Merge pull request #18 from LaurentRDC/develop

Preparation for 0.4.7 release

.appveyor/install-miniconda.ps1

@@ -5,10 +5,6 @@
 
 $MINICONDA_URL = "https://repo.continuum.io/miniconda/"
 
-$env:ASTROPY_LTS_VERSION = "1.0"
-$env:LATEST_ASTROPY_STABLE = "1.3"
-$env:LATEST_NUMPY_STABLE = "1.12"
-
 # We pin the version for conda as it's not the most stable package from
 # release to release. Add note here if version is pinned due to a bug upstream.
 if (! $env:CONDA_VERSION) {
@@ -110,7 +106,7 @@ if (! $env:CONDA_CHANNEL_PRIORITY) {
 # key may not yet exists
 conda config  --set channel_priority $CONDA_CHANNEL_PRIORITY
 
-# Create a conda environment using the astropy bonus packages
+# Create a conda environment 
 conda create -q -n test python=$env:PYTHON_VERSION
 activate test
 

.gitignore

@@ -9,7 +9,8 @@ __pycache__/
 # Visual studio cache
 *.vs/
 
-# Protein DataBank cache folder
+# Protein DataBank files and cache folder
+*.ent
 pdb_cache/
 
 # Jupyter notebooks

appveyor.yml

@@ -14,15 +14,13 @@ environment:
                         # to the matrix section.
       TEST_CMD: "python -m unittest discover --verbose"
 
+      CONDA_DEPENDENCIES: "numpy scipy cython scikit-image"
+      PIP_DEPENDENCIES: "pywavelets spglib pycifrw"
+
   matrix:
 
       - PYTHON_VERSION: "3.6"
-        CONDA_DEPENDENCIES: "numpy scipy cython scikit-image"
-        PIP_DEPENDENCIES: "pywavelets spglib pycifrw"
-
       - PYTHON_VERSION: "3.5"
-        CONDA_DEPENDENCIES: "numpy scipy cython scikit-image"
-        PIP_DEPENDENCIES: "pywavelets spglib pycifrw"
 
 matrix:
    fast_finish: true

docs/source/tutorials/image.rst

@@ -14,6 +14,7 @@ Contents
 
 * :ref:`streaming`
 * :ref:`alignment`
+* :ref:`symmetry`
 * :ref:`powder`
 
 .. _streaming:
@@ -211,6 +212,58 @@ Let's look at the difference:
 	plt.tight_layout()
 	plt.show()
 
+.. _symmetry:
+
+Image processing involving symmetry
+===================================
+
+Rotational symmetry
+-------------------
+Diffraction patterns exhibit rotational symmetry based on the crystal structure. We can
+take advantage of such symmetry to correct images in case of artifacts or defects. A useful
+routine is :func:`nfold`, which averages portions of a diffraction pattern with itself based on
+rotational symmetry.
+
+.. plot::
+
+    import matplotlib.pyplot as plt
+    from skimage.io import imread
+    from skued.image import nfold
+    import numpy as np
+
+    center = (1010, 1111)
+
+    mask = np.zeros((2048, 2048), dtype = np.bool)
+    mask[1100::, 442:480] = True # Artifact line
+    mask[0:1260, 900:1140] = True # beamblock
+
+    image = imread('graphite.tif')
+    av = nfold(image, mod = 6, center = center, mask = mask)
+
+    fig , (ax1, ax2, ax3) = plt.subplots(1,3, figsize = (9,3))
+    ax1.imshow(image, vmin = 0, vmax = 150)
+    ax2.imshow(mask, vmin = 0, vmax = 1)
+    ax3.imshow(av, vmin = 0, vmax = 150)
+
+    for ax in (ax1, ax2, ax3):
+        ax.xaxis.set_visible(False)
+        ax.yaxis.set_visible(False)
+
+    ax1.set_title('Graphite')
+    ax2.set_title('Mask')
+    ax3.set_title('Averaged')
+
+    plt.tight_layout()
+    plt.show()
+
+To use :func:`nfold`, all you need to know is the center of the diffraction pattern::
+
+    from skued.image import nfold
+    from skimage.io import imread
+
+    im = imread('graphite.tif')
+    av = nfold(im, mod = 6, center = center)    # mask is optional
+
 .. _powder:
 
 Image analysis on polycrystalline diffraction patterns
@@ -223,17 +276,22 @@ the center of those concentric rings is important. Let's load a test image:
 
 .. plot::
 
-	from skimage.io import imread
-	import matplotlib.pyplot as plt
-	path = 'Cr_1.tif'
+    from skimage.io import imread
+    import matplotlib.pyplot as plt
+    path = 'Cr_1.tif'
 
