This package provides Python bindings for the Murty algorithm, which is used for finding the k-best assignments in assignment problems.
pip install .pip install -e .- Python 3.8+
- NumPy
- Eigen3 (C++ library)
- CMake 3.15+
- C++14 compatible compiler
Here's how to use the murty function to get the k-best assignments.
import numpy as np
from murty import murty
# Create a cost matrix
cost_matrix = np.array([
[10, 20, 5],
[15, 5, 10],
[20, 15, 10]
], dtype=np.float64)
# Get the 3-best assignments
k = 3
assignments = murty(cost_matrix)
print(f"Top {k} assignments:")
for i, (cost, sol) in enumerate(assignments):
if i >= k:
break
print(f"Assignment {i+1}:")
print(f" Cost: {cost}")
print(f" Solution: {sol}")
# sol[i] = j means row i is assigned to column j
for row, col in enumerate(sol):
print(f" Row {row} -> Col {col} (Cost: {cost_matrix[row, col]})")This will output the top 3 assignments with their respective costs and solutions.
This package uses scikit-build-core for building. The C++ extension module _murty is built automatically when installing.
This code is an extraction of Jonatan Olofsson's implementation of Murty's algorithm from his MHT repo with pybind.
GPLv3