Pykov

Pykov is a tiny Python module on finite regular Markov chains.

View the Project on GitHub riccardoscalco/Pykov

Welcome to Pykov docs

Pykov is a tiny Python module on finite regular Markov chains.

You can define a Markov chain from scratch or read it from a text file according specific format. Pykov is versatile, being it able to manipulate the chain, inserting and removing nodes, and to calculate various kind of quantities, like the steady state distribution, mean first passage times, random walks, absorbing times, and so on.

Pykov is licensed under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.


Installation

Pykov can be installed/upgraded via pip

$ pip install git+git://github.com/riccardoscalco/Pykov@master #both Python2 and Python3
$ pip install --upgrade git+git://github.com/riccardoscalco/Pykov@master

Note that Pykov depends on numpy and scipy.


Getting started

Open your favourite Python shell and import pykov:

>>> import pykov
>>> pykov.__version__

Vector class

The Vector class inherits from python dict, which means it has the same behaviors and features of python dictionaries, with few exceptions. The states and the corresponding probabilities are the keys and the values of the dictionary, respectively.

Definition of a pykov.Vector():

>>> p = pykov.Vector()

You can get and set states in many ways:

>>> p['A'] = .2
>>> p
{'A': 0.2}

>>> p = pykov.Vector({'A':.3, 'B':.7})
>>> p
{'A':0.3, 'B':0.7}

>>> pykov.Vector(A=.3, B=.6, C=.1)
{'A':0.3, 'B':0.6, 'C':0.1}

States not belonging to the vector have zero probability, moreover states with zero probability are not shown:

>>> q = pykov.Vector(C=.4, B=.6)
>>> q['C']
0.4
>>> q['Z']
0.0
>>> 'Z' in q
False

>>> q['Z']=.2
>>> q
{'C': 0.4, 'B': 0.6, 'Z': 0.2}
>>> 'Z' in q
True
>>> q['Z']=0
>>> q
{'C': 0.4, 'B': 0.6}
>>> 'Z' in q
False

Vector operations

Sum

A pykov.Vector() instance can be added or substracted to another pykov.Vector() instance:

>>> p = pykov.Vector(A=.3, B=.7)
>>> q = pykov.Vector(C=.5, B=.5)
>>> p + q
{'A': 0.3, 'C': 0.5, 'B': 1.2}
>>> p - q
{'A': 0.3, 'C': -0.5, 'B': 0.2}
>>> q - p
{'A': -0.3, 'C': 0.5, 'B': -0.2}
Product

A pykov.Vector() instance can be multiplied by a scalar. The dot product with another pykov.Vector() or pykov.Matrix() instance is also supported.

>>> p = pykov.Vector(A=.3, B=.7)
>>> p * 3
{'A': 0.9, 'B': 2.1}
>>> 3 * p
{'A': 0.9, 'B': 2.1}
>>> q = pykov.Vector(C=.5, B=.5)
>>> p * q
0.35
>>> T = pykov.Matrix({('A','B'): .3, ('A','A'): .7, ('B','A'): 1.})
>>> p * T
{'A': 0.91, 'B': 0.09}
>>> T * p
{'A': 0.42, 'B': 0.3}

Vector methods

sum()

Sum the probability values.

>>> p = pykov.Vector(A=.3, B=.7)
>>> p.sum()
1.0
sort(reverse=False)

Return a list of tuples (state, probability) sorted according the probability values.

>>> p = pykov.Vector({'A':.3, 'B':.1, 'C':.6})
>>> p.sort()
[('B', 0.1), ('A', 0.3), ('C', 0.6)]
>>> p.sort(reverse=True)
[('C', 0.6), ('A', 0.3), ('B', 0.1)]
choose()

Choose a state at random, according to its probability.

>>> p = pykov.Vector(A=.3, B=.7)
>>> p.choose()
'B'
>>> p.choose()
'B'
>>> p.choose()
'A'
normalize()

Normalize the pykov.Vector, after normalization the probabilities sum to 1.

>>> p = pykov.Vector({'A':3, 'B':1, 'C':6})
>>> p.sum()
10.0
>>> p.normalize()
>>> p
{'A': 0.3, 'C': 0.6, 'B': 0.1}
>>> p.sum()
1.0
copy()

Return a shallow copy.

>>> p = pykov.Vector(A=.3, B=.7)
>>> q = p.copy()
>>> p['C'] = 1.
>>> q
{'A': 0.3, 'B': 0.7}
entropy()

Return the Shannon entropy, defined as $H(p) = \sum_i p_i \ln p_i$.

