Xu Yang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Xu Yang has a total of 74 co-authors. Here are the top co-authors by number of shared publications:
- Yueqiang Liu
General Atomics
12 shared publications - Yuling He
Key Laboratory of Atomic and Subatomic Structure and Quantum Control (Ministry of Education), Guangdong Basic Research Center of Excellence for Structure and Fundamental Interactions of Matter, School of Physics, South China Normal University 4 , Guangzhou 510006
7 shared publications Li Li
College of Science, Donghua University, Shanghai 201620, People’s Republic of China
6 shared publications- Guoliang Xia
UKAEA, Culham Campus 6 , Abingdon, OX14 3DB
5 shared publications - Lina Zhou
School of Science, Dalian Maritime University, Dalian 116026, People’s Republic of China
5 shared publications - Guanqi Dong
Southwestern Institute of Physics
4 shared publications
Latest Publications
- 1 September 2026
- Published by IOP Publishing in Plasma Physics and Controlled Fusion
- Vol. 68 (9) , 095030
Neural network surrogate model for drift-kinetic energy perturbation for internal kink instability
Multi-layer perceptron based neural networks (NNs) are trained to predict the ideal internal kink (IK) instability in tokamak plasmas, with particular emphasis on the drift-kinetic energy perturbation associated with thermal particles. The NN surrogate model is trained and validated on a semi-analytic database and tested on both held-out semi-analytic data and five EU DEMO design cases. A well-trained NN predicts the amplitude and phase of the perturbed drift-kinetic potential energy with the coefficient of determination ( R 2 ) reaching 0.88, and 0.98, respectively. The NN also achieves high R 2 values of 0.94 and 0.92 for the fluid growth rate and fluid perturbed potential energy, respectively. These results, together with low mean absolute errors, demonstrate the effectiveness of the semi-analytic training approach and establish a foundation for real-time IK stability monitoring and sawtooth period estimation in tokamak experiments. The physics-prediction reliability is further confirmed through case study of an EU DEMO negative-triangularity equilibrium, where the NN successfully reproduces the theoretical linear relationship between the fluid growth rate and the fluid perturbed potential energy and captures the dependence of IK instability on multiple plasma parameters. Collectively, these results confirm that the trained NN effectively captures essential physical characteristics of the IK instability.
- 1 September 2026
- Published by IOP Publishing in Plasma Physics and Controlled Fusion
- Vol. 68 (9) , 095030
Neural network surrogate model for drift-kinetic energy perturbation for internal kink instability
Multi-layer perceptron based neural networks (NNs) are trained to predict the ideal internal kink (IK) instability in tokamak plasmas, with particular emphasis on the drift-kinetic energy perturbation associated with thermal particles. The NN surrogate model is trained and validated on a semi-analytic database and tested on both held-out semi-analytic data and five EU DEMO design cases. A well-trained NN predicts the amplitude and phase of the perturbed drift-kinetic potential energy with the coefficient of determination ( R 2 ) reaching 0.88, and 0.98, respectively. The NN also achieves high R 2 values of 0.94 and 0.92 for the fluid growth rate and fluid perturbed potential energy, respectively. These results, together with low mean absolute errors, demonstrate the effectiveness of the semi-analytic training approach and establish a foundation for real-time IK stability monitoring and sawtooth period estimation in tokamak experiments. The physics-prediction reliability is further confirmed through case study of an EU DEMO negative-triangularity equilibrium, where the NN successfully reproduces the theoretical linear relationship between the fluid growth rate and the fluid perturbed potential energy and captures the dependence of IK instability on multiple plasma parameters. Collectively, these results confirm that the trained NN effectively captures essential physical characteristics of the IK instability.
