Modernizing the AMPL Stack

This repository contains supporting materials for a two-part webinar series on modernizing the AMPL workflow. It is designed for academics, researchers, and students who want to transition from legacy AMPL usage to a modern, Python-based integration.

Overview

The webinar is divided into two logical phases:

1. Part One: Conceptual Evolution

   • Discussion on the evolution of the AMPL ecosystem.
   • Identifying what remains foundational and what has been modernized.
   • Located in the part-one/ directory.

2. Part Two: Hands-on Migration

   • A practical guide to migrating from standalone .mod, .dat, and .run files to a Pythonic workflow using amplpy.
   • Demonstration of data management with Pandas and visualization with Matplotlib.
   • Located in the part-two/ directory.

Repository Structure

modern-ampl/
├── part-one/
│   └── ampl_part1.pdf          # Presentation slides/document for Part 1
└── part-two/
    ├── diet.mod                # AMPL Model file (Diet Problem)
    ├── diet.dat                # Legacy AMPL Data file
    ├── diet.run                # Legacy AMPL Run file
    └── webinar_part_two.ipynb  # Main Hands-on Jupyter Notebook

Key Features & Topics Covered

1. Modern Python Integration

Move away from the rigid standalone IDE and leverage the full power of the Python ecosystem. The amplpy library allows you to:

• Load AMPL models directly into Python.
• Use Pandas DataFrames instead of .dat files for dynamic and robust data handling.
• Solve models using various solvers (HiGHS, Gurobi, CPLEX, etc.) with simple Python commands.

2. Visualization and Results Analysis

• Extract results directly into Pandas structures for post-processing.
• Create professional plots and dashboards using Matplotlib and Streamlit.
• Perform "What-If" analysis by updating parameters dynamically in Python.

3. Advanced Diagnostic Tools

• IIS (Irreducible Infeasible Subset): Learn how to diagnose infeasible models using modern solver features.
• Dual Values & Sensitivity: Access dual values for economic interpretation and sensitivity analysis.
• Snapshots: Save and restore complete AMPL sessions for warm-starts or grading purposes.
• ampls: Explore lightweight interfaces for advanced solver interactions and custom callbacks.

Getting Started

To follow the hands-on session:

1. Navigate to the part-two/ directory.
2. Open webinar_part_two.ipynb in a Jupyter environment (JupyterLab, VS Code, or Google Colab).
3. Install the necessary requirements (included in the notebook):

   pip install amplpy pandas matplotlib

4. Follow the step-by-step instructions in the notebook to migrate the "Diet Problem" to a modern stack.

Additional Resources

• AMPL Model Colaboratory: colab.ampl.com - A library of ready-to-run notebooks.
• AMPL Python API Docs: amplpy.ampl.com
• AMPL Resources: ampl.com/resources
• AMPL Streamlit Apps: amplopt.streamlit.app

---

Created for the "Moving Your Course to Modern AMPL" Webinar series.