Torrin Bechtel Amara
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Torrin Bechtel Amara has a total of 34 co-authors. Here are the top co-authors by number of shared publications:
- Joseph McClenaghan
General Atomics, PO Box 85608, San Diego, CA 92186-5608, United States of America
4 shared publications - Lang L. Lao
General Atomics 2 , P.O. Box 85608, San Diego, California 92186-5608, USA
4 shared publications - Scott E. Kruger
Tech-X Corp., Boulder, CO (United States)
4 shared publications - Cihan Akcay
General Atomics, San Diego, CA, United States of America
3 shared publications - Xuan Sun
General Atomics, San Diego, CA 92121, United States of America
3 shared publications - Anirban Samaddar
Argonne National Laboratory, Lemont, IL, United States of America
2 shared publications
Latest Publications
- Open Access
- 12 August 2026
- Published by IOP Publishing in Nuclear Fusion
- Vol. 66 (9) , 096033
Achievement of a high-density, high-confinement, and high-beta tokamak plasma regime in DIII-D, and implications for a lower-current path for ITER and FPP
Experiments on DIII-D have demonstrated a density-confinement synergy that enables sustainment of high performance in a previously unattained parameter regime of simultaneous very high energy confinement quality (H 98y2 ⩾ 1.5), very high line-average density Greenwald fraction ( ƒ Gr = πa 2 < n >/ I P ⩾ 1.4), and high toroidal beta ( β T ⩾ 3%). Tokamak operation in this regime is essential for a compact steady-state FPP, as well as for Q = 10 with 500 MW of fusion power in ITER at I P ≪ 15 MA. These experiments leveraged the knowledge that, in the high-poloidal-beta ( β P ) regime, density gradients can enhance turbulence stabilization caused by high α MHD ( α MHD ∼ d β P / d r ). This was described by theoretical predictions and gyrokinetic transport simulations (Kotschenreuther et al 2024 Nucl. Fusion 64 076033), and later confirmed by experiments on DIII-D (Ding et al 2024 Nature 629 555). To increase both β P and β T , the new experiments increased the ideal-wall stability β N -limit by using a smaller plasma-outer wall distance and higher triangularity in the plasma cross section (top/bottom average δ ∼ 0.9), enabled by the recent ‘shape & volume rise’ (SVR) modification to the DIII-D divertor. The higher triangularity also contributed to achieving higher ƒ Gr by enabling higher pedestal density. At high density, the pedestal is ballooning limited and exhibits small and frequent ELMs, while the divertor is near detachment even without any impurity seeding. Highplasma performance was attained and sustained reproducibly, with the eventual terminations brought about by an MHD mode destabilized as the current profile slowly continued to evolve. A path to stationary fully noninductive operation might include ECH injection to reduce both core impurity accumulation and the electron collisionality, thus increasing the bootstrap current. These experiments provide the first experimental demonstration of the ƒ Gr , H 98 y 2 , and β T values required simultaneously for ITER Q = 10 at I P < 10 MA, pointing to practical ways to improve the energy confinement in a fusion reactor.
- 12 August 2026
- Published by IOP Publishing in Nuclear Fusion
- Vol. 66 (9) , 096033
Open AccessAchievement of a high-density, high-confinement, and high-beta tokamak plasma regime in DIII-D, and implications for a lower-current path for ITER and FPP
Experiments on DIII-D have demonstrated a density-confinement synergy that enables sustainment of high performance in a previously unattained parameter regime of simultaneous very high energy confinement quality (H 98y2 ⩾ 1.5), very high line-average density Greenwald fraction ( ƒ Gr = πa 2 < n >/ I P ⩾ 1.4), and high toroidal beta ( β T ⩾ 3%). Tokamak operation in this regime is essential for a compact steady-state FPP, as well as for Q = 10 with 500 MW of fusion power in ITER at I P ≪ 15 MA. These experiments leveraged the knowledge that, in the high-poloidal-beta ( β P ) regime, density gradients can enhance turbulence stabilization caused by high α MHD ( α MHD ∼ d β P / d r ). This was described by theoretical predictions and gyrokinetic transport simulations (Kotschenreuther et al 2024 Nucl. Fusion 64 076033), and later confirmed by experiments on DIII-D (Ding et al 2024 Nature 629 555). To increase both β P and β T , the new experiments increased the ideal-wall stability β N -limit by using a smaller plasma-outer wall distance and higher triangularity in the plasma cross section (top/bottom average δ ∼ 0.9), enabled by the recent ‘shape & volume rise’ (SVR) modification to the DIII-D divertor. The higher triangularity also contributed to achieving higher ƒ Gr by enabling higher pedestal density. At high density, the pedestal is ballooning limited and exhibits small and frequent ELMs, while the divertor is near detachment even without any impurity seeding. Highplasma performance was attained and sustained reproducibly, with the eventual terminations brought about by an MHD mode destabilized as the current profile slowly continued to evolve. A path to stationary fully noninductive operation might include ECH injection to reduce both core impurity accumulation and the electron collisionality, thus increasing the bootstrap current. These experiments provide the first experimental demonstration of the ƒ Gr , H 98 y 2 , and β T values required simultaneously for ITER Q = 10 at I P < 10 MA, pointing to practical ways to improve the energy confinement in a fusion reactor.
