feat: shortest_path_tools and relative demo

[shiyudawei]

Key Features

• Multi-agent System: Implements a collaborative multi-agent system based on the OxyGent framework.
• Shortest Path Calculation: Uses Google OR-Tools' min_cost_flow algorithm for efficient shortest path computation.
• Network Data Import: Supports importing network topology data from Excel/CSV files.
• Path Visualization: Utilizes Matplotlib and NetworkX to visualize network topologies and shortest paths.

Core Components

1. Shortest Path Algorithm Module (shortest_path.py):

   • Implements the shortest path using Google OR-Tools’ min_cost_flow.
   • Supports basic, inter-city, and constrained shortest path calculations.
   • Provides path visualization features.

2. Agent Configuration (shortest_path_demo.py):

   • Configures the multi-agent system.
   • Sets up LLM model connections.
   • Initiates the application.

Technology Stack

• Python: Core programming language.
• Google OR-Tools: Provides shortest path algorithm implementation.
• OxyGent: Multi-agent system framework.
• Pandas: Data processing and Excel file reading.
• Matplotlib & NetworkX: Network visualization.

Data Model

The system processes network topology data containing fields like city names, start and end nodes, distances, and optional costs.

Implementation Details

• Shortest Path Algorithm and Visualization:

  • Uses Google OR-Tools’ min_cost_flow for precise shortest path computation.
  • shortest_path.py serves as a tool file registered in OxyGent for use as an agent tool.

• Agent Configuration:

  • Demonstrates setting up agents like shortest_path_agent for path calculations, excel_agent for Excel file operations, and master_agent for coordinating other agents.

Example

Running the application with python shortest_path_demo.py uses a USnet topology, allowing users to query the shortest path between nodes and visualize the results.

Summary

OxyGent, combined with JoyCode, offers efficient development possibilities for large model applications. The shortest_path_demo provides a simple application scenario, demonstrating how large models can aid in reducing problem complexity and operational learning costs.