lagrangian function convex_optimization

1 files changed, 11 insertions, 11 deletions

diff --git a/jpa_test/algorithms.py b/jpa_test/algorithms.py
index 0d5f154..5d28300 100644
--- a/jpa_test/algorithms.py
+++ b/jpa_test/algorithms.py
@@ -35,11 +35,11 @@ def greedy_prediction(G, cascades):
     return G_hat
 
 
-def recovery_l1obj_l2constraint(G, cascades):
+def recovery_l1obj_l2constraint(G, cascades, floor_cstt, passed_function,
+                                *args, **kwargs):
     """
-    Returns estimated graph from following convex program:
-    min |theta_1| + lbda | exp(M theta) -(1- w)|
-    where theta = log (1 - p); w = 1_{infected}; lbda = lagrange cstt
+    Returns estimated graph from convex program specified by passed_function
+    passed_function should have similar structure to ones in convex_optimation
     """
     G_hat = cascade_creation.InfluenceGraph(max_proba=None)
     G_hat.add_nodes_from(G.nodes())
@@ -47,9 +47,9 @@ def recovery_l1obj_l2constraint(G, cascades):
         print node
         try:
             M, w = cascade_creation.icc_matrixvector_for_node(cascades, node)
-            p_node, __ = convex_optimization.l1obj_l2constraint(M,w)
+            p_node, __ = passed_function(M,w, *args, **kwargs)
             G_hat = cascade_creation.add_edges_from_proba_vector(G=G_hat,
-                        p_node=p_node, node=node, floor_cstt=.01)
+                        p_node=p_node, node=node, floor_cstt=floor_cstt)
         except timeout.TimeoutError:
             print "TimeoutError, skipping to next node"
     return G_hat
@@ -75,14 +75,14 @@ def test():
     """
     unit test
     """
-    G = cascade_creation.InfluenceGraph(max_proba = .5)
+    G = cascade_creation.InfluenceGraph(max_proba = .8)
     G.erdos_init(n = 100, p = .3)
     import time
     t0 = time.time()
-    A = cascade_creation.generate_cascades(G, .2, 100)
-
-    G_hat = recovery_l1obj_l2constraint(G, A)
-
+    A = cascade_creation.generate_cascades(G, .2, 10000)
+    G_hat = recovery_l1obj_l2constraint(G, A,
+            passed_function=convex_optimization.l1obj_l2penalization,
+            floor_cstt=.1, lbda=100)
     correctness_measure(G, G_hat, print_values=True)