Copyright (c) 2013 - Michael Nelson [email protected].
Geometric Algebra provides a different way of thinking about vectors, geometry and their applications to physics. You can read more on the Wikipedia article but be warned that it gets (unnecessarily) complicated rather quickly. I've created this JavaScript port of an old Python geometric algebra library that I wrote a few years ago with the aim of eventually creating an interactive introduction to Geometric Algebra - highlighting it's simplicity and beauty.
$ git clone https://github.com/absoludity/geometricalgebra.js.git
$ npm install
$ bin/shell
This will drop you in a nodejs REPL with the multivector library already loaded into the namespace as mv, where you can create vectors:
var v1 = new mv.MultiVector("2x + 3y");
var v2 = new mv.MultiVector("x + 2y");
and multiply them together:
v1.mul(v2).toString()
'8 + 1xy'
For example, if you multiply two vectors of unit length together, your can use the result to rotate any other vectors. Here we'll create a unit vector along the x-axis, and a unit vector that is 45 degrees towards the y-axis. In this case it's easier to create the terms explicitly rather than relying on the parsing of a string:
var x = new mv.MultiVector([new mv.MultiVectorTerm(1, [1])]);
var x45y = new mv.MultiVector([
new mv.MultiVectorTerm(1.0 / Math.sqrt(2), [1]),
new mv.MultiVectorTerm(1.0 / Math.sqrt(2), [2]),
]);
Verify that the second vector is indeed unit length (well, with floating-point errors) and then create the rotation by simply multiplying the two vectors:
x45y.mul(x45y).toString()
'0.9999999999999998'
var rotate45 = x45y.mul(x);
rotate45.toString()
'0.7071067811865475 - 0.7071067811865475xy'
We can now apply the rotation one or more times to see the results:
rotate45.mul(x).toString()
'0.7071067811865475x + 0.7071067811865475y'
rotate45.mul(rotate45).mul(x).toString()
'0.9999999999999998y'
rotate45.mul(rotate45).mul(rotate45).mul(x).toString()
'-0.7071067811865474x + 0.7071067811865474y'
You can run the unit-tests with:
$ bin/grunt test
Or if you run bin/grunt
in a console, it will by default run tests
and lint the code on each file save.
Debugging test output is much easier in a browser - the same tests will render in your favourite browser simply by opening the tests/index.html file.
This module uses the excellent RequireJS library to keep module loading sane. You can view the tests/main.js which is loaded by the tests/index.html for an example. With a similar setup you can use the multivector module with:
require(['multivector'], function(mv) {
var myvec = new mv.MultiVector("x + y");
// etc.
});
Most formal definitions of geometric algebra define the geometric product as a combination of an inner and outer product. This implementation doesn't define these products separately (although they can be derived), instead it assumes only two axioms:
Given any set of basis vectors (technically, unit-length orthogonal vectors):
- The product of a basis vector with itself is unity (ie. xx == 1 == yy).
- The product of two different basis vectors is anti-communative (ie. xy == - yx).
With only these two assumptions the library can demonstrate other properties of geometric algebra, such as the associativity of the geometric product for vectors:
var v3 = new mv.MultiVector("-5x - 1y");
v1.mul(v2.mul(v3)).toString()
'-41x - 3y'
(v1.mul(v2)).mul(v3).toString()
'-41x - 3y'
But these two assumptions also ensure that an outer product is not associative - just like the traditional cross-product of two vectors - but unlike all definitions that I've seen of Geometric Algebra, which define the outer product as an associative product:
- Imaginary numbers are not real: The Geometric Algebra of Spacetime: See equation 2.1 of 'The Geometric Algebra of Spacetime' which is prefixed with 'By construction, the outer product is associative'.
- Geometric Algebra and Applications to Physics: Equation 1.22 then scroll back to the start of the derivation at equation 1.16a (the first one, there are two equation 1.16as on page 15) where it seems to assume the associativity of the outer product.
Consider yourself warned - this geometricalgebra.js lib could be crack.