rip_condition

1 files changed, 26 insertions, 15 deletions

diff --git a/jpa_test/cascade_creation.py b/jpa_test/cascade_creation.py
index 09f0c45..a5c38cd 100644
--- a/jpa_test/cascade_creation.py
+++ b/jpa_test/cascade_creation.py
@@ -1,8 +1,8 @@
 import networkx as nx
 import numpy as np
 import collections
-
 from itertools import izip
+from sklearn.preprocessing import normalize
 
 class InfluenceGraph(nx.Graph):
     """
@@ -76,21 +76,32 @@ def icc_matrixvector_for_node(cascades, node):
     Returns M, w in matrix form where rows of M are i = t + k.T
     Excludes all (t,k) after node infection time; w = 1_{infected}
     """
-    w = []
-    M = []
-    for cascade in cascades:
-        t_i = cascade.infection_time(node)[0]
-        if t_i != 0:
-            indicator = np.zeros(len(cascade[:t_i]))
-            if t_i > 0:
-                indicator[-1] = 1
-            w.append(indicator)
-            M.append(np.array(cascade[:t_i]))
-    M = np.vstack(M)
-    w = np.hstack(w)
+    #TODO: you need to remove the variable corresponding to the node you are solving for!!!!
+    if node is None:
+        return np.vstack(cascades), None
+    else:
+        w = []
+        M = []
+        for cascade in cascades:
+            t_i = cascade.infection_time(node)[0]
+            if t_i != 0:
+                indicator = np.zeros(len(cascade[:t_i]))
+                if t_i > 0:
+                    indicator[-1] = 1
+                w.append(indicator)
+                M.append(np.array(cascade[:t_i]))
+        M = np.vstack(M)
+        w = np.hstack(w)
     return M, w
 
 
+def normalize_matrix(M):
+    """
+    returns M with normalized (L_1 norm) columns
+    """
+    return normalize(M.astype("float32"), axis=0, norm="l2")
+
+
 def icc_cascade(G, p_init):
     """
     Returns boolean vectors for one cascade
@@ -99,14 +110,14 @@ def icc_cascade(G, p_init):
     """
     susceptible = np.ones(G.number_of_nodes(), dtype=bool)
     active = np.random.rand(G.number_of_nodes()) < p_init
-    susceptible = susceptible - active
+    susceptible = susceptible & np.logical_not(active)
     cascade = Cascade()
     while active.any():
         cascade.append(active)
         active = np.exp(np.dot(G.logmat, active)) \
                     < np.random.rand(G.number_of_nodes())
         active = active & susceptible
-        susceptible = susceptible - active
+        susceptible = susceptible & np.logical_not(active)
     if not cascade:
         print "Empty cascade, consider changing p_init or n_nodes. Retrying."
         return icc_cascade(G, p_init)