>>> p = pykov.Vector(A=.3, B=.7)
>>> p.entropy()
0.6108643020548935

For further details, have a look at Khinchin A. I., Mathematical Foundations of Information Theory Dover, 1957.

dist(q)

Return the distance to another pykov.Vector, defined as $d(p,q) = \sum_i | p_i - q_i |$.

>>> p = pykov.Vector(A=.3, B=.7)
>>> q = pykov.Vector(C=.5, B=.5)
>>> p.dist(q)
1.0
relative_entropy(q)

Return the Kullback-Leibler distance, defined as $d(p,q) = \sum_i p_i \ln (p_i/q_i)$.

>>> p = pykov.Vector(A=.3, B=.7)
>>> q = pykov.Vector(A=.4, B=.6)
>>> p.relative_entropy(q) #d(p,q)
0.02160085414354654
>>> q.relative_entropy(p) #d(q,p)
0.022582421084357485

Note that the Kullback-Leibler distance is not symmetric.


Matrix class

The pykov.Matrix() class inherits from python dictionaries. Dict keys are tuple of states, dict values are the matrix entries. Indexes do not need to be int, they can be string, as the states of a pykov.Vector().

Definition of pykov.Matrix():

>>> T = pykov.Matrix()

You can get and set items in many ways:

>>> T = pykov.Matrix()
>>> T[('A','B')] = .3
>>> T
{('A', 'B'): 0.3}
>>> T['A','A'] = .7
>>> T
{('A', 'B'): 0.3, ('A', 'A'): 0.7}

>>> T = pykov.Matrix({('A','B'): .3, ('A','A'): .7, ('B','A'): 1.})
>>> T[('A','B')]
0.3
>>> T['A','B']
0.3

Items not belonging to the matrix have value equal to zero, moreover items with value equal to zero are not shown:

>>> T = pykov.Matrix({('A','B'): .3, ('A','A'): .7, ('B','A'): 1.})
>>> T['B','B']
0.0

>>> T = pykov.Matrix({('A','B'): .3, ('A','A'): .7, ('B','A'): 1.})
>>> T
{('A', 'B'): 0.3, ('A', 'A'): 0.7, ('B', 'A'): 1.0}
>>> T['A','A'] = 0
>>> T
{('A', 'B'): 0.3, ('B', 'A'): 1.0}

Matrix operations

Sum

A pykov.Matrix() instance can be added or substracted to another pykov.Matrix() instance.

>>> T = pykov.Matrix({('A','B'): .3, ('A','A'): .7, ('B','A'): 1.})
>>> I = pykov.Matrix({('A','A'):1, ('B','B'):1})
>>> T + I
{('B', 'A'): 1.0, ('A', 'B'): 0.3, ('A', 'A'): 1.7, ('B', 'B'): 1.0}
>>> T - I
{('B', 'A'): 1.0, ('A', 'B'): 0.3, ('A', 'A'): -0.3, ('B', 'B'): -1}
Product

A pykov.Matrix() instance can be multiplied by a scalar, the dot product with a pykov.Vector() or another pykov.Matrix() instance is also supported.

>>> T = pykov.Matrix({('A','B'): .3, ('A','A'): .7, ('B','A'): 1.})
>>> T * 3
{('B', 'A'): 3.0, ('A', 'B'): 0.9, ('A', 'A'): 2.1}

>>> p = pykov.Vector(A=.3, B=.7)
>>> T * p
{'A': 0.42, 'B': 0.3}

>>> W = pykov.Matrix({('N', 'M'): 0.5, ('M', 'N'): 0.7,
                      ('M', 'M'): 0.3, ('O', 'N'): 0.5,
                      ('O', 'O'): 0.5, ('N', 'O'): 0.5})
>>> W * W
{('N', 'M'): 0.15, ('M', 'N'): 0.21, ('M', 'O'): 0.35,
 ('M', 'M'): 0.44, ('O', 'M'): 0.25, ('O', 'N'): 0.25,
 ('O', 'O'): 0.5, ('N', 'O'): 0.25, ('N', 'N'): 0.6}

Matrix methods

states()

Return the set of states.

>>> T = pykov.Matrix({('A','B'): .3, ('A','A'): .7, ('B','A'): 1.})
>>> T.states()
{'B', 'A'}

States without ingoing or outgoing transitions are removed from the set of states.

