3️⃣ Heuristic Crossover and Mutation

The crossover operator prioritizes shared parent adjacencies, mutual best-match buddy pairs, and low-dissimilarity neighboring pieces. Swap mutation preserves permutation validity while introducing new local adjacencies.

4️⃣ Dots and Boxes Minimax Agent

The Dots and Boxes agent evaluates states from the agent perspective, searches with Minimax, supports optional alpha-beta pruning, and orders moves by immediate captures, safe moves, and risky moves.

5️⃣ Evaluation and Analysis

The notebook includes compact verification runs for both assignment parts. Full experiment grids are available through notebook flags and helper functions but are disabled by default because the complete assignment-scale runs are computationally expensive.

Repository Structure

The project is organized as follows:

AI_CA2_Genetic_Minimax/
├── data/                         # Assignment image assets
│   └── images/                   # Input JPG images for puzzle experiments
├── docs/                         # Original assignment specification
│   └── AI-F04-CA2.pdf            # Computer Assignment 2 PDF
├── notebooks/                    # Standardized Jupyter Notebook
│   └── AI_CA2_810102576.ipynb    # Main implementation and analysis notebook
├── reports/                      # Exported notebook report
│   └── AI_CA2_810102576.html     # HTML export of the executed notebook
├── src/                          # Reusable Python implementation
│   ├── __init__.py               # Package initializer
│   ├── ai_ca2.py                 # Genetic Algorithm and Minimax core logic
│   └── dots_and_boxes_gui.py     # Optional Tkinter GUI wrapper
├── tests/                        # Automated sanity tests
│   └── test_core_algorithms.py   # Core algorithm tests
├── .gitignore                    # Ignored local/generated files
├── Makefile                      # Test, notebook, export, and cleanup commands
├── README.md                     # Project documentation
└── requirements.txt              # Python dependencies

Setup & Usage

Create a virtual environment and install the required packages:

python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

Launch the standardized notebook:

jupyter notebook notebooks/AI_CA2_810102576.ipynb

Run the automated tests:

make test

Export the notebook to HTML:

make export

Run the optional Dots and Boxes GUI from a graphical Python environment:

python -m src.dots_and_boxes_gui

Results

The executed notebook verifies the main implementation with compact reproducible runs:

• The compact genetic-algorithm smoke run on starry.jpg completed successfully and returned a best fitness of approximately 0.000449895633.
• The Minimax sanity check selected the final edge of an almost-complete box and returned a positive evaluation.
• The compact Dots and Boxes benchmark showed fewer evaluated nodes with alpha-beta pruning than without pruning at the same search depth.

These results validate execution and algorithm integration, but they do not replace the full assignment-scale experiment grid.

Notes

• Full genetic-algorithm image sweeps can be enabled with RUN_FULL_GA_EXPERIMENTS = True inside the notebook.
• Full Minimax repeated benchmarks can be enabled with RUN_FULL_MINIMAX_EXPERIMENTS = True inside the notebook.
• The Tkinter GUI is isolated from the notebook so that automated notebook execution does not open a graphical window.
• No license file was present in the original submission, so no reuse license is declared in this repository.

Author

• Meraj Rastegar