from random import choice
line = list(25* 'a') + list(25 * 'b') + list(50* 'c')
x = choice(range(100))
line[x] #is the sampled value

def prior_sample(bn):
   # X1,...,XN are the variables in bn, ordered
   #   so that a node does not appear to the right of a descendent
   x = []
   for Var in [X1,..., XN]:
      x.append(random sample from P-dist(Var|parents(Var) in x))
   return x

def rejection_sample(X,e,N,bn):
   # X is the query var; e is the set of evidence variables
   # with their values; N is the number of samples
   Cts - a vector of counts for each value of X, all initially 0
   for j = 1 to N:
      y = prior_sample(bn)
      if y is consistent with e then:
         Cts[xi] += 1, where xi is the value of X in y
   return normalize(Cts)

def normalize(lst):
   s = sum(lst)
   i = 0
   while i < len(lst):
     lst[i] = lst[i] / s
     i += 1
   return lst

def likelihood_weighting(X,e,N,bn):
   # X is the query var; e is the set of evidence variables
   # with their values; N is the number of samples
   Cts - a vector of counts for each value of X, all initially 0
   for j = 1 to N:
      y,w = weighted_sample(bn,e)
      Cts[xi] += w, where xi is the value of X in y
   return normalize(Cts)   

def weighted_sample(bn,e):
   w = 1
   x = [e] #e.g., if e = {MC=T, JC=F},
           # x = [MC=T,JC=F]
   # X1,...,XN are the variables in bn, ordered
   #   so that a node does not appear to the right of a descendent
   for each variable Xi:
      if Xi is an evidence variable with value xi in e:
          w = w * P(Xi=xi|parents(Xi) in x)
      else:
          x.append(random sample from P-dist(Xi|parents(Xi)))             
   return x,w