release 0.0.2

Author: pedroramaciotti

cloudtropy/entropy_functions.py

@@ -5,7 +5,7 @@
 def testfunc():
     return 1;
 
-def entropy(X,base=2,N=None,d=None,delta_c=None):
+def entropy(X,base=2,N=None,d=None,delta_c=None,lims=None):
     """
     Take a cloud X of P points in Q dimensions (i.e., X.shape->(P,Q)) and compute
     the entropy of a  discrete-support probability mass function on a Q-dimensional grid.
@@ -44,5 +44,5 @@ def entropy(X,base=2,N=None,d=None,delta_c=None):
     Journal of Physics Communications 3, no. 9 (2019): 095011.
     """
 
-    p = pmf(X,N=N,d=d,delta_c=delta_c,return_grid=False)
+    p = pmf(X,N=N,d=d,delta_c=delta_c,return_grid=False,lims=lims)
     return entpy(np.ravel(p),base=base);

cloudtropy/pmfs.py

@@ -2,7 +2,7 @@
 from sklearn.metrics import pairwise_distances
 
 
-def pmf(X,N=None,d=None,delta_c=None,return_grid=True):
+def pmf(X,N=None,d=None,delta_c=None,return_grid=True,lims=None):
 	"""
     Take a cloud X of P points in Q dimensions (i.e., X.shape->(P,Q)) and compute
 	a discrete-support probability mass function on a Q-dimensional grid.
@@ -25,6 +25,9 @@ def pmf(X,N=None,d=None,delta_c=None,return_grid=True):
 		If not given, computed as the greatest width divided by N.
 	return_grid : bool, default=True
         If True, return the support of the N**Q grid points.
+	lims : array of 2-tuples
+		Spatial limits of the computed probability distribution
+		function. Must have Q 2-tuples
 
     Returns
     -------
@@ -52,30 +55,36 @@ def pmf(X,N=None,d=None,delta_c=None,return_grid=True):
 	    raise ValueError('N and d parameters cannot be given at the same time.')
 	# if (d is None) and (N is None):
 	#     raise ValueError('Either N or d must be given as inputs.')
+	if lims is not None:
+		if len(list(lims))!=X.shape[1]:
+			raise ValueError('lims must contain the same number of 2-tuples as the dimensions of the points in the cloud (%d).'%X.shape[1])
+		if any([len(tuple)!=2 for tuple in lims]):
+			raise ValueError('lims must only contain 2-tuples.')
+
+	# Retrieving parameters of the points cloud
+	(N_points,dims) = X.shape
+	# N_points = X.shape[0]
+	# dims = X.shape[1]
 
 	# Treating input
+	if lims is None:
+		lims = []
+		for dim in range(dims):
+		    lims.append( (np.min(X[:,dim]),np.max(X[:,dim])) ) 
 	if (d is not None) and (N is None):
-		N = np.max( (X.max(axis=1)-X.min(axis=1))/d )
+		N = int(np.ceil(np.max([np.abs(lim[1]-lim[0]) for lim in lims]) / d ))
 	if (d is None) and (N is None):
 	    N=20
 	if delta_c is None:
 		delta_c = np.max( (X.max(axis=1)-X.min(axis=1)))/N
 
-	# Retrieving parameters of the points cloud
-	(N_points,dims) = X.shape
-	# N_points = X.shape[0]
-	# dims = X.shape[1]
-
 	# Generating grid
-	lims = []
-	for dim in range(dims):
-	    lims.append( (np.min(X[:,dim]),np.max(X[:,dim])) )   
 	dim_arrays = []
 	for dim in range(dims):
-	    if d is not None:
-	        dim_arrays.append(np.arange(start=lims[dim][0],stop=lims[dim][1]+d,step=d))
-	    elif N is not None:
-	        dim_arrays.append(np.linspace(start=lims[dim][0],stop=lims[dim][1],num=N))
+	    # if d is not None:
+	    #     dim_arrays.append(np.arange(start=lims[dim][0],stop=lims[dim][1]+d,step=d))
+	    # elif N is not None:
+	    dim_arrays.append(np.linspace(start=lims[dim][0],stop=lims[dim][1],num=N))
 	grid = np.meshgrid(*dim_arrays)
 	# Creating the points of the grid
 	Y = np.vstack(map(np.ravel, grid)).transpose()
@@ -86,7 +95,7 @@ def pmf(X,N=None,d=None,delta_c=None,return_grid=True):
 	# Computing the PMF
 	flat_pmf = (pd<delta_c).sum(axis=0)
 	flat_pmf = flat_pmf/flat_pmf.sum()
-	pmf = np.reshape(flat_pmf,newshape=[N for i in range(dims)])
+	pmf = np.reshape(flat_pmf,newshape=[int(N) for i in range(dims)])
 
