Cythonizing

Implement a function in Cython that computes x**3 - 4x + 2 on each element of a large floating point array. Is this function faster than NumPy by default? If so, can you explain why?

Benchmarking the NumPy code in IPython:

import numpy as np
x = np.linspace(0, 1, 10000)
%timeit x**3 - 4*x + 2

(do the same for your Cython code)

Wrapping C

In the wrapping directory is a simple C library computing the sin elementwise for a Numpy array. The header stuff.h defines a function with the signature:

void compute(int n, double *input, double *output)

which takes two C arrays of doubles containing n doubles.

Write a Cython wrapper:

def do_compute(double[::1] input_array):
    ...
    return output_array

for this function. Note: whenever ::1 is specified for an axis, it means that that dimension should be contiguous in memory. Also, unless you call np.asarray on your output, a MemoryView object instead of an array will be returned.

You can find the pointer to an array with, e.g., &input_array[0, 0]. E.g., this will yield the equivalent of double *, ready for use in C.

The behaviour of the function can be verified with run.py.

Conway's Game of Life

The Game of Life is a well-known cellular automaton that exhibits interesting behavior. It is defined on a grid of cells, where each cell has 8 neighbors:

.........
.........
...ooo...
...oxo...
...ooo...
.........
.........

The update rule to get a new state from the old one is:

• Cells with with fewer than 2 or more than 3 live neighbors die.
• A live cell with two or three live neighbors lives on to the next generation.
• A dead cell with exactly 3 live neighbors becomes alive.

Write a Cython function life_update(old_state, new_state) that takes an N x N Numpy array old_state of type int containing the old state, and writes the new state to a similar array new_state. Just use four nested for-loops, but remember to add type declarations.

Image files containing well-known starting shapes are supplied (use matplotlib.pyplot.imread to read them into Numpy arrays). Assign them at the center of a big grid, and see what happens!

• glider.png: the glider
• glider_gun.png: the Gosper glider gun
• breeder.png: one sort of a breeder
• ... and others!

These examples come from the Game of Life simulator Golly.

L-Systems

Implement an L-system in Cython. Try, for example, to build a Sierpinski Triangle or the Dragon Curve. If you want, you can first implement it in pure Python, and then add the type information later.