- Open Access
- 10 November 2025
- Published by IOP Publishing in Nuclear Fusion
- Vol. 66 (1) , 016015
Reconstruction of current profiles from external magnetic and MSE measurements on DIII-D
This work introduces a new current model into the EFIT code for plasma current profile reconstruction. Using this model, edge-localized current density profiles are reconstructed solely from external magnetic measurements. Quantitative comparisons show that edge current profiles obtained with the new current model align closely with those from kinetic equilibrium reconstruction. The new model enables current density reconstruction with fewer diagnostics than kinetic equilibrium reconstruction (Lao et al 1990 Nucl. Fusion 30 1035–49), demonstrating that edge-localized current density information can be obtained from external magnetic measurements only, without the help of kinetic profiles and internal diagnostics, and without using the bootstrap current model. Furthermore, with the constraints of external magnetic and MSE measurements, current profiles in full radius obtained from the new model are in good agreement with the results of kinetic equilibrium reconstruction, which not only improves the efficiency of current reconstruction but also helps avoid possible errors caused by different kinds of diagnoses. Possible applications are also discussed, and a brief analysis of the evolution of the full-radius current density is performed. Noticeable decreases in both the magnitude and gradient of the edge current around the time of ELM bursts are observed. Moreover, analysis of the full radius current density profiles ( ρ N = 0 − 1.0 , ρ N denotes to square root of the normalized toroidal flux) reveals that as the edge current changes, the inner current also changes following a sinusoidal-like pattern at almost all time slices during four type-I ELMs in discharge 161493. The current evolution in the inner region ( ρ N = 0.2 − 0.6 ) is synchronized with the boundary region ( ρ N = 0.89 − 1.0 ), while the intermediate region ( ρ N = 0.6 − 0.89 ) shows opposite behavior. This sinusoidal-like variations in full-radius current density profile indicates strong core-pedestal coupling, suggesting ELMs significantly affect core plasmas, though current changes manifest differently across radial regions.
- 10 November 2025
- Published by IOP Publishing in Nuclear Fusion
- Vol. 66 (1) , 016015
Open AccessReconstruction of current profiles from external magnetic and MSE measurements on DIII-D
This work introduces a new current model into the EFIT code for plasma current profile reconstruction. Using this model, edge-localized current density profiles are reconstructed solely from external magnetic measurements. Quantitative comparisons show that edge current profiles obtained with the new current model align closely with those from kinetic equilibrium reconstruction. The new model enables current density reconstruction with fewer diagnostics than kinetic equilibrium reconstruction (Lao et al 1990 Nucl. Fusion 30 1035–49), demonstrating that edge-localized current density information can be obtained from external magnetic measurements only, without the help of kinetic profiles and internal diagnostics, and without using the bootstrap current model. Furthermore, with the constraints of external magnetic and MSE measurements, current profiles in full radius obtained from the new model are in good agreement with the results of kinetic equilibrium reconstruction, which not only improves the efficiency of current reconstruction but also helps avoid possible errors caused by different kinds of diagnoses. Possible applications are also discussed, and a brief analysis of the evolution of the full-radius current density is performed. Noticeable decreases in both the magnitude and gradient of the edge current around the time of ELM bursts are observed. Moreover, analysis of the full radius current density profiles ( ρ N = 0 − 1.0 , ρ N denotes to square root of the normalized toroidal flux) reveals that as the edge current changes, the inner current also changes following a sinusoidal-like pattern at almost all time slices during four type-I ELMs in discharge 161493. The current evolution in the inner region ( ρ N = 0.2 − 0.6 ) is synchronized with the boundary region ( ρ N = 0.89 − 1.0 ), while the intermediate region ( ρ N = 0.6 − 0.89 ) shows opposite behavior. This sinusoidal-like variations in full-radius current density profile indicates strong core-pedestal coupling, suggesting ELMs significantly affect core plasmas, though current changes manifest differently across radial regions.