- 22 June 2026
- Published by IOP Publishing in Plasma Physics and Controlled Fusion
- Vol. 68 (7)
TORQSNet: A physics-informed spectral neural network for predicting toroidal torque density in tokamak devices
Fast prediction of torque density profiles driven by three-dimensional (3D) magnetic fields is required for accurate real-time control of plasma toroidal rotation. A toroidal torque spectral neural network (TORQSNet) is developed to enable rapid and accurate prediction of radial profiles of neoclassical toroidal viscosity (NTV) torque, electromagnetic torque, and Reynolds stress torque under the resonant magnetic perturbations (RMP). TORQSNet is trained and validated using a numerical database of 3D equilibria computed using the MARS-F code for ASDEX Upgrade (AUG), DIII-D, MAST, and ITER. In oscillatory radial regions, torque density profiles are parameterized by Legendre polynomial expansion coefficients, which are predicted by the network and are used as interpretable features linked to resonant response characteristics in magnetohydrodynamic equilibria. Across all four devices and the investigated toroidal mode numbers, coefficients of determination of R2>0.946 are obtained for reconstructed NTV torque density profiles. Strong cross-device generalization is demonstrated, with models trained on a single device accurately predicting radial torque profiles in the other three devices. The combined influence of plasma rotation and resistivity is captured by the trained TORQSNet. Core torque is reduced while edge torque is increased, which supports optimization of RMP-based edge localized mode suppression by promoting pedestal penetration and mitigating core flow damping.
- 22 June 2026
- Published by IOP Publishing in Plasma Physics and Controlled Fusion
- Vol. 68 (7)
TORQSNet: A physics-informed spectral neural network for predicting toroidal torque density in tokamak devices
Fast prediction of torque density profiles driven by three-dimensional (3D) magnetic fields is required for accurate real-time control of plasma toroidal rotation. A toroidal torque spectral neural network (TORQSNet) is developed to enable rapid and accurate prediction of radial profiles of neoclassical toroidal viscosity (NTV) torque, electromagnetic torque, and Reynolds stress torque under the resonant magnetic perturbations (RMP). TORQSNet is trained and validated using a numerical database of 3D equilibria computed using the MARS-F code for ASDEX Upgrade (AUG), DIII-D, MAST, and ITER. In oscillatory radial regions, torque density profiles are parameterized by Legendre polynomial expansion coefficients, which are predicted by the network and are used as interpretable features linked to resonant response characteristics in magnetohydrodynamic equilibria. Across all four devices and the investigated toroidal mode numbers, coefficients of determination of R2>0.946 are obtained for reconstructed NTV torque density profiles. Strong cross-device generalization is demonstrated, with models trained on a single device accurately predicting radial torque profiles in the other three devices. The combined influence of plasma rotation and resistivity is captured by the trained TORQSNet. Core torque is reduced while edge torque is increased, which supports optimization of RMP-based edge localized mode suppression by promoting pedestal penetration and mitigating core flow damping.
- 1 April 2026
- Published by AIP Publishing in Physics of Plasmas
- Vol. 33 (4)
Two-fluid plasma response to resonant magnetic perturbations with varying plasma elongation and triangularity
Two-fluid effects on the plasma response to applied resonant magnetic perturbations of toroidal mode numbers n = 1–4 are numerically investigated, with systematic variation of plasma boundary shaping (elongation and triangularity) for both the single-null and double-null (DN) divertor-like tokamak configurations. Two-fluid response models yield similar optimum coil phasings for best edge-localized modes (ELMs) control, to those predicted by the single-fluid counterpart. This explains the previous success of the modeled spectrum optimization for ELMs control experiments using the single-fluid approach. Both single- and two-fluid models predict reduced optimal coil phasing at enhanced plasma elongation, while triangularity variations have minimal influence. However, two-fluid models predict a stronger screening of edge resonant field perturbations than single-fluid counterpartment, due to faster E × B and/or electron flows enclosed in the former in the edge pedestal region. Finally, both predict a weaker edge-peeling response in DN plasmas, consistent with the experimental findings of difficulties in suppressing ELMs in DN plasmas.