-	im = imread(path, plugin = 'tifffile')
-	mask = np.zeros_like(im, dtype = np.bool)
-	mask[0:1250, 950:1250] = True
+    im = imread(path, plugin = 'tifffile')
+    mask = np.zeros_like(im, dtype = np.bool)
+    mask[0:1250, 950:1250] = True
 
-	im[mask] = 0
-	plt.imshow(im, vmin = 0, vmax = 200)
-	plt.show()
+    im[mask] = 0
+    fig = plt.figure()
+    ax = fig.add_subplot(111)
+    ax.imshow(im, vmin = 0, vmax = 200)
+
+    ax.xaxis.set_visible(False)
+    ax.yaxis.set_visible(False)
+    plt.show()
 
 This is a noisy diffraction pattern of polycrystalline vanadium dioxide. 
 Finding the center of such a symmetry pattern can be done with the 
@@ -249,25 +307,32 @@ Finding the center of such a symmetry pattern can be done with the
 	rr = np.sqrt((ii - ic)**2 + (jj - jc)**2)
 	im[rr < 100] = 0
 
-	plt.imshow(im, vmax = 1200)
+	plt.imshow(im, vmax = 200)
 	plt.show()
 
 .. plot::
 
-	from skimage.io import imread
-	import numpy as np
-	import matplotlib.pyplot as plt
-	path = 'Cr_1.tif'
-	im = imread(path, plugin = 'tifffile')
-	from skued.image import powder_center
-	mask = np.zeros_like(im, dtype = np.bool)
-	mask[0:1250, 950:1250] = True
-	ic, jc = powder_center(im, mask = mask)
-	ii, jj = np.meshgrid(np.arange(im.shape[0]), np.arange(im.shape[1]),indexing = 'ij')
-	rr = np.sqrt((ii - ic)**2 + (jj - jc)**2)
-	im[rr < 100] = 1e6
-	plt.imshow(im, vmin = 0, vmax = 200)
-	plt.show()
+    from skimage.io import imread
+    import numpy as np
+    import matplotlib.pyplot as plt
+    path = 'Cr_1.tif'
+    im = imread(path, plugin = 'tifffile')
+    from skued.image import powder_center
+    mask = np.zeros_like(im, dtype = np.bool)
+    mask[0:1250, 950:1250] = True
+    ic, jc = powder_center(im, mask = mask)
+    ii, jj = np.meshgrid(np.arange(im.shape[0]), np.arange(im.shape[1]),indexing = 'ij')
+    rr = np.sqrt((ii - ic)**2 + (jj - jc)**2)
+    im[rr < 100] = 1e6
+
+    fig = plt.figure()
+    ax = fig.add_subplot(111)
+    ax.imshow(im, vmin = 0, vmax = 200)
+
+    ax.xaxis.set_visible(False)
+    ax.yaxis.set_visible(False)
+
+    plt.show()
 
 Angular average
 ---------------
@@ -295,21 +360,27 @@ First, we create a test image::
 
 .. plot::
 
-	import numpy as np
-	import matplotlib.pyplot as plt
-	from skued import gaussian
-	image = np.zeros( (256, 256) )
-	xc, yc = image.shape[0]/2, image.shape[1]/2	# center
-	extent = np.arange(0, image.shape[0])
-	xx, yy = np.meshgrid(extent, extent)
-	rr = np.sqrt((xx - xc)**2 + (yy-yc)**2)
-	image += gaussian([xx, yy], center = [xc, yc], fwhm = 200)
-	image[np.logical_and(rr < 40, rr > 38)] = 1
-	image[np.logical_and(rr < 100, rr > 98)] = 0.5
-	image /= image.max()	# Normalize max to 1
-	image += np.random.random(size = image.shape)
-	plt.imshow(image)
-	plt.show()
+    import numpy as np
+    import matplotlib.pyplot as plt
+    from skued import gaussian
+    image = np.zeros( (256, 256) )
+    xc, yc = image.shape[0]/2, image.shape[1]/2	# center
+    extent = np.arange(0, image.shape[0])
+    xx, yy = np.meshgrid(extent, extent)
+    rr = np.sqrt((xx - xc)**2 + (yy-yc)**2)
+    image += gaussian([xx, yy], center = [xc, yc], fwhm = 200)
+    image[np.logical_and(rr < 40, rr > 38)] = 1
+    image[np.logical_and(rr < 100, rr > 98)] = 0.5
+    image /= image.max()	# Normalize max to 1
+    image += np.random.random(size = image.shape)
+
+    fig = plt.figure()
+    ax = fig.add_subplot(111)
+    ax.imshow(image)
+
+    ax.xaxis.set_visible(False)
+    ax.yaxis.set_visible(False)
+    plt.show()
 