- Open Access
- 1 September 2024
- Published by AIP Publishing in Physics of Plasmas
- Vol. 31 (9)
EFIT-Prime: Probabilistic and physics-constrained reduced-order neural network model for equilibrium reconstruction in DIII-D
We introduce EFIT-Prime, a novel machine learning surrogate model for EFIT (Equilibrium FIT) that integrates probabilistic and physics-informed methodologies to overcome typical limitations associated with deterministic and ad hoc neural network architectures. EFIT-Prime utilizes a neural architecture search-based deep ensemble for robust uncertainty quantification, providing scalable and efficient neural architectures that comprehensively quantify both data and model uncertainties. Physically informed by the Grad–Shafranov equation, EFIT-Prime applies a constraint on the current density Jtor and a smoothness constraint on the first derivative of the poloidal flux, ensuring physically plausible solutions. Furthermore, the spatial location of the diagnostics is explicitly incorporated in the inputs to account for their spatial correlation. Extensive evaluations demonstrate EFIT-Prime's accuracy and robustness across diverse scenarios, most notably showing good generalization on negative-triangularity discharges that were excluded from training. Timing studies indicate an ensemble inference time of 15 ms for predicting a new equilibrium, offering the possibility of plasma control in real-time, if the model is optimized for speed.
- 1 September 2024
- Published by AIP Publishing in Physics of Plasmas
- Vol. 31 (9)
Open AccessEFIT-Prime: Probabilistic and physics-constrained reduced-order neural network model for equilibrium reconstruction in DIII-D
We introduce EFIT-Prime, a novel machine learning surrogate model for EFIT (Equilibrium FIT) that integrates probabilistic and physics-informed methodologies to overcome typical limitations associated with deterministic and ad hoc neural network architectures. EFIT-Prime utilizes a neural architecture search-based deep ensemble for robust uncertainty quantification, providing scalable and efficient neural architectures that comprehensively quantify both data and model uncertainties. Physically informed by the Grad–Shafranov equation, EFIT-Prime applies a constraint on the current density Jtor and a smoothness constraint on the first derivative of the poloidal flux, ensuring physically plausible solutions. Furthermore, the spatial location of the diagnostics is explicitly incorporated in the inputs to account for their spatial correlation. Extensive evaluations demonstrate EFIT-Prime's accuracy and robustness across diverse scenarios, most notably showing good generalization on negative-triangularity discharges that were excluded from training. Timing studies indicate an ensemble inference time of 15 ms for predicting a new equilibrium, offering the possibility of plasma control in real-time, if the model is optimized for speed.
- Open Access
- 15 July 2024
- Published by IOP Publishing in Nuclear Fusion
- Vol. 64 (8) , 086065
Impact of various DIII-D diagnostics on the accuracy of neural network surrogates for kinetic EFIT reconstructions
Kinetic equilibrium reconstructions make use of profile information such as particle density and temperature measurements in addition to magnetics data to compute a self-consistent equilibrium. They are used in a multitude of physics-based modeling. This work develops a multi-layer perceptron (MLP) neural network (NN) model as a surrogate for kinetic Equilibrium Fitting (EFITs) and trains on the 2019 DIII-D discharge campaign database of kinetic equilibrium reconstructions. We investigate the impact of including various diagnostic data and machine actuator controls as input into the NN. When giving various categories of data as input into NN models that have been trained using those same categories of data, the predictions on multiple equilibrium reconstruction solutions (poloidal magnetic flux, global scalars, pressure profile, current profile) are highly accurate. When comparing different models with different diagnostics as input, the magnetics-only model outputs accurate kinetic profiles and the inclusion of additional data does not significantly impact the accuracy. When the NN is tasked with inferring only a single target such as the EFIT pressure profile or EFIT current profile, we see a large increase in the accuracy of the prediction of the kinetic profiles as more data is included. These results indicate that certain MLP NN configurations can be reasonably robust to different burning-plasma-relevant diagnostics depending on the accuracy requirements for equilibrium reconstruction tasks.
- 15 July 2024
- Published by IOP Publishing in Nuclear Fusion
- Vol. 64 (8) , 086065
Open AccessImpact of various DIII-D diagnostics on the accuracy of neural network surrogates for kinetic EFIT reconstructions
Kinetic equilibrium reconstructions make use of profile information such as particle density and temperature measurements in addition to magnetics data to compute a self-consistent equilibrium. They are used in a multitude of physics-based modeling. This work develops a multi-layer perceptron (MLP) neural network (NN) model as a surrogate for kinetic Equilibrium Fitting (EFITs) and trains on the 2019 DIII-D discharge campaign database of kinetic equilibrium reconstructions. We investigate the impact of including various diagnostic data and machine actuator controls as input into the NN. When giving various categories of data as input into NN models that have been trained using those same categories of data, the predictions on multiple equilibrium reconstruction solutions (poloidal magnetic flux, global scalars, pressure profile, current profile) are highly accurate. When comparing different models with different diagnostics as input, the magnetics-only model outputs accurate kinetic profiles and the inclusion of additional data does not significantly impact the accuracy. When the NN is tasked with inferring only a single target such as the EFIT pressure profile or EFIT current profile, we see a large increase in the accuracy of the prediction of the kinetic profiles as more data is included. These results indicate that certain MLP NN configurations can be reasonably robust to different burning-plasma-relevant diagnostics depending on the accuracy requirements for equilibrium reconstruction tasks.
