Merge branch 'develop'

idtxl/data.py

@@ -34,7 +34,7 @@ class Data():
         >>> data_2.set_data(data_new, 's')
 
     Note:
-        Realisations are stored as attribute 'data'. This can only be set via 
+        Realisations are stored as attribute 'data'. This can only be set via
         the 'set_data()' method.
 
     Args:
@@ -828,16 +828,26 @@ def generate_mute_data(self, n_samples=1000, n_replications=10):
 
         Generate example data and overwrite the instance's current data. The
         network is used as an example the paper on the MuTE toolbox (Montalto,
-        PLOS ONE, 2014, eq. 14). The network consists of five autoregressive
-        (AR) processes with model orders 2 and the following (non-linear)
-        couplings:
+        PLOS ONE, 2014, eq. 14) and was orginially proposed by Baccala &
+        Sameshima (2001). The network consists of five autoregressive (AR)
+        processes with model orders 2 and the following (non-linear) couplings:
 
         0 -> 1, u = 2 (non-linear)
         0 -> 2, u = 3
         0 -> 3, u = 2 (non-linear)
         3 -> 4, u = 1
         4 -> 3, u = 1
 
+        References:
+
+        - Montalto, A., Faes, L., & Marinazzo, D. (2014) MuTE: A MATLAB toolbox
+          to compare established and novel estimators of the multivariate
+          transfer entropy. PLoS ONE 9(10): e109462.
+          https://doi.org/10.1371/journal.pone.0109462
+        - Baccala, L.A. & Sameshima, K. (2001). Partial directed coherence: a
+          new concept in neural structure determination. Biol Cybern 84:
+          463–474. https://doi.org/10.1007/PL00007990
+
         Args:
             n_samples : int
                 number of samples simulated for each process and replication

idtxl/estimators_pid.py

@@ -5,7 +5,6 @@
 
 Bertschinger, N., Rauh, J., Olbrich, E., Jost, J., & Ay, N. (2014). Quantifying
 Unique Information. Entropy, 16(4), 2161–2183. http://doi.org/10.3390/e16042161
-
 """
 import numpy as np
 from . import synergy_tartu
@@ -17,9 +16,9 @@
 class SydneyPID(Estimator):
     """Estimate partial information decomposition of discrete variables.
 
-    Fast implementation of the partial information decomposition (PID)
-    estimator for discrete data. The estimator does not require JAVA or GPU
-    modules to run.
+    Fast implementation of the BROJA partial information decomposition (PID)
+    estimator for discrete data (Bertschinger, 2014). The estimator does not
+    require JAVA or GPU modules to run.
 
     The estimator finds shared information, unique information and
     synergistic information between the two inputs s1 and s2 with respect to
@@ -34,6 +33,12 @@ class SydneyPID(Estimator):
     CMI decreases; and an outer loop which decreases the size of the
     probability mass increment the virtualised swapping utilises.
 
+    References
+
+    - Bertschinger, N., Rauh, J., Olbrich, E., Jost, J., & Ay, N. (2014).
+      Quantifying unique information. Entropy, 16(4), 2161–2183.
+      http://doi.org/10.3390/e16042161
+
     Args:
         settings : dict
             estimation parameters

idtxl/idtxl_io.py

@@ -375,7 +375,7 @@ class for directed graphs (DiGraph). Multiple options for the weight are
     available (see documentation of method get_adjacency_matrix for details).
 
     Args:
-        adjacency_matrix : 2D numpy array
+        adjacency_matrix : AdjacencyMatrix instances
             adjacency matrix to be exported, returned by get_adjacency_matrix()
             method of Results() class
         weights : str
@@ -389,9 +389,9 @@ class for directed graphs (DiGraph). Multiple options for the weight are
     """
     # use 'weights' parameter (string) as networkx edge property name and use
     # adjacency matrix entries as edge property values
-    custom_type = [(weights, type(adjacency_matrix[0, 0]))]
-    custom_npmatrix = np.matrix(adjacency_matrix, dtype=custom_type)
-    return nx.from_numpy_matrix(custom_npmatrix, create_using=nx.DiGraph())
+    G = nx.DiGraph()
+    G.add_weighted_edges_from(adjacency_matrix.get_edge_list(), weights)
+    return G
 
 
 def export_networkx_source_graph(results, target, sign_sources=True, fdr=True):
@@ -410,7 +410,7 @@ def export_networkx_source_graph(results, target, sign_sources=True, fdr=True):
         target : int
             target index
         sign_sources : bool [optional]
-            add only sources with significant information contribution
+            add sources with significant information contribution only
             (default=True)
         fdr : bool [optional]
             return FDR-corrected results (default=True)
@@ -494,7 +494,7 @@ def export_brain_net_viewer(adjacency_matrix, mni_coord, file_name, **kwargs):
       https://doi.org/10.1371/journal.pone.0068910
 
     Args:
-        adjacency_matrix : 2D numpy array
+        adjacency_matrix : AdjacencyMatrix instance
             adjacency matrix to be exported, returned by get_adjacency_matrix()
             method of Results() class
         mni_coord : numpy array
@@ -516,15 +516,13 @@ def export_brain_net_viewer(adjacency_matrix, mni_coord, file_name, **kwargs):
     """
     # Check input and get default settings for plotting. The default for
     # node labels is a list of '-' (no labels).
-    n_nodes = adjacency_matrix.shape[0]
-    n_edges = np.sum(adjacency_matrix > 0)
+    n_nodes = adjacency_matrix.n_nodes()
+    n_edges = adjacency_matrix.n_edges()
     labels = kwargs.get('labels', ['-' for i in range(n_nodes)])
     node_color = kwargs.get('node_color', np.ones(n_nodes))
     node_size = kwargs.get('node_size', np.ones(n_nodes))
     if n_edges == 0:
         Warning('No edges in results file. Nothing to plot.')
-    assert adjacency_matrix.shape[0] == adjacency_matrix.shape[1], (
-        'Adjacency matrix must be quadratic.')
     assert mni_coord.shape[0] == n_nodes and mni_coord.shape[1] == 3, (
         'MNI coordinates must have shape [n_nodes, 3].')
     assert len(labels) == n_nodes, (
@@ -551,6 +549,6 @@ def export_brain_net_viewer(adjacency_matrix, mni_coord, file_name, **kwargs):
     with open('{0}.edge'.format(file_name), 'w') as text_file:
         for i in range(n_nodes):
             for j in range(n_nodes):
-                print('{0}\t'.format(adjacency_matrix[i, j]),
+                print('{0}\t'.format(adjacency_matrix._edge_matrix[i, j]),
                       file=text_file, end='')
             print('', file=text_file)

idtxl/network_comparison.py

@@ -129,8 +129,8 @@ def compare_links_within(self, settings, link_a, link_b, network, data):
             settings=self.settings,
             union_network=self.union,
             results={
-                'cmi_diff_abs': {link_a[1]: [np.abs(self.cmi_diff)],
-                                 link_b[1]: [np.abs(self.cmi_diff)]},
+                'cmi_diff_abs': {link_a[1]: np.abs(self.cmi_diff),
+                                 link_b[1]: np.abs(self.cmi_diff)},
                 'a>b': {link_a[1]: [te_a > te_b], link_b[1]: [te_a > te_b]},
                 'pval': {link_a[1]: [pvalue], link_b[1]: [pvalue]},
                 'cmi_surr': self.cmi_surr,