src.py

import os
import numpy as np
import dmtools
from dmtools import arrange
from assistant.help import get_metadata
import random
random.seed(3699)


def drunk_walk(k:int, x:int, y:int, M:np.ndarray):
    """Do a k-step drunk walk from (x,y) on M.

    Args:
        k (int): number of steps to take.
        x (int): x-coordinate of start location.
        y (int): y-coordinate of start location.
        M (np.ndarray): board on which to walk.
    """
    n,m = M.shape
    for i in range(k):
        move = random.randint(0,3)
        if move == 0:
            x += 1
        if move == 1:
            x -= 1
        if move == 2:
            y += 1
        if move == 3:
            y -= 1
        M[x % n][y % m] += 1
    return M


def drunk_walk_image(n:int, k:int, d:int):
    """Create a d*d drunk walk with n steps on k gradients.

    Args:
        n (int): number of steps to take.
        k (int): number of gradients.
        d (int): dimension of the board on which to walk.
    """
    M = np.zeros((d,d))
    x = random.randint(0,d)
    y = random.randint(0,d)
    M = drunk_walk(n,x,y,M)
    # This implementation makes more sense but not as interesting..
    # M = np.clip(M, 0, k)
    M = np.array(list(map(lambda x: x % k, M)))
    return M.astype(int)


def drunk_walk_series(n:int, k:int, d:int,
                      w:int, h:int, b:int) -> np.ndarray:
    """Create a series of d*d drunk walk with n steps on k gradients.

    Args:
        n (int): number of steps to take.
        k (int): number of gradients.
        d (int): dimension of the board on which to walk.
        w (int): number of boards per row of the grid.
        h (int): number of boards per column of the grid.
        b (int): width of the border/margin.

    Returns:
        np.ndarray: Image of the series of drunk walks.
    """
    images = [drunk_walk_image(n,k,d) / k for i in range(w*h)]
    grid_image = arrange.image_grid(images,w,h,b)
    return grid_image


# COMPILE PIECES | 2021-03-17

pieces = [(4096, 4, 128, 3, 3, 8),
          (4, 4, 2, 3, 3, 1),
          (1024, 4, 32, 3, 3, 4),
          (256, 4, 16, 3, 3, 2),
          (64, 4, 8, 3, 3, 2),
          (16, 4, 4, 3, 3, 1)]

os.makedirs('output', exist_ok=True)
for n, k, d, w, h, b in pieces:
    image = drunk_walk_series(n, k, d, w, h, b)
    path = 'output/%d_step_drunk_walk.pgm' % n
    dmtools.write_netpbm(image, k, path, metadata=get_metadata())