centralityindices 0.1.0

Python library for computing Bundle, Pivotal, and Weighted Bundle centrality indices.

CentralityIndices

CentralityIndices is a Python library for computing centrality indices for weighted directed graphs:

• Bundle Index (BI)
• Pivotal Index (PI)
• Weighted Bundle Index 1 (wBI1)
• Weighted Bundle Index 2 (wBI2)

The library accepts dense numpy.ndarray adjacency matrices and sparse SciPy matrices. For sparse graphs it can store the adjacency matrix in CSC format, which is efficient for extracting incoming edges of a target vertex.

Installation

Install the package in editable mode from the repository root:

pip install -e .

For running tests:

pip install -e ".[test]"

Quick Start

import numpy as np

from centralityindices import CentralityIndices

W = np.array(
    [
        [0, 0, 0],
        [50, 0, 0],
        [70, 40, 0],
    ],
    dtype=float,
)

centrality = CentralityIndices(W)

print(centrality.backend)
print(centrality.bundle_index(i=2, q=60, k=3))
print(centrality.pivotal_index(i=2, q=60, k=3))
print(centrality.weighted_bundle_index_1(i=2, q=60, k=3))
print(centrality.weighted_bundle_index_2(i=2, q=60, k=3))
print(centrality.all_indices(i=2, q=60, k=3))

Sparse Matrix Support

The density_threshold parameter controls backend selection. If the fraction of non-zero matrix entries is below the threshold, the graph is stored as scipy.sparse.csc_array; otherwise it is stored as a dense NumPy array.

from scipy.sparse import csc_array

W_sparse = csc_array(W)
centrality = CentralityIndices(W_sparse, density_threshold=0.1)
print(centrality.backend)

Tests

pytest tests/test_coverage.py