2017-10-15 17:36:19 +00:00
|
|
|
#!/usr/bin/env python3
|
|
|
|
|
# -*- coding: utf-8 -*-
|
|
|
|
|
"""
|
|
|
|
|
Created on Mon May 8 16:02:18 2017
|
|
|
|
|
|
|
|
|
|
@author: eddie
|
|
|
|
|
"""
|
|
|
|
|
|
|
|
|
|
from mcMatrix import markovTreeMatrix as mc
|
|
|
|
|
from totalVariationNorm import totalVariation1 as TV
|
|
|
|
|
import numpy as np
|
|
|
|
|
import matplotlib.patches as mpatches
|
|
|
|
|
import matplotlib.pyplot as plt
|
|
|
|
|
|
|
|
|
|
np.random.seed(666)
|
|
|
|
|
|
|
|
|
|
arity = 2 # arity of tree
|
|
|
|
|
depth = 3 # depth of tree
|
|
|
|
|
|
2017-11-03 20:07:40 +00:00
|
|
|
P = mc(depth, arity) # mc(depth of tree, arity of tree 2 by default)
|
2017-10-15 17:36:19 +00:00
|
|
|
|
|
|
|
|
pi = np.random.randn(len(P)) # the initial randomly generated vector, to show
|
|
|
|
|
# that in fact pi will converge to stationary.
|
|
|
|
|
|
|
|
|
|
stat = []
|
|
|
|
|
for x in range(0,len(P)): # uniform distribution of the tree
|
|
|
|
|
stat.append(1/arity)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
near = []
|
|
|
|
|
|
|
|
|
|
for x in range(100):
|
|
|
|
|
pi = P.dot(pi)
|
|
|
|
|
near.append(TV(pi,stat))
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
red_patch = mpatches.Patch(color='red', label='TV convergence to stationary dist')
|
|
|
|
|
plt.legend(handles=[red_patch])
|
|
|
|
|
plt.plot(near,'ro', ms='0.8')
|
|
|
|
|
plt.show()
|
