensure you're only passing one-dimensional vectors. That's either a bug or a feature, depending on how you think about it.

Function Operation Description
swap(x_r,x_i,y_r,y_i) swap Swap the elements of x and y
scal(alpha_r,alpha_i,x_r,x_i) scal Multiple vector x by scalar alpha
copy(x_r,x_i,y_r,y_i) copy Copy x into y
axpy(alpha_r,alpha_i, x_r,x_i, y_r,y_i) axpy Multiple x by alpha and add it to y
cpsc(alpha_r,alpha_i, x_r,x_i, y_r,y_i) cpsc Multiply x by alpha and assign it to y
dotu(x_r,x_i,y_r,y_i) dot Calculate the product transpose(x) * y.
doth(x_r,x_i,y_r,y_i) dot Calculate the product conj(x) * y.
nrm2(x_r,x_i) nrm2 Calculate the 2-norm of x
asum(x_r,x_i) asum Calculate the 1-norm of x
iamax(x_r,x_i) Not yet implemented

A note on working with complex ndarrays

ndarrays only hold real numbers of varying types and javascript has no native complex type, so the best we can do for now is to try to encapsulate a decent amount of that. This library deals with vectors, but to start with the more general case of storing, for example, the matrix

sample matrix,

here are two methods:

  • Store the real and imaginary components in multiple arrays:
var a_r = ndarray([1,3,7, -2,1,-5], [2,3]),
    a_i = ndarray([2,4,8,  4,-2,6], [2,3]);
  • Interleave the real and imaginary components:
var a = ndarray([1,2,3,4,7,8,-2,4,1,-2,-5,6], [2,3,2]),
    a_r = a.pick(null,null,0),
    a_i = a.pick(null,null,1);

In this example, there's an additional final dimension of the array. This applies to vectors, matrices, and higher-dimensional arrays.

I won't comment on the relative effiency of each method.


Usage should be pretty straightforward. There aren't really any options or variations.

var cblas1 = require('ndarray-blas-level1-complex');
var x = ndarray([1,2,3,5,6,7],[3,2]);
var y = ndarray([3,4,5,2,3,1],[3,2]);
var x_r = x.pick(null,0),
    x_i = x.pick(null,1),
    y_r = y.pick(null,0),
    y_i = y.pick(null,1);
cblas1.axpy( 2, 3, x_r, x_i, y_r, y_i );


(c) 2015 Ricky Reusser. MIT License