MITIM: a toolbox for modeling tasks in plasma physics and fusion energy

The MITIM (MIT Integrated Modeling) is a versatile and user-friendly Python library designed for plasma physics and fusion energy researchers, distributed as the MITIM-fusion GitHub repository. Developed in 2018 by Pablo Rodriguez-Fernandez at the MIT Plasma Science and Fusion Center, this light-weight, command-line, object-oriented toolbox streamlines the execution and interpretation of physics models and simplifies complex optimization tasks.

MITIM stands out for its modular nature, making it particularly useful for integrating models with optimization workflows. The toolbox has been instrumental in various high-impact research projects, such as VITALS and PORTALS, as well as optimization tasks during the design of the SPARC tokamak. MITIM’s ongoing development and maintenance by the MFE-IM group at MIT ensure its relevance and utility in cutting-edge research, with a focus on core transport, simulation and optimization.

Researchers and developers interested in contributing can find the project on GitHub at MITIM-fusion. The repository welcomes contributions and provides guidelines for those looking to enhance MITIM’s capabilities.

Users Agreement: License and contributions

Warning

The authors are not responsible for any errors or omissions, or for the results obtained from the use of this repository. All scripts and coding examples in this repository are provided “as is”, with no guarantee of completeness, accuracy or of the results obtained.

The intended use of this repository and the capabilities it provides is to accelerate the learning curve of main transport codes, specially for students and young researchers. For publication-quality results, the user is advised to understand every step behind the wheels of MITIM, and to write custom workflows and routines to test and verify results.

The users are strongly encouraged to contribute to the code by submitting issues, requesting features or finding bugs. The Users Agreement applies to any forked version of the repository.

Overview

Developed at the MIT Plasma Science and Fusion Center, MITIM emerged in 2023 as a progression from the PORTALS project (Performance Optimization of Reactors via Training of Active Learning Surrogates). This evolution marks a significant enhancement in our approach to transport and optimization in plasma physics research.

MITIM’s core functionality revolves around the standalone execution of codes and the nuanced interpretation of results through object-oriented Python scripts. This enables researchers to seamlessly integrate these scripts into custom surrogate-based optimization frameworks, significantly boosting the efficiency and effectiveness of their studies (see Standalone Capabilities and Optimization Capabilities for more details).

Contents

• Installation
  • Instructions
  • User configuration
  • License and contributions
    • License
• Standalone Capabilities
  • TGLF
    • Plot TGLF results
    • Run TGLF from input.gacode
    • Run TGLF from TRANSP results file
    • Run TGLF from input.tglf file
    • Run 1D scans of TGLF input parameter
    • TGLF aliases
  • TGYRO
    • Run TGYRO from an input.gacode file
    • Read TGYRO results that were run externally
    • TGYRO aliases
  • TRANSP
    • Run TRANSP
    • Read results from external TRANSP run
    • TRANSP aliases
  • Miscellaneous
    • Interpret Tokamak equilibrium
    • Interpret input.gacode files
• Optimization Capabilities
  • Fusion applications
    • VITALS
      • 1. Preparation of TGLF class
      • 2. VITALS Run
      • 3. VITALS Interpretation
    • PORTALS
      • Run a standard profile prediction with PORTALS-TGLF
      • Run a standard profile prediction with PORTALS-CGYRO
    • FREEGS-U
  • Optimize a custom function
    • Understanding the MITIM namelist
    • Understanding the MITIM outputs
• Frequently Asked Questions
  • Issues during MITIM installation
  • Issues during MITIM tests
  • Issues during PORTALS simulations

Note

Language enhancements and code refinements provided by the assistance of OpenAI’s ChatGPT-4.