added image_analysis code

Author: saforem2

.gitignore

@@ -4,6 +4,7 @@ data_copy/*
 data_01_11_0549/*
 worm_algorithm/src/worm_ising_2d
 worm_algorithm/5itergraphter.ipynb
+worm_algorithm/image_analysis/data/*
 
 *.pyc
 *.tex

worm_algorithm/count_bonds.py

@@ -65,6 +65,9 @@ def __init__(self, L, block_val=None, num_blocks=10,
                 )
             else:
                 self._save_dir = save_dir
+        if not os.path.exists(self._save_dir):  # ensure save_dir exists
+            os.makedirs(self._save_dir)         # create directory if not
+
         config_files = os.listdir(self._data_dir)
         self._config_files = sorted([
             self._data_dir + i for i in config_files if i.endswith('.txt')
@@ -182,11 +185,26 @@ def count_bonds(self):
             self.bond_stats[key] = [val[0], err[0], val[1], err[1]]
             del(data)
 
+    #  def _save_batch(self, data):
+        """Save in progress bond_stats data to .txt file."""
+        #  save_file = self._save_dir + f'bond_stats_{self._L}_in_progress.txt'
+        #  save_file_copy = save_file + '.bak'
+        #  orig_exists = os.path.exists(save_file)
+        #  copy exists = os.path.exists(save_file_copy)
+        #  if orig exists:
+        #      shutil.copy(save_file, save_file_copy)
+        #      if copy_exists:
+        #          os.remove(save_file)
+        #  ordered_bond_stats = OrderedDict(sorted(self.bond_stats.items(),
+        #                                          key=lambda t: t[0]))
+        #  with open(save_file, 'w') as f:
+        #      for key, val in ordered_bond_stats.items():
+        #          f.write(f'{key} {val[0]} {val[1]} {val[2]} {val[3]}')
+
+
 
     def _save(self):
         """Save bond_stats data to .txt file."""
-        if not os.path.exists(self._save_dir):
-            os.makedirs(self._save_dir)
         save_file = self._save_dir + 'bond_stats_{}.txt'.format(self._L)
         save_file_copy = save_file + '.bak'
         orig_exists = os.path.exists(save_file)

worm_algorithm/generate_data.ipynb

@@ -0,0 +1,82 @@
+import numpy as np
+import matplotlib.pyplot as plt
+import scipy.misc
+import math
+import itertools
+
+def show(image):
+    fig, ax = plt.subplots()
+    img = ax.imshow(image, cmap='Greys', aspect=1, interpolation='none')
+    plt.colorbar(img)
+    return fig, ax
+
+class Image(object):
+    """Class to identify boundaries between black and white pixels in image.
+    """
+    def __init__(self, image_file):
+        #self.cutoff = cutoff
+        try:
+            self._image_orig = scipy.misc.imread(image_file, "mode=L")
+        except FileNotFoundError:
+            raise f"Unable to load from {image_file}"
+        self._Nx, self._Ny = self._image_orig.shape
+        self._image_flat = self._image_orig.flatten()
+        self._image = np.reshape(1 - self._image_flat / 255.0,
+                                 (self._Nx, self._Ny))
+        self._image_cropped = None
+        self._image_bw = None
+        self._image_links = None
+        #self._original = self._image.copy()
+
+    #  def show(self, image=None):
+    #      """Create figure containing image, return figure and axes instances."""
+    #      if image is None:
+    #          image = self._image
+    #      fig, ax = plt.subplots()
+    #      img = ax.imshow(image, cmap='Greys', aspect=1,
+    #                      interpolation='none')
+    #      plt.colorbar(img)
+    #      return fig, ax
+
+    def crop(self, x_start, y_start, x_end, y_end):
+        """Crop image from (x_start, y_start) to (x_end, y_end)."""
+        cropped_image = self._image[x_start:x_end, y_start:y_end]
+        self._image_cropped = cropped_image
+        #self._image = cropped_image
+        return cropped_image
+
+    def _cutoff(self, cutoff, image=None):
+        """Convert image from greyscale to exclusively black and white,
+        determined by the value of cutoff."""
+        if image is None:
+            image = self._image
+        image_flat = np.ndarray.flatten(image)
+        Nx, Ny = image.shape
+        #n_sites = image.shape[0] * image.shape[1]
+        image_bw = np.array([-1 if i < cutoff else 1 for i in image_flat])
+        image_bw = (np.reshape(image_bw, (Nx, Ny)) + 1) / 2.0
+        self._image_bw = image_bw
+        return image_bw
+
+    def get_boundaries(self, cutoff, image=None):
+        """Identify boundaries separating black and white regions of image."""
+        if image is None:
+            image = self._image
+        image_bw = self._cutoff(cutoff, image)
+        Nx, Ny = image_bw.shape
+        links_arr = np.zeros((Nx, Ny, 2))
+        for x, y in itertools.product(range(Nx), range(Ny)):
+            links_arr[(x+1)%Nx, y, 1] = (int(round(image_bw[x, y]))
+                                         + int(round(image_bw[(x+1)%Nx, y])))%2
+            links_arr[x, (y+1)%Ny, 0] = (int(round(image_bw[x, y]))
+                                         + int(round(image_bw[x, (y+1)%Ny])))%2
+        image_links = np.zeros((2*Nx, 2*Ny), dtype=int)
+        for x, y in itertools.product(range(Nx), range(Ny)):
+            image_links[2*x+1, 2*y] = links_arr[x, y, 0]
+            image_links[2*x, 2*y+1] = links_arr[x, y, 1]
+            link_sum = (links_arr[x, y, 0] + links_arr[x, y, 1]
+                        + links_arr[(x-1)%Nx, y, 0] + links_arr[x, (y-1)%Ny, 1])
+            if link_sum != 0:
+                image_links[2*x, 2*y] = 1
+        self._image_links = image_links
+        return image_links

worm_algorithm/image_analysis/__init__.py

@@ -0,0 +1,23 @@
+import os
+import subprocess
+import sys
+import argparse
+
+def download_mnist():
+    """Download mnist dataset into 'mnist/' directory."""
+    current_dir = os.getcwd()
+    mnist_data_dir = '../data/mnist/'
+    if not os.path.exists(mnist_data_dir):
+        os.makedirs(mnist_data_dir)
+    os.chdir(mnist_data_dir)
+    os.system('curl -O http://deeplearning.net/data/mnist/mnist.pkl.gz')
+    os.chdir(current_dir)
+    mnist_data_file = mnist_data_dir.join('mnist.pkl.gz')
+    if os.path.isfile(mnist_data_file):
+        print(f"Successfully downloaded mnist data file to: {mnist_data_file}")
+
+
+if __name__ == '__main__':
+    download_mnist()
+
+