ADEPT

ADEPT is a set of A utomatic D ifferentiation E nabled P lasma T ransport solvers.

Examples

Examples can be found in the tests folder or in the adept-notebooks repository - http://github.com/ergodicio/adept-notebooks. Example configuration files are also provided in configs/

---

Documentation

Getting Started

• Usage
  • Installation
  • Run an Example
  • Solver-Specific Guides

Solvers

• Available Solvers
  • Kinetic Solvers
  • Spectral Solvers
  • Particle Solvers
  • Envelope Solvers
  • Fluid Solvers
  • External Code Wrappers
• Vlasov 1D1V Solver
  • Things You Might Care About
  • Equations and Quantities
  • Multispecies Support
  • Solver Options
  • Boundary Conditions
  • Forcing and Drivers
  • What Gets Saved
  • Running on Multiple GPUs
  • Practical Notes
  • Configuration Reference
• Vlasov 1D2V Solver
  • Things You Might Care About
  • Equations and Quantities
  • Collisions
  • Boundary Conditions
  • Forcing and Drivers
  • What Gets Saved
  • Solver Options
  • Practical Notes
  • Configuration Reference
• Vlasov-2D Overview
  • Numerics
  • Time-step ordering (one full dt)
  • Boundary Conditions
  • Forcing and Drivers
  • What Gets Saved
  • See also
• Vlasov-Fokker-Planck 1D Solver
  • Equations and Quantities
  • Staggered Spatial Grid
  • Boundary Conditions
  • Collision Operators
  • Electric Field Solver
  • Solver Algorithm
  • Boundary Conditions
  • Forcing and Drivers
  • What Gets Saved
  • Configuration Reference
• Spectrax 1D Solver
  • Equations and Quantities
  • Solver Options
  • Entry Points
  • Boundary Conditions
  • Forcing and Drivers
  • What Gets Saved
  • Configuration Reference
• Mixed Hermite-Legendre 1D Solver
  • Equations and Quantities
  • Numerics
  • Initialization
  • Boundary Conditions
  • Forcing and Drivers
  • What Gets Saved
  • Configuration Reference
• PIC 1D Solver
  • Equations and Quantities
  • Solver Options
  • Transverse Drivers
  • Limitations
  • Boundary Conditions
  • Forcing and Drivers
  • What Gets Saved
  • Practical Notes
  • Configuration Reference
• LPSE 2D (Envelope-2D) Solver
  • Equations and Quantities
  • Boundary Conditions
  • Forcing and Drivers
  • What Gets Saved
  • Configuration Reference
• Two-Fluid 1D Solver
  • Equations and Quantities
  • Closures
  • Machine-Learned Closures
  • Things You Might Care About
  • Boundary Conditions
  • Forcing and Drivers
  • What Gets Saved
  • Practical Notes
  • Configuration Reference
• OSIRIS adept module — usage overview
  • File structure
  • End-to-end data flow
  • How to run
  • Manifest schema
  • What lands in MLflow
  • Concurrent H5 → NetCDF conversion (osiris.stream_convert)
  • Ramdisk staging (osiris.stage_root)
  • Canned plots (plots/ artifacts)
  • Programmatic use
  • Adding new metrics
  • Adding a new test problem

Configuration Reference

• Vlasov-1D Configuration Reference
  • Top-Level Structure
  • units
  • density
  • grid
  • save
  • mlflow
  • drivers
  • diagnostics
  • terms
  • Complete Examples
• Vlasov-1D2V Configuration Reference
  • Velocity-space geometry
  • grid
  • save
  • diagnostics
  • terms
• Vlasov-2D Configuration Reference
  • Top-level keys
  • units
  • grid
  • density
  • terms
  • drivers
  • save
  • mlflow
  • Example: 2D Landau damping
• VFP-1D Configuration Reference
  • Top-Level Structure
  • units
  • density
  • grid
  • save
  • mlflow
  • drivers
  • terms
  • diagnostics
  • Complete Examples
• Spectrax-1D Configuration Reference
  • Top-Level Structure
  • physics
  • grid
  • drivers
  • save
• Mixed Hermite-Legendre 1D Configuration Reference
  • Top-Level Structure
  • physics
  • grid
  • initialization
  • drivers (optional)
  • save
  • Example: two-stream instability
• PIC-1D Configuration Reference
  • Top-Level Structure
  • grid
  • terms
  • Shared Blocks
  • Complete Example
• LPSE-2D (Envelope-2D) Configuration Reference
  • Top-Level Structure
  • units
  • density
  • grid
  • save
  • mlflow
  • drivers
  • terms
  • Complete Example
  • Example Configurations
• Two-Fluid-1D Configuration Reference
  • Top-level keys
  • mlflow
  • units
  • grid
  • save
  • physics
  • drivers
  • models
• OSIRIS Configuration Reference
  • Top-Level Structure
  • mlflow
  • osiris
  • output
  • Complete example
  • Adding a new test problem

Reference

• FAQ
• API Guide
  • ergoExo
  • ADEPTModule
• Developer Guide
  • The two classes
  • Typical usage
  • The setup sequence
  • Taking gradients
  • Adding a solver
  • Libraries
• Tests
  • Running the tests
  • Test Suite
  • Continuous integration

Note

This project is under active development.

Contributing Guide

The contributing guide is in development but for now, just make an issue / pull request and we can go from there :)

Citation

If you are using this package for your research, please cite

A. Joglekar and A. Thomas, “ADEPT - automatic differentiation enabled plasma transport,” ICML - SynS & ML Workshop (https://syns-ml.github.io/2023/contributions/), 2023

References

[1] A. S. Joglekar & A. G. R. Thomas. “Unsupervised discovery of nonlinear plasma physics using differentiable kinetic simulations.” J. Plasma Phys. 88, 905880608 (2022).

[2] A. S. Joglekar and A. G. R. Thomas, “Machine learning of hidden variables in multiscale fluid simulation,” Mach. Learn.: Sci. Technol., vol. 4, no. 3, p. 035049, Sep. 2023, doi: 10.1088/2632-2153/acf81a.