Data

Tools

Indexes

Applications

Experiments

Awesome

An open API service indexing awesome lists of open source software.

awesome-ml-in-plasma-physics

⚛️ ML in fusion industry/science 🍩⚡
https://github.com/kharitonov-ivan/awesome-ml-in-plasma-physics

Last synced: about 20 hours ago
JSON representation

  • Research Papers

    • DOI - Abstract Equilibrium reconstruction is crucial in nuclear fusion
    • DOI - Abstract We present a semi-automated algorithm for designing
    • DOI - 4326/ac121b) <!-- imported-from-bib -->
    • DOI - Abstract We propose a novel flux-surface parameterization
    • DOI - Abstract A neural network, BES-ELMnet, predicting a
    • DOI - Abstract This article reviews applications of Bayesian inference
    • DOI - Abstract A deep learning-based disruption prediction algorithm
    • DOI - AbstractFluid-based scrape-off layer transport codes, such as
    • DOI - Abstract We explore the possibility of fully replacing a plasma
    • DOI - Abstract Artificial neural network models have been developed to
    • DOI - Disruptions are sudden and unavoidable losses of confinement that
    • DOI - Stable, high-performance operation of a tokamak requires several
    • arXiv - The drive to control tokamaks, a prominent technology in
    • arXiv - Generative deep learning techniques are employed in a novel
    • DOI - The tokamak offers a promising path to fusion energy, but
    • arXiv - Understanding and accounting for uncertainty helps to ensure
    • arXiv - The path of tokamak fusion and ITER is maintaining
    • OpenReview - Reinforcement learning (RL) is capable of training
    • DOI - Abstract Equilibrium reconstruction is crucial in nuclear fusion
    • arXiv - This study investigates the feasibility of reconstructing the
    • arXiv - Precise control of plasma shape and position is essential for
    • arXiv - Generative artificial intelligence methods are employed for
    • DOI - 024-07313-3) - The tokamak approach, utilizing a toroidal magnetic field
    • DOI - Abstract We present a semi-automated algorithm for designing
    • arXiv - In the quest for controlled thermonuclear fusion, tokamaks
    • arXiv - Machine learning models are exceptionally effective in
    • Thesis - Fusion in a magnetically confined plasma is still in the realm of
    • DOI - For stable and efficient fusion energy production using a tokamak
    • arXiv - This research explores the application of Deep Reinforcement
    • Link - In this work, we extend the database of 50,000 2D UEDGE
    • arXiv - Although tokamaks are one of the most promising devices for
    • DOI - Abstract A deep learning-based disruption prediction algorithm
    • DOI - Abstract We explore the possibility of fully replacing a plasma
    • DOI - For decades, plasma transport simulations in tokamaks have used
    • arXiv - Machine learning (ML) provides a broad spectrum of tools and
    • DOI - - imported-from-bib -->
    • arXiv - Rapid reconstruction of 2D plasma profiles from line-integral
    • DOI - Abstract Artificial neural network models have been developed to
    • arXiv - Bayesian optimisation (BO) is a powerful framework for global
    • DOI - In many practical applications of reinforcement learning (RL), it
    • arXiv - This paper presents a sparsified Fourier neural operator for
    • arXiv - Machine Learning guided data augmentation may support the
    • DOI - Abstract A neural network, BES-ELMnet, predicting a
    • arXiv - The fusion research facility ITER is currently being
    • arXiv - We present TORAX, a new, open-source, differentiable tokamak
    • arXiv - Fusion power production in tokamaks uses discharge configurations