>>> T['A','C']=1
>>> T.states()
{'A', 'C', 'B'}
>>> T['A','C']=0
>>> T.states()
{'A', 'B'}
pred(key=None)

Return the precedessors of a state (if not indicated, of all states). In matrix notation, return the column of the indicated state.

>>> T = pykov.Matrix({('A','B'): .3, ('A','A'): .7, ('B','A'): 1.})
>>> T.pred()
{'A': {'A': 0.7, 'B': 1.0}, 'B': {'A': 0.3}}
>>> T.pred('A')
{'A': 0.7, 'B': 1.0}
succ(key=None)

Return the successors of a state (if not indicated, of all states). In matrix notation, return the row of the indicated state.

>>> T = pykov.Matrix({('A','B'): .3, ('A','A'): .7, ('B','A'): 1.})
>>> T.succ()
{'A': {'A': 0.7, 'B': 0.3}, 'B': {'A': 1.0}}
>>> T.succ('A')
{'A': 0.7, 'B': 0.3}
copy()

Return a shallow copy.

>>> T = pykov.Matrix({('A','B'): .3, ('A','A'): .7, ('B','A'): 1.})
>>> W = T.copy()
>>> T[('B','B')] = 1.
>>> W
{('B', 'A'): 1.0, ('A', 'B'): 0.3, ('A', 'A'): 0.7}
remove(states)

Return a shallow copy of the matrix without the indicated states. All the links where the states appear are deleted, so that the result will not be in general a stochastic matrix.

>>> T = pykov.Matrix({('A','B'): .3, ('A','A'): .7, ('B','A'): 1.})
>>> T.remove(['B'])
{('A', 'A'): 0.7}
>>> T = pykov.Matrix({('A','B'): .3, ('A','A'): .7, ('B','A'): 1.,
                     ('C','D'): .5, ('D','C'): 1., ('C','B'): .5})
>>> T.remove(['A','B'])
{('C', 'D'): 0.5, ('D', 'C'): 1.0}
stochastic()

Change the pykov.Matrix() instance in a right stochastic matrix. Set the sum of every row equal to one, raise PykovError if not possible.

>>> T = pykov.Matrix({('A','B'): 3, ('A','A'): 7, ('B','A'): .2})
>>> T.stochastic()
>>> T
{('B', 'A'): 1.0, ('A', 'B'): 0.3, ('A', 'A'): 0.7}

>>> T[('A','C')]=1
>>> T.stochastic()
pykov.PykovError: 'Zero links from node C'
transpose()

Return the transpose of the pykov.Matrix() instance.

>>> T = pykov.Matrix({('A','B'): .3, ('A','A'): .7, ('B','A'): 1.})
>>> T.transpose()
{('B', 'A'): 0.3, ('A', 'B'): 1.0, ('A', 'A'): 0.7}
>>> T
{('A', 'B'): 0.3, ('A', 'A'): 0.7, ('B', 'A'): 1.0}
eye()

Return the Identity matrix.

>>> T = pykov.Matrix({('A','B'): .3, ('A','A'): .7, ('B','A'): 1.})
>>> T.eye()
{('A', 'A'): 1., ('B', 'B'): 1.}
>>> type(T.eye())
<class 'pykov.Matrix'>
ones()

Return a pykov.Vector() instance with entries equal to 1.

>>> T = pykov.Matrix({('A','B'): .3, ('A','A'): .7, ('B','A'): 1.})
>>> T.ones()
{'A': 1.0, 'B': 1.0}
>>> type(T.ones())
<class 'pykov.Vector'>
trace()

Return the matrix trace.

>>> T = pykov.Matrix({('A','B'): .3, ('A','A'): .7, ('B','A'): 1.})
>>> T.trace()
0.7

Chain class

The pykov.Chain class inherits from pykov.Matrix class. The dict key is a tuple of states, the dict value is the transition probability to go from the first state to the second state, in other words pykov describes the transitions of a Markov chain with a right stochastic matrix.

Chain methods

adjacency()

Return the adjacency matrix.

>>> T = pykov.Chain({('A','B'): .3, ('A','A'): .7, ('B','A'): 1.})
>>> T.adjacency()
{('B', 'A'): 1, ('A', 'B'): 1, ('A', 'A'): 1}
>>> type(T.adjacency())
<class 'pykov.Matrix'>
pow(p, n)

Find the probability distribution after n steps, starting from an initial pykov.Vector() p.

