Cubic Regularization: Modular Python Implementation

This project implements advanced optimization algorithms for regularized logistic regression, including Gradient Descent, Conjugate Gradient, BFGS, ARC, and Cubic Newton, in a modular, production-grade Python package.

Features

• Modular, extensible codebase
• Robust error handling and logging
• Statistical analysis and plotting utilities
• Ready for research and enterprise use

Project Structure

cubic_regularization/
│
├── src/
│   ├── cubic_regularization/
│   │   ├── __init__.py
│   │   ├── data.py
│   │   ├── optimizers/
│   │   │   ├── __init__.py
│   │   │   ├── base.py
│   │   │   ├── gradient_descent.py
│   │   │   ├── conjugate_gradient.py
│   │   │   ├── bfgs.py
│   │   │   ├── arc.py
│   │   │   └── cubic_newton.py
│   │   ├── experiment.py
│   │   ├── utils.py
│   │   └── logging_config.py
│   └── main.py
│
├── tests/
│   ├── __init__.py
│   ├── test_data.py
│   ├── test_optimizers.py
│   └── test_experiment.py
│
├── requirements.txt
├── .gitignore
├── README.md
└── setup.py

Setup

1. Create and activate a virtual environment:

   python -m venv venv
   source venv/bin/activate  # On Windows: venv\Scripts\activate

2. Install dependencies:

   pip install -r requirements.txt

Usage

• Run experiments:

  python -m src.main

• Add your data files (e.g., a9a, a9a.t) to the project root.

Testing

• Run tests with pytest or unittest:

  pytest tests/

Running from the Project Root

If you want to run the experiment from the project root (not inside src/), set the PYTHONPATH so Python can find the package:

On Windows PowerShell:

$env:PYTHONPATH = ".\src"; python -m src.main

On Linux/macOS:

PYTHONPATH=./src python -m src.main

License

MIT