⬆️ Algorithmic Bias Model steady state iteration method

ndlib/models/opinions/AlgorithmicBiasModel.py

@@ -2,6 +2,7 @@
 import numpy as np
 import future.utils
 from collections import defaultdict
+import tqdm
 
 __author__ = ["Alina Sirbu", "Giulio Rossetti"]
 __email__ = ["alina.sirbu@unipi.it", "giulio.rossetti@isti.cnr.it"]
@@ -178,3 +179,34 @@ def iteration(self, node_status=True):
         else:
             return {"iteration": self.actual_iteration - 1, "status": {},
                     "node_count": node_count.copy(), "status_delta": status_delta.copy()}
+
+    def steady_state(self, max_iterations, nsteady=1000, sensibility=0.00001, node_status=True):
+        """
+        Execute a bunch of model iterations
+
+        :param max_iterations: the maximum number of iterations to execute
+        :param nsteady: number of required stable states
+        :param sensibility: sensibility check for a steady state
+        :param node_status: if the incremental node status has to be returned.
+
+        :return: a list containing for each iteration a dictionary {"iteration": iteration_id, "status": dictionary_node_to_status}
+        """
+        system_status = []
+        steady_it = 0
+        for it in tqdm.tqdm(range(0, max_iterations)):
+            its = self.iteration(node_status)
+
+            if it > 0:
+                old = np.array(list(system_status[-1]['status'].values()))
+                actual = np.array(list(its['status'].values()))
+                res = np.abs(old-actual)
+                if np.all((res < sensibility)):
+                    steady_it += 1
+                else:
+                    steady_it = 0
+
+            system_status.append(its)
+            if steady_it == nsteady:
+                return system_status[:-nsteady]
+
+        return system_status

ndlib/test/test_ndlib.py

@@ -19,9 +19,9 @@
 
 def get_graph(er=False):
     if not er:
-        g = nx.complete_graph(250)
+        g = nx.complete_graph(100)
     else:
-        g = nx.erdos_renyi_graph(250, 0.1)
+        g = nx.erdos_renyi_graph(1000, 0.1)
     gi = ig.Graph(directed=False)
     gi.add_vertices(list(g.nodes()))
     gi.add_edges(list(g.edges()))
@@ -129,15 +129,19 @@ def test_utldr(self):
 
     def test_algorithmic_bias_model(self):
 
-        for g in get_graph(er=True):
+        for g in get_graph():
             model = opn.AlgorithmicBiasModel(g, seed=0)
             config = mc.Configuration()
             config.add_model_parameter("epsilon", 0.32)
             config.add_model_parameter("gamma", 1)
             model.set_initial_status(config)
+            iterations = model.iteration_bunch(10)
+            self.assertEqual(len(iterations), 10)
             iterations = model.iteration_bunch(100, node_status=False)
             self.assertEqual(len(iterations), 100)
 
+            _ = model.steady_state(max_iterations=100)
+
     def test_voter_model(self):
         for g in get_graph():
             model = opn.VoterModel(g, seed=0)