>>> T = pykov.Chain({('A','B'): .3, ('A','A'): .7, ('B','A'): 1.})
>>> p = pykov.Vector(A=1)
>>> T.pow(p,3)
{'A': 0.7629999999999999, 'B': 0.23699999999999996}
>>> p * T * T * T #not efficient
{'A': 0.7629999999999999, 'B': 0.23699999999999996}
move(state)

Do one step from the indicated state to one of its successors, chosen at random according to the transition probability. Return the new state.

>>> T = pykov.Chain({('A','B'): .3, ('A','A'): .7, ('B','A'): 1.})
>>> T.move('A')
'B'
walk(steps, start=None, stop=None)

Return a random walk of n steps, starting and stopping at the indicated states. If not indicated, then the starting state is chosen according to the steady state probability. If the stopping state is reached before to do n steps, then the walker stops.

>>> T = pykov.Chain({('A','B'): .3, ('A','A'): .7, ('B','A'): 1.})
>>> T.walk(10)
['B', 'A', 'B', 'A', 'A', 'B', 'A', 'A', 'A', 'B', 'A']
>>> T.walk(10,'B','B')
['B', 'A', 'A', 'A', 'A', 'A', 'B']
walk_probability(walk)

Return the logarithm of the walk probability (see walk() method). Impossible walks have probability equal to zero.

>>> T = pykov.Chain({('A','B'): .3, ('A','A'): .7, ('B','A'): 1.})
>>> T.walk_probability(['A','A','B','A','A'])
-1.917322692203401
>>> probability = math.exp(-1.917322692203401)
>>> probability
0.147

>>> p = T.walk_probability(['A','B','B','B','A'])
>>> math.exp(p)
0.0
steady()

Return the steady state, i.e. the equilibrium distribution of the chain.

>>> T = pykov.Chain({('A','B'): .3, ('A','A'): .7, ('B','A'): 1.})
>>> T.steady()
{'A': 0.7692307692307676, 'B': 0.23076923076923028}

Since Pykov describes the chain with a right stochatic matrix, the steady state $x$ satisfies at the condition $p=pT$ and it is calculated with the inverse iteration method $Q^t x = e$, where $Q = I - T$ and $e = (0,0,...,1)$. Moreover, the Markov chain is assumed to be ergodic, i.e. the transition matrix must be irreducible and acyclic. You can easily test such properties by means of NetworkX, let's see how:

>>> import networkx as nx

>>> T = pykov.Chain({('A','B'): .3, ('A','A'): .7, ('B','A'): 1.})
>>> G = nx.DiGraph(list(T.keys()))
>>> nx.is_strongly_connected(G) # is irreducible
True
>>> nx.is_aperiodic(G)
True

>>> T = pykov.Chain({('A','B'): .3, ('A','A'): .7, ('B','B'): 1.})
>>> G = nx.DiGraph(list(T.keys()))
>>> nx.is_strongly_connected(G) # is irreducible
False
>>> nx.is_aperiodic(G)
True

>>> T = pykov.Chain({('A','B'): 1, ('B','C'): 1., ('C','A'): 1.})
>>> G = nx.DiGraph(list(T.keys()))
>>> nx.is_strongly_connected(G)
True
>>> nx.is_aperiodic(G)
False

Often, Markov chains created from raw data are not irreducibles. In such cases, the Markov chain may be defined by means of the largest strongly connected component of the associated graph. Strongly connected components can be found with NetworkX:

>>> nx.strongly_connected_components(G)

For further details on the inverse iteration method, have a look at W. Stewart, Introduction to the Numerical Solution of Markov Chains, Princeton University Press, Chichester, West Sussex, 1994.

mixing_time(cutoff=0.25, jump=1, p=None)

Return the mixing time, defined as the number of steps needed to have $|pT^n - \pi|<0.25$, where $\pi$ is the steady state $\pi = \pi T$.

If the initial distribution p is not indicated, then the iteration starts from the less probable state of the steady distribution. The parameter jump controls the iteration step, for example with jump=2 n has values 2,4,6,8,..

>>> d = {('R','R'):1./2, ('R','N'):1./4, ('R','S'):1./4,
         ('N','R'):1./2, ('N','N'):0., ('N','S'):1./2,
         ('S','R'):1./4, ('S','N'):1./4, ('S','S'):1./2}
>>> T = pykov.Chain(d)
>>> T.mixing_time()
2
entropy(p=None, norm=False)

Return the Chain entropy, defined as $H = \sum_i \pi_i H_i$, where $H_i=\sum_j T_{ij}\ln T_{ij}$. If p is not None, then the entropy is calculated with the indicated probability pykov.Vector().