 	if return_grid:
 		return grid, pmf;

setup.py

@@ -13,13 +13,13 @@
 ]
 
 setup(name='cloudtropy',
-      version='0.0.1',
+      version='0.0.2',
       description='Empirical probability mass functions and entropies of N-dimensional clouds of points.',
       author='Pedro Ramaciotti Morales',
       author_email='[email protected]',
       url = 'https://github.com/pedroramaciotti/Cloudtropy',
       download_url = 'https://github.com/pedroramaciotti/Cloudtropy/archive/0.0.1.tar.gz',
-      keywords = ['entropy','probabilities'],
+      keywords = ['entropy','probabilities','entropy of points'],
       packages=find_packages(),
       data_files=[('', ['LICENSE'])],
       install_requires=INSTALL_REQUIRES)

tests/all.png

@@ -0,0 +1,10 @@
+import numpy as np
+import matplotlib.pyplot as plt
+
+import sys
+sys.path.append('..')
+import cloudtropy
+
+##################
+# Simulated data #
+##################

tests/contour_simple.png

@@ -1,5 +1,4 @@
 import numpy as np
-from sklearn.metrics import pairwise_distances
 import matplotlib.pyplot as plt
 
 from mpl_toolkits import mplot3d
@@ -8,7 +7,7 @@
 
 
 import sys
-sys.path.append('..')
+sys.path.append('../')
 import cloudtropy
 
 
@@ -19,7 +18,7 @@
 lims = (-2,6)
 
 scale = 0.2
-X = np.random.uniform(low=lims[0],high=lims[1],size=(10000,2)) # back
+X = np.random.uniform(low=lims[0],high=lims[1],size=(10000,2)) # background
 X = np.concatenate([X,scale*np.random.randn(gen_N,gen_dim)+np.array([0,0])] )
 X = np.concatenate([X,scale*np.random.randn(gen_N,gen_dim)+np.array([4,0])] )
 X = np.concatenate([X,scale*np.random.randn(gen_N,gen_dim)+np.array([0,4])] )
@@ -31,10 +30,13 @@
 N_grid = 80
 delta_c = 0.35
 
-grid,pmf = cloudtropy.pmf(X,N=N_grid,delta_c=delta_c)
-entropy = cloudtropy.entropy(X,base=2,N=N_grid,delta_c=delta_c)
+# grid,pmf = cloudtropy.pmf(X,N=N_grid,delta_c=delta_c,lims=[(-2,6),(-2,6)])
+grid,pmf = cloudtropy.pmf(X,d=0.1,delta_c=delta_c,lims=[(-2,6),(-2,6)])
+entropy = cloudtropy.entropy(X,base=2,N=N_grid,delta_c=delta_c,lims=[(-3,7),(-3,7)])
+
+
+print(cloudtropy.entropy(X,base=2,d=0.1,delta_c=delta_c,lims=[(-3,7),(-3,7)]))
 
-entropy_readme = cloudtropy.entropy(X)
 
 ############## All in one
 

tests/parameters_test.py

@@ -0,0 +1,36 @@
+import numpy as np
+import matplotlib.pyplot as plt
+
+from mpl_toolkits import mplot3d
+
+# from scipy.stats import entropy
+
+
+import sys
+sys.path.append('..')
+import cloudtropy
+
+
+# data
+
+gen_dim = 2
+gen_N = 300
+lims = (-2,6)
+
+scale = 0.2
+X = np.random.uniform(low=lims[0],high=lims[1],size=(10000,2)) # background
+X = np.concatenate([X,scale*np.random.randn(gen_N,gen_dim)+np.array([0,0])] )
+X = np.concatenate([X,scale*np.random.randn(gen_N,gen_dim)+np.array([4,0])] )
+X = np.concatenate([X,scale*np.random.randn(gen_N,gen_dim)+np.array([0,4])] )
+X = np.concatenate([X,scale*np.random.randn(gen_N,gen_dim)+np.array([4,4])] )
+
+
+# input parameters
+
+N_grid = 80
+delta_c = 0.35
+
+grid,pmf = cloudtropy.pmf(X,N=N_grid,delta_c=delta_c)
+entropy = cloudtropy.entropy(X,base=2,N=N_grid,delta_c=delta_c)
+
+entropy_readme = cloudtropy.entropy(X)