- Open Access
- 1 May 2024
- Published by AIP Publishing in Physics of Plasmas
- Vol. 31 (5)
Thinking Bayesian for plasma physicists
Bayesian statistics offers a powerful technique for plasma physicists to infer knowledge from the heterogeneous data types encountered. To explain this power, a simple example, Gaussian Process Regression, and the application of Bayesian statistics to inverse problems are explained. The likelihood is the key distribution because it contains the data model, or theoretic predictions, of the desired quantities. By using prior knowledge, the distribution of the inferred quantities of interest based on the data given can be inferred. Because it is a distribution of inferred quantities given the data and not a single prediction, uncertainty quantification is a natural consequence of Bayesian statistics. The benefits of machine learning in developing surrogate models for solving inverse problems are discussed, as well as progress in quantitatively understanding the errors that such a model introduces.
- 1 May 2024
- Published by AIP Publishing in Physics of Plasmas
- Vol. 31 (5)
Open AccessThinking Bayesian for plasma physicists
Bayesian statistics offers a powerful technique for plasma physicists to infer knowledge from the heterogeneous data types encountered. To explain this power, a simple example, Gaussian Process Regression, and the application of Bayesian statistics to inverse problems are explained. The likelihood is the key distribution because it contains the data model, or theoretic predictions, of the desired quantities. By using prior knowledge, the distribution of the inferred quantities of interest based on the data given can be inferred. Because it is a distribution of inferred quantities given the data and not a single prediction, uncertainty quantification is a natural consequence of Bayesian statistics. The benefits of machine learning in developing surrogate models for solving inverse problems are discussed, as well as progress in quantitatively understanding the errors that such a model introduces.
- Open Access
- 12 November 2023
- Published by Association for Computing Machinery (ACM)
- p. 1939-1948
Performance-Portable GPU Acceleration of the EFIT Tokamak Plasma Equilibrium Reconstruction Code
This paper presents the steps followed to GPU-offload parts of the core solver of EFIT-AI, an equilibrium reconstruction code suitable for tokamak experiments and burning plasmas. For this work, we will focus on the fitting procedure that consists of a Grad–Shafranov (GS) equation inverse solver that calculates equilibrium reconstructions on a grid. We will show profiling results of the original code (CPU-baseline), as well as the directives used to GPU-offload the most time-consuming function, initially to compare OpenACC and OpenMP on NVIDIA and AMD GPUs and later on to assess OpenMP performance portability on NVIDIA, AMD and Intel GPUs. We will make a performance comparison for different spatial grid sizes and show the speedup achieved on NVIDIA A100 (Perlmutter-NERSC), AMD MI250X (Frontier-OLCF) and Intel PVC GPUs (Sunspot-ALCF). Finally, we will draw some conclusions and recommendations to achieve high-performance portability for an equilibrium reconstruction code on the new HPC architectures.
- 12 November 2023
- Published by Association for Computing Machinery (ACM)
- p. 1939-1948
Open AccessPerformance-Portable GPU Acceleration of the EFIT Tokamak Plasma Equilibrium Reconstruction Code
This paper presents the steps followed to GPU-offload parts of the core solver of EFIT-AI, an equilibrium reconstruction code suitable for tokamak experiments and burning plasmas. For this work, we will focus on the fitting procedure that consists of a Grad–Shafranov (GS) equation inverse solver that calculates equilibrium reconstructions on a grid. We will show profiling results of the original code (CPU-baseline), as well as the directives used to GPU-offload the most time-consuming function, initially to compare OpenACC and OpenMP on NVIDIA and AMD GPUs and later on to assess OpenMP performance portability on NVIDIA, AMD and Intel GPUs. We will make a performance comparison for different spatial grid sizes and show the speedup achieved on NVIDIA A100 (Perlmutter-NERSC), AMD MI250X (Frontier-OLCF) and Intel PVC GPUs (Sunspot-ALCF). Finally, we will draw some conclusions and recommendations to achieve high-performance portability for an equilibrium reconstruction code on the new HPC architectures.