randomize weights: added checking eighted network

netrw/rewire/randomized_weights.py

@@ -0,0 +1,93 @@
+from .base import BaseRewirer
+import copy
+import itertools as it
+import random
+import networkx as nx
+import numpy as np
+
+
+class RandomizedWeightCM_swap(BaseRewirer):
+    """
+    Swap weights of a weighted network without rewiring edges.
+
+    - rewire_step: Swap weights bewteen two randomly chosen edges
+    - rewire: Over the list of edges, permutate the list of associated weigths
+
+    4th method is from Ghavasieh, A.; De Domenico, M.
+    "Multiscale Information Propagation in Emergent Functional Networks".
+    Entropy 2021, 23, 1369. https://doi.org/10.3390/e23101369
+    """
+
+    def edge_pair_random_choice(self, G):
+        e_list = list(G.edges(data=True))
+        e_1 = random.choice(e_list)
+        e_list.remove(e_1)
+        e_2 = random.choice(e_list)
+
+        return e_1, e_2
+
+    def step_rewire(self, G, copy_graph=True):
+        if copy_graph:
+            G = copy.deepcopy(G)
+
+        e_1, e_2 = self.edge_pair_random_choice(G)
+
+        w_1 = e_1[2]["weight"]
+        w_2 = e_2[2]["weight"]
+
+        G.edges[e_1[0], e_1[1]]["weight"] = w_2
+        G.edges[e_2[0], e_2[1]]["weight"] = w_1
+
+        return G
+
+    def full_rewire(self, G, copy_graph=True):
+        if copy_graph:
+            G = copy.deepcopy(G)
+
+        e_list = list(G.edges())
+        w_list = [x[2]["weight"] for x in list(G.edges(data=True))]
+
+        w_list = np.random.permutation(w_list)
+        nx.set_edge_attributes(G, dict(zip(e_list, w_list)), "weight")
+
+        return G
+
+
+class RandomizedWeightCM_redistribution(BaseRewirer):
+    """
+    Redistribute weights of a weighted network without rewiring edges.
+
+    - rewire_step: the total sum of weight of a randomly chosen pair of links is randomly re-distributed over this two links
+    - rewire: The total sum of weights of all links in the netwrok is randomly distributed over the links
+    """
+
+    def step_rewire(self, G, copy_graph=True):
+        if copy_graph:
+            G = copy.deepcopy(G)
+
+        e_1, e_2 = self.edge_pair_random_choice(G)
+
+        a_1 = random.random()
+
+        w_sum = e_1[2]["weight"] + e_2[2]["weight"]
+
+        G.edges[e_1[0], e_1[1]]["weight"] = a_1 * w_sum
+        G.edges[e_2[0], e_2[1]]["weight"] = (1 - a_1) * w_sum
+
+        return G
+
+    def full_rewire(self, G, copy_graph=True):
+        if copy_graph:
+            G = copy.deepcopy(G)
+
+        alphas = np.random.rand(len(G.edges()))
+        alphas = alphas / np.sum(alphas)
+
+        w = [x[2]["weight"] for x in list(G.edges(data=True))]
+        w_sum = np.sum(w)
+        aw = alphas * w_sum
+
+        e_list = list(G.edges())
+        nx.set_edge_attributes(G, dict(zip(e_list, aw)), "weight")
+
+        return G