wirewalk 0.0.3

Python3 Implementation of the WireWalk Algorithm

wirewalk

References

Wire Before You Walk. T. Asmara, D. Bhaskar, I. Adelstein, S. Krishnaswamy, M. Perlmutter. In proceedings for Asilomar 2023.

Asilomar 2023 Proceedings

To reproduce the results, please see the instructions in the Asilomar_2023.py file.

Install

pip install wirewalk

Prerequisites

You will need

• Python3
• Networkx
• Numpy
• Gensim
• editdistance

I highly recommend installing an Anaconda environment. Future versions of WireWalk will be available on PyPI and conda.

How to use

import networkx as nx
from wirewalk.core import WireWalk, jaccard_coefficient, max_flow

# Create a graph
graph = nx.fast_gnp_random_graph(n=10, p=0.5)

# Instantiate a WireWalk object
wireWalk = WireWalk(graph, dimensions = 128, window = 10, walk_length = 80, num_walks = 10, workers = 1)

# Compute transition probabilities using jaccard coefficient transformation, generate walks, and embed nodes
model = wireWalk.fit(jaccard_coefficient)

# **MAX_FLOW and MIN_COST_MAX_FLOW ONLY WORK WITH GIVEN capacity**
# If weight exists, then  
# nx.set_edge_attributes(graph, nx.get_edge_attributes(graph, "weight"), "capacity").
# Otherwise,
nx.set_edge_attributes(graph, 1, "capacity")
model = wireWalk.fit(max_flow)