metaforge 1.0.3

MetaForge: A modular optimization framework for solving Job Shop Scheduling problems using metaheuristics and learning-based approaches.

🔧 MetaForge

MetaForge is a modular Python toolkit for solving Job Shop Scheduling Problems (JSSP) using classic metaheuristics and modern reinforcement learning methods.

🚀 From Tabu Search and Genetic Algorithms to Deep Q-Networks (DQN) and Neuroevolution — MetaForge brings together the best of optimization and AI in one clean, extensible framework.

---

🎯 Key Features

• ✅ Solve classic benchmark problems (OR-Library, JSON)
• 🧠 Built-in solvers:
  • Tabu Search (TS)
  • Simulated Annealing (SA)
  • Genetic Algorithm (GA)
  • Ant Colony Optimization (ACO)
  • Q-Learning
  • DQN (with and without replay buffer)
  • Neuroevolution
• 📊 Beautiful convergence and Gantt chart visualizations
• 🤖 Reinforcement Learning support out-of-the-box
• 🧪 Designed for research, education, and real-world production scheduling

---

📦 Installation

From PyPI:

pip install metaforge

From GitHub (latest):

pip install git+https://github.com/Mageed-Ghaleb/MetaForge.git

---

📁 Quick Start

from metaforge.problems.benchmark_loader import load_job_shop_instance
from metaforge.metaforge_runner import run_solver

# Load a benchmark from URL
url = "https://raw.githubusercontent.com/Mageed-Ghaleb/MetaForge/main/data/benchmarks/ft06.txt"
problem = load_job_shop_instance(url)

# Run a solver
result = run_solver("ts", problem)

# View makespan
print("Best Makespan:", result["makespan"])

---

📊 Visualizations

from metaforge.utils.visualization import plot_gantt_chart
schedule = result["schedules"][-1]
plot_gantt_chart(schedule, num_machines=problem.num_machines, num_jobs=len(problem.jobs))

---

📓 Notebooks

Name	Description	Launch
MetaForge_Quick_Start.ipynb	Light demo: install, run, visualize
MetaForge_Complete_Testing.ipynb	Full testing suite across all solvers

---

📚 Documentation

• 📖 Usage Guide
• 🧠 Solvers Overview
• 🧠 Solvers In Details
• 📂 Benchmark Format

---

🧠 Why MetaForge?

Most libraries focus on one type of solver. MetaForge unifies traditional algorithms and deep reinforcement learning into one clean package. Whether you’re teaching, publishing, or scheduling in a factory — MetaForge is your launchpad. 🚀

---

🔧 Benchmarks Supported

• FT06, FT10, FT20 (OR-Library)
• LA01–LA05
• JSON format coming soon

---

📈 Contributing

We're just getting started! Feel free to:

• Suggest solvers or enhancements
• Fork and extend
• Submit PRs — code, docs, notebooks, anything

---

📄 License

MIT License — free for academic and commercial use.

---

👨‍💻 Author

Mageed Ghaleb
📧 [e-mail]
🔗 LinkedIn
🔗 GitHub

---

Built with ❤️ for solvers, schedules, and scientific curiosity.

---

🔎 Keywords (for discoverability)

MetaForge is designed for:

• Job Shop Scheduling Problems (JSSP)
• Metaheuristics (Tabu Search, Genetic Algorithm, ACO, SA)
• Reinforcement Learning in Scheduling (Q-Learning, DQN)
• Production Scheduling Optimization
• Flexible Flowshops & Real-world Scheduling
• Benchmark Comparisons and Solver Visualization