Fastest possible contagion algorithm with iGraph

I think you are duplicating your work. At each time step you check whether the vertex at hand infects others or not, namely you run countSimilarNeigh only for the vertex at hand. Instead you run it for all the neighbors of the vertex. Here is what I think the following code might work well. I have also changed the logic of the code. Now it's focused on the susceptibles and iterate through them. It's faster now but one has to check the results for integrity. My change in the countSimilarNeigh might have also made it a little bit faster.

from igraph import *
from random import *
from time import *

def countSimilarNeigh(g,v):
    return float(g.vs(g.neighbors(v))['state'].count(True))/g.degree(v)

def contagion(g):
    contagious = True
    while contagious:
        for v in g.vs():
            contagious = False
            if v['contagious'] == False:
                if countSimilarNeigh(g,v.index) > 0.1:
                    v['state'] = True
                    v['contagious'] = True
                    contagious = True

def init_graph(n = 60, p = .1):
    g = Graph.Erdos_Renyi(n,p)                
    while g.is_connected == False:
        g = Graph.Erdos_Renyi(n,p)
    g.simplify(multiple=True, loops=True)
    return g


def score(g,repl = 200):
    for c in range(repl):
        cc = 0
        for i in g.vs():
            i['contagious'] = False
            i['state'] = False
            if random() < .1 and cc < 4:
                i['state'] = True
                i['contagious'] = True
                cc += 1
        contagion(g)

t0 = time()    
score(init_graph())
print time()-t0