    • arXiv - Generative deep learning techniques are employed in a novel
    • DOI - The tokamak offers a promising path to fusion energy, but
    • arXiv - Understanding and accounting for uncertainty helps to ensure
    • OpenReview - Reinforcement learning (RL) is capable of training
    • DOI - Abstract Equilibrium reconstruction is crucial in nuclear fusion
    • arXiv - This study investigates the feasibility of reconstructing the
    • arXiv - Precise control of plasma shape and position is essential for
    • arXiv - Generative artificial intelligence methods are employed for
    • DOI - The force-balanced state of magnetically confined plasmas heated
    • arXiv - A large number of magnetohydrodynamic (MHD) equilibrium
    • arXiv - In this paper, we present a new static and time-dependent
    • Link - A large set of 2D UEDGE simulations with currents and cross-field
    • arXiv - We present a fast and accurate data-driven surrogate model
    • DOI - Generating energy from nuclear fusion in a tokamak may highly
    • DOI - AbstractNuclear fusion using magnetic confinement, in
    • arXiv - Neural networks (NNs) offer a path towards synthesizing and
    • arXiv - Edge plasma turbulence is critical to the performance of
    • DOI - Disruptions are sudden and unavoidable losses of confinement that
    • arXiv - Predicting disruptions across different tokamaks is a great
    • DOI - 217/5840) - In this paper we present results from the first use of neural
    • DOI - The detachment regime has a high potential to play an important
    • arXiv - Within integrated tokamak plasma modeling, turbulent transport
    • DOI - Abstract We propose a novel flux-surface parameterization
    • arXiv - We present an ultrafast neural network model, QLKNN, which
    • arXiv - the-electron-temperature-and-identifying) - A machine learning approach has been implemented to measure
    • DOI - Nuclear fusion power delivered by magnetic-confinement tokamak
    • arXiv - Nuclear fusion is the process that powers the sun, and it is
    • arXiv - Neural networks provide powerful approaches of dealing with
    • DOI - Stable, high-performance operation of a tokamak requires several
    • DOI - 024-07313-3) - The tokamak approach, utilizing a toroidal magnetic field
    • DOI - Abstract We present a semi-automated algorithm for designing
    • arXiv - In the quest for controlled thermonuclear fusion, tokamaks
    • arXiv - Machine learning models are exceptionally effective in
    • Thesis - Fusion in a magnetically confined plasma is still in the realm of
    • DOI - For stable and efficient fusion energy production using a tokamak
    • arXiv - Rapid reconstruction of 2D plasma profiles from line-integral
    • arXiv - This research explores the application of Deep Reinforcement
    • Link - In this work, we extend the database of 50,000 2D UEDGE
    • arXiv - Although tokamaks are one of the most promising devices for
    • arXiv - The drive to control tokamaks, a prominent technology in
    • arXiv - The path of tokamak fusion and ITER is maintaining
    • DOI - Abstract A deep learning-based disruption prediction algorithm
    • DOI - Abstract We explore the possibility of fully replacing a plasma
    • DOI - For decades, plasma transport simulations in tokamaks have used
    • arXiv - Machine learning (ML) provides a broad spectrum of tools and
    • DOI - - imported-from-bib -->
    • DOI - Abstract Artificial neural network models have been developed to
    • arXiv - Bayesian optimisation (BO) is a powerful framework for global
    • DOI - In many practical applications of reinforcement learning (RL), it
