BMCToolkit 0.6.6

Simulate and analyze trajectories of Block Markov Chains.

This Python package provides tools to simulate and analyze trajectories of Block Markov Chains (BMCs).

Contents

A compiled C++ library is distributed, together with stubs.

Usage

Here is an example on how to use the library:

import random
import numpy as np
import BMCToolkit as BMCToolkit

if __name__ == "__main__":

    num_states = 10*30
    num_clusters = 2
    path_length = 1500

    current_state = 0
    frequency_matrix = np.zeros((num_states,num_states))
    trajectory = []
    for t in range(path_length):
        past_state = current_state
        if t % 2 == 0:
            if current_state <= num_states / 3:
                current_state = random.randint(num_states/3,num_states-1)
            else:
                current_state = random.randint(0,num_states/3-1)
        else:
            current_state = random.randint(0,num_states-1)    
        trajectory.append(current_state)
        frequency_matrix[past_state,current_state] += 1
    
    print("Testing compute_clusters_from_trajectory...")
    clustering = BMCToolkit.compute_clusters_from_trajectory( trajectory, num_states, num_clusters )
    
    print(clustering)

Related projects

The compiled C++ library utilizes the following other C++ libraries unmodified:

• the Eigen library, available at https://eigen.tuxfamily.org/;
• the Sparse Eigenvalue Computation Toolkit as a Redesigned ARPACK (SPECTRA) library, available at https://spectralib.org/.