Welcome to the documentation for Co-optimized Machine-Learned Manifolds (CMLM)!

CMLM includes scripts for generating and testing machine learned manifold models for LES of turbulent combustion that use neural networks. The included machine learning approaches were initially published in a Combustion and Flame article. It also includes scripts for generating physics-based manifold models with tabulated chemistry data. This package generates models in a format usable by the PeleLMeX reacting flow solver, with much of the implementation done through the PelePhysics library.

Head over to the Getting Started page to see how to install and begin using CMLM.

Source code available on GitHub. The repository and this documentation are a work in progress. Adding this documentation is the first step toward making it a usable piece of software with some semblance of organization, testing, dependency management, etc.

Warning

For now, CMLM is under active development. Changes to interfaces, file formats, outputs, behavior, etc. should all be expected and may be made without warning.

Contents:

• Getting Started
  • Dependencies & Installation
  • Using CMLM
• Library/API
  • Neural Network Models
  • Tabulated Models
    • cmlm.ctable_tools
  • General Utilities
    • cmlm.utils.input_file
    • cmlm.utils.timer
• Command Line Tools
  • ctable_tools
• Run Scripts
  • create_dummy_table_nd.py
    • Usage
    • Input File
  • create_spray_table_nd.py
    • Usage
    • Input File
• Tutorials
• Questions & Contributing
  • Discussions
  • Contributions
    • Formatting Guidelines
    • Building the Documentation
• Citation
  • Software
  • Model