    • arXiv - This paper presents a sparsified Fourier neural operator for
    • arXiv - Machine Learning guided data augmentation may support the
    • DOI - Abstract A neural network, BES-ELMnet, predicting a
    • arXiv - The fusion research facility ITER is currently being
    • arXiv - We present TORAX, a new, open-source, differentiable tokamak
    • arXiv - Fusion power production in tokamaks uses discharge configurations
    • arXiv - The physical sciences require models tailored to specific
    • DOI - A Bayesian optimization framework is used to investigate
    • arXiv - Grid decarbonization for climate change requires dispatchable
    • DOI - AbstractTokamaks are the most promising way for nuclear fusion
    • arXiv - A deep neural network is developed and trained on magnetic
    • arXiv - The physical sciences require models tailored to specific
    • DOI - A Bayesian optimization framework is used to investigate
    • arXiv - Grid decarbonization for climate change requires dispatchable
    • DOI - AbstractTokamaks are the most promising way for nuclear fusion
    • arXiv - A deep neural network is developed and trained on magnetic
    • arXiv - Predicting plasma evolution within a Tokamak is crucial to
    • arXiv - While fusion reactors known as tokamaks hold promise as a
    • DOI - Abstract This article reviews applications of Bayesian inference
    • Link - This presentation will introduce a multi-fidelity neural network
    • arXiv - Turbulence in fluids, gases, and plasmas remains an open
    • arXiv - Reinforcement learning (RL) has shown promising results for
    • DOI - AbstractFluid-based scrape-off layer transport codes, such as
    • arXiv - Edge plasma turbulence is critical to the performance of
    • arXiv - Predicting disruptions across different tokamaks is a great
    • DOI - The detachment regime has a high potential to play an important
    • arXiv - Within integrated tokamak plasma modeling, turbulent transport
    • DOI - Abstract We propose a novel flux-surface parameterization
    • arXiv - We present an ultrafast neural network model, QLKNN, which
    • arXiv - the-electron-temperature-and-identifying) - A machine learning approach has been implemented to measure
    • DOI - Nuclear fusion power delivered by magnetic-confinement tokamak
    • arXiv - Nuclear fusion is the process that powers the sun, and it is
    • arXiv - Neural networks provide powerful approaches of dealing with
    • DOI - Stable, high-performance operation of a tokamak requires several
    • DOI - Disruptions are sudden and unavoidable losses of confinement that
    • DOI - 217/5840) - In this paper we present results from the first use of neural
    • arXiv - Predicting plasma evolution within a Tokamak is crucial to
    • arXiv - While fusion reactors known as tokamaks hold promise as a
    • DOI - Abstract This article reviews applications of Bayesian inference
    • Link - This presentation will introduce a multi-fidelity neural network
    • arXiv - Turbulence in fluids, gases, and plasmas remains an open
    • arXiv - Reinforcement learning (RL) has shown promising results for
    • arXiv - A large number of magnetohydrodynamic (MHD) equilibrium
    • arXiv - In this paper, we present a new static and time-dependent
    • Link - A large set of 2D UEDGE simulations with currents and cross-field
    • arXiv - We present a fast and accurate data-driven surrogate model
    • DOI - Generating energy from nuclear fusion in a tokamak may highly
    • DOI - AbstractNuclear fusion using magnetic confinement, in
    • arXiv - Neural networks (NNs) offer a path towards synthesizing and

