Argonaut - Discrete Choice Optimization Software

Argonaut is an interactive software for decentralized multiagent discrete choice optimization. It supports the full design-optimize-visualize loop: users construct agents and feasible plans, run alternative optimization algorithms, and inspect how the decision space is traversed, pruned, and steered toward high quality collective solutions.

Argonaut is applicable to domains such as energy demand management, shared mobility, transport planning, logistics optimization, IoT data sharing, and drone surveillance, where many agents coordinate choices under local and global objectives.

Contents:

• Quick Start
  • 1. Open Argonaut
  • 2. Select a Dataset
  • 3. Select Agents, Plans, and Dimensions
  • 4. Configure the Algorithm
  • 5. Run
• Concepts
  • Agents
  • Plans
  • Collective Solution
  • Objectives
• Algorithms
  • Tree-Based Iterative
  • Brute Force Exhaustive Search
  • Both Tree-Based + Brute Force
• Settings
  • Dataset Source
  • Agent Selection
  • Plan Selection
  • Dimension Selection
  • Plan Editor
  • Algorithm Parameters
  • Weighted Total Cost
  • Cost Functions
• Outputs
  • Tree-Based Files
  • Tree-Based Logs
  • Brute Force Files
  • Brute Force Logs
  • Combined Mode Outputs
• Visualizer
  • Open Visualizer
  • Tree-Based Visualizer
  • Compare View
  • Brute Force Visualizer
  • Combined Visualizer

Simulator URL

The tested deployment is available at:

https://argonautsim-382548405389.us-central1.run.app/

Tested Example

The documentation examples use the following small run:

• Dataset: Gaussian [Subset]

• Agents: 3

• Plans per agent: 2

• Dimensions: 100

• Iterations: 40

• Children per node: 2

• Simulations: 1

• Global cost: VAR - Variance

• Local cost: INDEX - Plan index

• Objective weights: \(\alpha = 0\), \(\beta = 0\)

The three algorithm modes were exercised with these settings: Tree-Based Iterative, Brute Force Exhaustive search, and Both Tree-Based + Brute Force.