hasnainroopawalla / ant-colony-optimization

A Python package to find the shortest path in a graph using Ant Colony Optimization

Ant Colony Optimization

A Python package to find the shortest path in a graph using Ant Colony Optimization (ACO).

➡️ Check out my Medium article for a detailed walkthrough 🚀

The Ant colony Optimization algorithm is a probabilistic technique for solving computational problems which can be reduced to finding good paths through graphs (source).

This implementation of the ACO algorithm uses the NetworkX graph environment.

🏁 Getting Started

To install the package directly from PyPi:

$ pip install aco_routing

🎈 Usage

Check out: example.py

Import all the dependencies:

from aco_routing import ACO
import networkx as nx

Create a NetworkX.Graph object with nodes and edges:

G = nx.DiGraph()

G.add_edge("A", "B", cost=2)
G.add_edge("B", "C", cost=2)
G.add_edge("A", "H", cost=2)
G.add_edge("H", "G", cost=2)
G.add_edge("C", "F", cost=1)
G.add_edge("F", "G", cost=1)
G.add_edge("G", "F", cost=1)
G.add_edge("F", "C", cost=1)
G.add_edge("C", "D", cost=10)
G.add_edge("E", "D", cost=2)
G.add_edge("G", "E", cost=2)

Use ACO to find the shortest path and cost between the source and destination:

aco = ACO(G, ant_max_steps=100, num_iterations=100, ant_random_spawn=True)

aco_path, aco_cost = aco.find_shortest_path(
    source="A",
    destination="D",
    num_ants=100,
)

Output:

ACO path: A -> H -> G -> E -> D
ACO path cost: 8.0

📦 Contents

Ant

aco_routing.Ant

• An Ant that traverses the graph.

ACO

aco_routing.ACO

• The traditional Ant Colony Optimization algorithm that spawns ants at various nodes in the graph and finds the shortest path between the specified source and destination (pseudo code).

Contributing

• Post any issues and suggestions on the GitHub issues page.
• To contribute, fork the project and then create a pull request back to master.

ant-colony-optimization's Issues

line 136, in get_node_edges return self.graph[id].edges KeyError: ' '

It might be a problem with my data set but I don't know why Dijkstra works fine but the ant colony it has a problem
it gives me this error above
It does not really specify anything that is wrong except the keyError : ""

Any way you can help would be appreciate it thank you for tour time

Dijkstra is missing when pip installing

In the example.py, it links to a code which can't be used. For example, the Dijkstra

Unable to find optimal solution for Undirected Graph

Thanks for solving the previous issue.

I'm trying to replicate the paper "Application of Ant Colony Algorithm in Finding
Shortest Paths in Mobile Games".

I notice that this library is specifically meant to deal with directed graphs, so perhaps I am trying to use the wrong tool.

from aco_routing import ACO
import networkx as nx
import matplotlib.pyplot as plt

G = nx.Graph()
# North of map
G.add_edge("HQ", "A", cost=1)
G.add_edge("A", "B", cost=1)
G.add_edge("B", "?North", cost=1)
G.add_edge("?North", "G", cost=1)
G.add_edge("G", "OGAS", cost=1)
# South of map
G.add_edge("HQ", "C", cost=1)
G.add_edge("C", "D", cost=1)
G.add_edge("D", "Helipad", cost=1)
G.add_edge("Helipad", "OGAS", cost=1)
G.add_edge("C", "E", cost=1)
G.add_edge("E", "?South", cost=1)
G.add_edge("?South", "Helipad", cost=1)

source = "OGAS"
destination = "HQ"

aco = ACO(G, ant_max_steps=100, num_iterations=100, ant_random_spawn=True)
# aco.evaporation_rate = 0
# aco.beta = 0

dijkstra_path = nx.dijkstra_path(G, source, destination)
dijkstra_cost = nx.path_weight(G, dijkstra_path, "cost")

aco_path, aco_cost = aco.find_shortest_path(
    source=source,
    destination=destination,
    num_ants=100,
)

print(f"ACO - path: {aco_path}, cost: {aco_cost}")
print(f"Dijkstra - path: {dijkstra_path}, cost: {dijkstra_cost}")
# nx.draw(G, with_labels=True, font_weight='bold')
# plt.show()

In this code, aco always returns the unoptimal solution. Is it due to the fact that the code is meant for directed graphs?

AntNet algorithm

Implement the AntNet algorithm: source

Use NetworkX package

Replace the custom graph implementation with a NetworkX graph: source