Rock-solid code demo

This repo is a demo of best practices for reliable scientific code as described in a series of blog posts.

The code in this repo implements Gaussian elimination without pivoting. Given a square matrix, it constructs factors L and U that are lower-triangular and strictly-upper-triangular (respectively) such that LU = A. The algorithm is not numerically stable and should not be used for large matrices due to the tendency for roundoff errors to be amplified. The implementation is also very slow, since it uses Python loops.

Dependencies

This code requires Python and numpy. If you have Python and pip, you can get numpy via

pip install numpy

Installation

To use this code, you should first copy it to your computer via

git clone https://github.com/ketch/rock-solid-code-demo.git

Usage

Then you can invoke it from the IPython prompt via

    cd rock-solid-code-demo
    import factor
    import numpy as np
    A = np.random.rand(3,3)
    L, U = factor.LU(A)