First commit of MAMMULT code

4 files changed, 308 insertions, 0 deletions

diff --git a/models/nullmodels/model_MDM.py b/models/nullmodels/model_MDM.py
new file mode 100644
index 0000000..5b0a969
--- /dev/null
+++ b/models/nullmodels/model_MDM.py
@@ -0,0 +1,66 @@
+####
+##
+##
+## This is the vertical participation model. For each node i, we use
+## exactly the same value of B_i as in the original network, but we
+## choose at random the layers in which node i will be active. This
+## breaks down intra-layer correlations. 
+##
+## We get as input a file which reports, for each value of B_i, the
+## number of nodes in the original network which have that value, i the format:
+##
+## B_i N(B_i)
+##
+##
+##
+## The output is the obtained distribution of bit-strings.
+##
+##
+
+import sys
+import random
+
+
+def to_binary(l):
+    s = 0
+    e = 0
+    for v in l:
+        s += v * pow(2,e)
+        e +=1
+    return s
+
+
+if len(sys.argv) < 3:
+    print "Usage: %s <Bi_file> <M>" % sys.argv[0]
+    sys.exit(1)
+
+M = int(sys.argv[2])
+
+layers = range(M)
+
+distr = {}
+
+with open(sys.argv[1], "r") as f:
+    for l in f:
+        if l[0] == "#":
+            continue
+        val, num = [int(x) for x in l.strip(" \n").split(" ")]
+        for j in range(num):
+            node_layers = random.sample(layers, val)
+            node_bitstring = [0 for x in range(M)]
+            #print node_bitstring, node_layers
+            for i in node_layers:
+                #print i,
+                node_bitstring[i] = 1
+                #print node_bitstring
+                
+            bs = to_binary(node_bitstring)
+            if bs in distr:
+                distr[bs] += 1
+            else:
+                distr[bs] = 1
+
+for k in distr:
+    bin_list = bin(k)
+    bin_num = sum([int(x) if x=='1' else 0 for x in bin_list[2:]])
+    sys.stderr.write("%d %0175s %d \n" % (bin_num, bin_list[2:], distr[k]))
diff --git a/models/nullmodels/model_MSM.py b/models/nullmodels/model_MSM.py
new file mode 100644
index 0000000..2c30df5
--- /dev/null
+++ b/models/nullmodels/model_MSM.py
@@ -0,0 +1,65 @@
+####
+##
+##
+## Create a synthetic multiplex network in which a node $i$ appears at
+## each layer $\alpha$ with a probability equal to $B_i$, which is the
+## fraction of layers in which node $i$ participate in the original
+## multiplex.
+##
+## Take a file of node binary participation indices, and sample a
+## multiplex compatible with that distribution
+##
+##
+## The input file is in the format:
+## 
+## nodeID_i B_i
+##
+## The output file is a node-layer participation file, in the format
+##
+## node_id1 layer_id1
+## node_id2 layer_id2
+## .....
+##
+
+import sys
+
+import random
+
+if len(sys.argv) < 3:
+    print "Usage: %s <filein> <M>" % sys.argv[0]
+    sys.exit(1)
+
+M = int(sys.argv[2])
+
+bin_index = {}
+node_ids = []
+
+lines = open(sys.argv[1]).readlines()
+
+
+for l in lines:
+    if l[0] == "#":
+        continue
+    elems = [int(x) for x in l.strip(" \n").split(" ")]
+    bin_index[elems[0]] = 1.0 * elems[1]/M
+    node_ids.append(elems[0])
+
+N = len(node_ids)
+
+node_layers = {}
+
+for alpha in range (M):
+    for i in node_ids:
+        val = random.random()
+        if val < bin_index[i]:
+            if  node_layers.has_key(i):
+                node_layers[i].append(alpha)
+            else:
+                node_layers[i] = [alpha]
+
+
+for i in node_ids:
+    if node_layers.has_key(i):
+        for j in range(len(node_layers[i])):
+            print i, node_layers[i][j]
+
diff --git a/models/nullmodels/model_hypergeometric.py b/models/nullmodels/model_hypergeometric.py
new file mode 100644
index 0000000..0a36237
--- /dev/null
+++ b/models/nullmodels/model_hypergeometric.py
@@ -0,0 +1,56 @@
+####
+##
+## This is the hypergeometric model. Each layer has the same number of
+## non-isolated nodes as the initial graph, but the nodes are
+## activated at random. The input is a file of number of non-isolated
+## nodes in each layer, and the total number of nodes in the multiplex.