 
 ... and we can easily compute an angular average::

docs/source/tutorials/structure.rst

@@ -28,8 +28,15 @@ To create an atom, simply provide its element and coordinates::
 
 	copper = Atom(element = 'Cu', coords = [0,0,0])
 
-:class:`Atom` objects are hashable; this means that they can be stored in a :func:`set`. Therefore,
-a list of atoms can be reduced into unique atoms using a :func:`set`.
+Since we are most concerned with atoms in crystals, the coordinates here are assumed to be fractional.
+The real-space position with respect to a :class:`Crystal` or :class:`Lattice` can be accessed using the 
+:meth:`xyz` method::
+
+    from skued.structure import graphite
+    
+    carbon = list(graphite)[-1]
+    fractional = carbon.coords
+    real = carbon.xyz(lattice = graphite)
 
 One important feature of the :class:`Atom` class is the possibility to compute the electrostatic
 potential across meshes::
@@ -66,11 +73,11 @@ After plot formatting:
 
 The :class:`Crystal` Class
 ==========================
-Diffraction experiments relying on the redundancy of crystals to get good experimental signals;
+Diffraction experiments rely on the redundancy of crystals to get good experimental signals;
 hence, handling crystal models is the main feature of the :mod:`skued.structure` subpackage.
 
-Constructing a :class:`Crystal` object from a file or database
---------------------------------------------------------------
+Constructing a :class:`Crystal` object
+--------------------------------------
 Creating a :class:`Crystal` object can be done most easily from a Crystal Information File (CIF, .cif)::
 	
 	from skued.structure import Crystal
@@ -132,7 +139,7 @@ The :class:`Crystal` object provides some interfaces for easy structure manipula
 	    print(atm.element, atm.coords)
 
 Note that iterating over the :attr:`crystal.atoms` attribute may or may not be equivalent to 
-:data:`iter(crystal)`, due to the way crystals are defined.
+:data:`iter(crystal)`, due to the way symmetry operators are defined.
 
 :class:`Crystal` objects also provide interoperability with :mod:`spglib`::
 

setup.py

@@ -4,6 +4,7 @@
 import os
 import re
 from setuptools import setup, find_packages
+from unittest import TestLoader
 #from Cython.Build import cythonize
 
 # To upload to pypi.org:
@@ -33,6 +34,9 @@
 if BASE_PACKAGE not in packages:
     packages.append(BASE_PACKAGE)
 
+def skued_test_suite():
+    return TestLoader().discover('.')
+
 if __name__ == '__main__':
     setup(
         name = 'scikit-ued',
@@ -52,6 +56,7 @@
         data_files = [('skued\\baseline\\data', wavelets)],
         include_package_data = True,
         zip_safe = False,
+        test_suite = 'setup.skued_test_suite', 
         classifiers = ['Environment :: Console',
                        'Intended Audience :: Science/Research',
                        'Topic :: Scientific/Engineering',

skued/image/streaming.py

@@ -45,6 +45,7 @@ def ialign(images, reference = None, mask = None, fill_value = 0.0):
 
 	yield from map(partial(align, reference = reference, mask = mask, fill_value =  fill_value), images)
 
+# TODO: handle NaNs by having array sum_of_weights and not counting NaNs
 def iaverage(images, weights = None):
     """ 
     Streaming average of diffraction images. This is equivalent

skued/image/symmetry.py

@@ -9,10 +9,9 @@
 from warnings import warn
 
 # TODO: out parameter?
-def nfold(im, mod, center = None, mask = None, **kwargs):
+def nfold(im, mod, center = None, mask = None, fill_value = 0.0, **kwargs):
     """ 
     Returns an images averaged according to n-fold rotational symmetry.
-    Keyword arguments are passed to skimage.transform.rotate()
 
     Parameters
     ----------
@@ -27,7 +26,11 @@ def nfold(im, mod, center = None, mask = None, **kwargs):
         Mask of `image`. The mask should evaluate to `True`
         (or 1) on invalid pixels. If None (default), no mask
         is used.
-    
+    fill_value : float, optional
+        In the case of a mask that overlaps with itself when rotationally averaged,
+        the overlapping regions will be filled with this value.
+    kwargs
+        Keyword arguments are passed to skimage.transform.rotate().
 
     Returns
     -------
@@ -57,7 +60,9 @@ def nfold(im, mod, center = None, mask = None, **kwargs):
 
     stack = np.dstack([rotate(im, angle, center = center, **rotate_kwargs) for angle in angles])
     avg = np.nanmean(stack, axis = 2)
-    return np.nan_to_num(avg)
+
+    avg[np.isnan(avg)] = fill_value
+    return avg
 
 def nfold_symmetry(*args, **kwargs):
     warn('nfold_symmetry() is deprecated. Please use nfold in \