  • Implementation Papers

    • Paper - Neural network model for cross-device line-integral diagnostics with physics constraints
    • Paper - Neural network model for cross-device line-integral diagnostics with physics constraints
    • Paper - Neural network model for cross-device line-integral diagnostics with physics constraints

  • Tools

    • Simulation and Modeling Frameworks

      • OpenPOPCON - Open source Plasma Operating CONtour analysis tool for tokamak performance prediction and optimization
      • CFS-POPCON - Plasma Operating CONtour analysis tool for tokamak performance prediction and optimization
      • TORAX - Differentiable tokamak core transport simulator in JAX with ML-surrogate integration (QLKNN neural networks), trajectory optimization, and real-time capable compilation
      • FreeGSNKE - Free boundary equilibrium solver for tokamaks
      • OpenFUSIONToolkit - Open source fusion simulation toolkit with comprehensive plasma physics modeling capabilities
      • RAPTOR - RApid Plasma Transport simulatOR for tokamaks
      • TGLF - Trapped Gyro-Landau Fluid model for tokamak transport
      • FUSE.jl - Fusion Simulation Engine in Julia
      • MXHEquilibrium.jl - MHD equilibrium solver in Julia
      • vmecpp - C++ implementation of the VMEC stellarator equilibrium code
      • OMFIT - One Modeling Framework for Integrated Tasks with over 110 physics modules, supporting machine learning reduced models and HPC workflow automation. Used by 400+ scientists across 25 institutions
      • OMAS - Ordered Multi-dimensional Arrays for Magnetic Confinement Fusion, a standardized Python library for storing and manipulating tokamak experimental and simulation data
      • TORAX - Differentiable tokamak core transport simulator in JAX with ML-surrogate integration (QLKNN neural networks), trajectory optimization, and real-time capable compilation
      • FreeGSNKE - Free boundary equilibrium solver for tokamaks
      • TokaMaker - An open-source time-dependent Grad-Shafranov tool for the design and modeling of axisymmetric fusion devices ([paper](https://arxiv.org/abs/2311.07719))
      • RAPTOR - RApid Plasma Transport simulatOR for tokamaks
      • TGLF - Trapped Gyro-Landau Fluid model for tokamak transport
      • FUSE.jl - Fusion Simulation Engine in Julia
      • MXHEquilibrium.jl - MHD equilibrium solver in Julia
      • vmecpp - C++ implementation of the VMEC stellarator equilibrium code
      • OMFIT - One Modeling Framework for Integrated Tasks with over 110 physics modules, supporting machine learning reduced models and HPC workflow automation. Used by 400+ scientists across 25 institutions
      • OMAS - Ordered Multi-dimensional Arrays for Magnetic Confinement Fusion, a standardized Python library for storing and manipulating tokamak experimental and simulation data
      • SIMSOPT - Flexible stellarator optimization framework in Python/C++ with interfaces to VMEC and SPEC, including ML-ready optimization routines and parallelized gradient calculations
      • QuaLiKiz / QLKNN - Quasi-linear gyrokinetic transport model for tokamak plasmas with neural network surrogate (QLKNN) for 10,000x faster predictions
      • Travis Code - IPP Max Planck Institute plasma physics code
      • SIMSOPT - Flexible stellarator optimization framework in Python/C++ with interfaces to VMEC and SPEC, including ML-ready optimization routines and parallelized gradient calculations
      • QuaLiKiz / QLKNN - Quasi-linear gyrokinetic transport model for tokamak plasmas with neural network surrogate (QLKNN) for 10,000x faster predictions
      • Travis Code - IPP Max Planck Institute plasma physics code

    • Machine Learning Frameworks for Plasma Physics

      • FRNN (Fusion Recurrent Neural Network) - Deep learning package for tokamak disruption prediction using recurrent neural networks with stateful LSTM training, multi-machine capabilities, and TensorBoard integration (PPPL)
      • disruption-py - Physics-based scientific framework for disruption analysis with AI/ML applications supporting multi-tokamak analysis (C-Mod, DIII-D compatibility) (MIT PSFC)
      • FRNN (Fusion Recurrent Neural Network) - Deep learning package for tokamak disruption prediction using recurrent neural networks with stateful LSTM training, multi-machine capabilities, and TensorBoard integration (PPPL)
      • disruption-py - Physics-based scientific framework for disruption analysis with AI/ML applications supporting multi-tokamak analysis (C-Mod, DIII-D compatibility) (MIT PSFC)
      • TokSearch - Search engine for fusion experimental data ([paper](https://www.sciencedirect.com/science/article/abs/pii/S0920379618301042))
      • DisruptionBench - First standardized benchmark for tokamak disruption prediction across DIII-D, EAST, and Alcator C-Mod with ~30,000 discharges focusing on model generalizability
      • TokSearch - Search engine for fusion experimental data ([paper](https://www.sciencedirect.com/science/article/abs/pii/S0920379618301042))
      • DisruptionBench - First standardized benchmark for tokamak disruption prediction across DIII-D, EAST, and Alcator C-Mod with ~30,000 discharges focusing on model generalizability
Programming Languages
Python 13 Julia 4 Jupyter Notebook 3 HTML 2 C++ 2
Categories
Research Papers 144 Tools 36 Implementation Papers 3
Sub Categories
Simulation and Modeling Frameworks 28 Data Platforms and Search Tools 4 Machine Learning Frameworks for Plasma Physics 4
Keywords
fusion 6 plasma 4 stellarator 4 plasma-physics 4 stellarators 4 mhd-simulation 2 data-storage-infrastructure 2 imas 2 iter 2 omfit 2 storage-systems 2 nuclear-fusion 2 optimization 2 deep-learning 2 keras 2 rnn 2 tensorflow 2