>>> T = pykov.Chain({('A','B'): .3, ('A','A'): .7, ('B','A'): 1.})
>>> T.entropy()
0.46989561696530169

With norm=True entropy belongs to [0,1].

>>> T.entropy(norm=True)
0.33895603665233132

For further details, have a look at Khinchin A. I., Mathematical Foundations of Information Theory, Dover, 1957.

mfpt_to(state)

Return the Mean First Passage Times of every state to the indicated state.

>>> d = {('R', 'N'): 0.25, ('R', 'S'): 0.25, ('S', 'R'): 0.25,
         ('R', 'R'): 0.5, ('N', 'S'): 0.5, ('S', 'S'): 0.5,
         ('S', 'N'): 0.25, ('N', 'R'): 0.5, ('N', 'N'): 0.0}
>>> T = pykov.Chain(d)
>>> T.mfpt_to('R')
{'S': 3.333333333333333, 'N': 2.666666666666667} #mfpt from 'S' to 'R' is 3.33

See also Kemeny J. G. and Snell, J. L., Finite Markov Chains. Springer-Verlag: New York, 1976.

absorbing_time(transient_set)

Mean number of steps needed to leave the transient_set, return the pykov.Vector() tau where tau[i] is the mean number of steps needed to leave the transient set starting from state i. The parameter transient_set is a subset of states (iterable).

>>> d = {('R','R'):1./2, ('R','N'):1./4, ('R','S'):1./4,
         ('N','R'):1./2, ('N','N'):0., ('N','S'):1./2,
         ('S','R'):1./4, ('S','N'):1./4, ('S','S'):1./2}
>>> T = pykov.Chain(d)
>>> p = pykov.Vector({'N':.3, 'S':.7})
>>> tau = T.absorbing_time(p.keys())
>>> tau
{'S': 3.333333333333333, 'N': 2.6666666666666665}

In other words, the mean number of steps in order to leave states 'S' and 'N' starting from 'S' is 3.33. It is sufficient to calculate p * tau in order to weight the mean times according an initial distribution p.

>>> p * tau
3.1333333333333329

In order to better understand the meaning of the method, the calculation of the previous example can be approximated by means of many random walkers:

>>> numpy.mean([len(T.walk(10000000,"S","R"))-1 for i in range(1000000)])
3.3326020000000001
>>> numpy.mean([len(T.walk(10000000,"N","R"))-1 for i in range(1000000)])
2.6665549999999998
absorbing_tour(p, transient_set=None)

Return a pykov.Vector() v, where v[i] is the mean time the process is in the transient state i before leaving the transient set.

Note that v.sum() is equal to p * tau (see absorbing_time() method). If not specified, the transient set is defined as the set of states in vector p.

>>> d = {('R','R'):1./2, ('R','N'):1./4, ('R','S'):1./4,
         ('N','R'):1./2, ('N','N'):0., ('N','S'):1./2,
         ('S','R'):1./4, ('S','N'):1./4, ('S','S'):1./2}
>>> T = pykov.Chain(d)
>>> p = pykov.Vector({'N':.3, 'S':.7})
>>> T.absorbing_tour(p)
{'S': 2.2666666666666666, 'N': 0.8666666666666669}
fundamental_matrix()

Return the fundamental matrix, have a look at Kemeny J. G. and Snell J. L., Finite Markov Chains. Springer-Verlag: New York, 1976 for further details.

>>> T = pykov.Chain({('A','B'): .3, ('A','A'): .7, ('B','A'): 1.})
>>> T.fundamental_matrix()
{('B', 'A'): 0.17751479289940991, ('A', 'B'): 0.053254437869822958,
('A', 'A'): 0.94674556213017902, ('B', 'B'): 0.82248520710059214}

Note that the fundamental matrix is not sparse.

kemeny_constant()

Return the Kemeny constant of the transition matrix.

>>> T = pykov.Chain({('A','B'): .3, ('A','A'): .7, ('B','A'): 1.})
>>> T.kemeny_constant()
1.7692307692307712

Utilities

Pykov comes with an utility useful to create a pykov.Chain() from a text file, let say file /mypath/mat, which contains the transition matrix defined with the following format:

A A .7
A B .3
B A 1

The pykov.Chain() instance is created with the command:

>>> P = pykov.readmat('/mypath/mat')
>>> P
{('B', 'A'): 1.0, ('A', 'B'): 0.3, ('A', 'A'): 0.7}

Docs written with StackEdit.