+##
+## The output file is a node-layer participation file, in the format
+##
+## node_id1 layer_id1
+## node_id2 layer_id2
+## .....
+##
+
+import sys
+import random
+
+if len(sys.argv) < 3:
+    print "Usage: %s <layer_N_file> <N>" % sys.argv[0]
+    sys.exit(1)
+
+N = int(sys.argv[2])
+
+layer_degs = []
+node_layers = {}
+
+lines = open(sys.argv[1]).readlines()
+
+M = 0
+
+
+for l in lines:
+    if l[0] == "#":
+        continue
+    n = [int(x) for x in l.strip(" \n").split(" ")][0]
+    layer_degs.append(n)
+    M += 1
+
+for i in range(M):
+    num = layer_degs[i]
+    added = []
+    n = 0
+    while n < num:
+        j = random.choice(range(N))
+        if j not in added:
+            n += 1
+            added.append(j)
+            if node_layers.has_key(j):
+                node_layers[j].append(i)
+            else:
+                node_layers[j] = [i]
+        
+for n in node_layers.keys():
+    for i in node_layers[n]:
+        print n,i
+
diff --git a/models/nullmodels/model_layer_growth.py b/models/nullmodels/model_layer_growth.py
new file mode 100644
index 0000000..46b4c0f
--- /dev/null
+++ b/models/nullmodels/model_layer_growth.py
@@ -0,0 +1,121 @@
+####
+##
+## layer-by-layer multiplex growth
+##
+## - We start from a multiplex with M_0 layers, with a certain number of 
+##   active nodes each
+##
+## - Each new layer arrives with a certain number N\lay{\alpha} of nodes
+##   to be activated (this number is sampled from the observed distribution 
+##   of N\lay{\alpha}, like in the airlines multiplex)
+## 
+## - Each node $i$ is activated with a probability proportional to the
+##   number of existing layers in which it is already active, added to an
+##   attractivity A :
+##
+## P_i(t) \propto A + B_i(t)
+##
+## - the hope is that A might tune the exponent of the resulting distribution
+##   of B_i.....
+## 
+##
+## This script takes as input a file which contains the degrees of the
+## layers, the total number of nodes in the multiplex, the initial
+## number M0 of layers in the multiplex and the attractiveness
+## parameter A. If "RND" is specified as a third parameter, then the
+## sequence of N\lay{\alpha} is shuffled
+##
+## Gives as output a list of node-layer participation
+##
+
+import sys
+import random
+
+if len(sys.argv) < 5:
+    print "Usage: %s <layers_N> <N> <M0> <A> [RND]" % sys.argv[0]
+    sys.exit(1)
+
+N = int(sys.argv[2])
+M0 = int(sys.argv[3])
+A = int(sys.argv[4])
+
+layer_degs = []
+
+
+if len(sys.argv) > 5 and sys.argv[5] == "RND":
+    randomize = True
+else:
+    randomize = False
+
+lines = open(sys.argv[1]).readlines()
+
+M = 0
+
+
+for l in lines:
+    if l[0] == "#":
+        continue
+    n = [int(x) for x in l.strip(" \n").split(" ")][0]
+    layer_degs.append(n)
+    M += 1
+
+
+if randomize:
+    random.shuffle(layer_degs)
+
+## the list node_Bi contains, at each time, the attachment
+## probabilities, i.e. it is a list which contains each node $i$
+## exactly $A + B_i$ times
+
+node_Bi = []
+
+#
+# initialize the distribution of attachment proibabilities, giving to
+# all nodes an attachment probability equal to A
+#
+
+for i in range(N):
+    for j in range(A):
+        node_Bi.append(i)
+        
+layers = []
+
+
+for i in range(M0):
+    N_alpha = layer_degs.pop()
+    node_list = []
+    num_nodes = 0
+    while num_nodes < N_alpha:
+        val = random.choice(range(N))
+        if val not in node_list:
+            node_list.append(val)
+            num_nodes += 1
+    for n in node_list:
+        node_Bi.append(n)
+    layers.append(node_list)
+    i += 1
+
+
+#sys.exit(1)
+
+while i < M:
+    N_alpha = layer_degs.pop()
+    node_list = []
+    num_nodes = 0
+    while num_nodes < N_alpha:
+        val = random.choice(node_Bi)
+        if val not in node_list:
+            node_list.append(val)
+            num_nodes += 1
+    for n in node_list:
+        node_Bi.append(n)
+    layers.append(node_list)
+    i += 1
+
+#print layers
+
+for i in range(M):
+    node_list = layers[i]
+    for n in node_list:
+        print n, i
+