- 1 April 2026
- Published by AIP Publishing in Physics of Plasmas
- Vol. 33 (4)
Two-fluid plasma response to resonant magnetic perturbations with varying plasma elongation and triangularity
Two-fluid effects on the plasma response to applied resonant magnetic perturbations of toroidal mode numbers n = 1–4 are numerically investigated, with systematic variation of plasma boundary shaping (elongation and triangularity) for both the single-null and double-null (DN) divertor-like tokamak configurations. Two-fluid response models yield similar optimum coil phasings for best edge-localized modes (ELMs) control, to those predicted by the single-fluid counterpart. This explains the previous success of the modeled spectrum optimization for ELMs control experiments using the single-fluid approach. Both single- and two-fluid models predict reduced optimal coil phasing at enhanced plasma elongation, while triangularity variations have minimal influence. However, two-fluid models predict a stronger screening of edge resonant field perturbations than single-fluid counterpartment, due to faster E × B and/or electron flows enclosed in the former in the edge pedestal region. Finally, both predict a weaker edge-peeling response in DN plasmas, consistent with the experimental findings of difficulties in suppressing ELMs in DN plasmas.
- Open Access
- 20 February 2026
- Published by Elsevier in SSRN Electronic Journal
RMP-NET: A Multi-Device Deep Learning Surrogate for Fast Prediction of Plasma Response to Resonant Magnetic Perturbations
Edge Localized Modes (ELMs) pose a major threat to the operational stability of future tokamaks, necessitating effective control strategies such as Resonant Magnetic Perturbations (RMP). However, the optimization of RMP parameters heavily relies on high-fidelity numerical codes such as MARS-F (a toroidal magnetohydrodynamic stability code). Although the code can simulate the plasma response under external RMP, each simulation takes hours to complete. This high computational cost prohibits rapid parameter scans and hinders the potential for real-time control. Deep learning provides a promising pathway for developing fast and accurate surrogate models to address this limitation. In this work, we introduce RMP-NET (Resonant Magnetic Perturbations-Network), a deep learning-based surrogate model that integrates 3D feature reconstruction and a Magnetic Field Perturbation Perception Parallel Attention (MF3PA) mechanism. Built upon the MARS-F framework, the model achieves orders-of-magnitude acceleration in predicting RMP-controlled ELMs. To improve generalizability beyond single-device data, we constructed a multi-device dataset consisting of 13,077 samples from four tokamaks: AUG, DIII-D, ITER, and MAST. Using this dataset, we developed the RMP-NET model for ELM prediction. Experimental results demonstrate that, compared to conventional numerical simulation approaches like the MARS-F code, RMP-NET achieves a dramatic reduction in computation time—from about an hour per simulation to millisecond-scale inference—albeit with a slight trade-off in physical accuracy. This advancement establishes a solid foundation for real-time ELM control. Collectively, RMP-NET offers a scalable deep learning framework tailored for real-time ELM forecasting and RMP optimization.
- 20 February 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessRMP-NET: A Multi-Device Deep Learning Surrogate for Fast Prediction of Plasma Response to Resonant Magnetic Perturbations
Edge Localized Modes (ELMs) pose a major threat to the operational stability of future tokamaks, necessitating effective control strategies such as Resonant Magnetic Perturbations (RMP). However, the optimization of RMP parameters heavily relies on high-fidelity numerical codes such as MARS-F (a toroidal magnetohydrodynamic stability code). Although the code can simulate the plasma response under external RMP, each simulation takes hours to complete. This high computational cost prohibits rapid parameter scans and hinders the potential for real-time control. Deep learning provides a promising pathway for developing fast and accurate surrogate models to address this limitation. In this work, we introduce RMP-NET (Resonant Magnetic Perturbations-Network), a deep learning-based surrogate model that integrates 3D feature reconstruction and a Magnetic Field Perturbation Perception Parallel Attention (MF3PA) mechanism. Built upon the MARS-F framework, the model achieves orders-of-magnitude acceleration in predicting RMP-controlled ELMs. To improve generalizability beyond single-device data, we constructed a multi-device dataset consisting of 13,077 samples from four tokamaks: AUG, DIII-D, ITER, and MAST. Using this dataset, we developed the RMP-NET model for ELM prediction. Experimental results demonstrate that, compared to conventional numerical simulation approaches like the MARS-F code, RMP-NET achieves a dramatic reduction in computation time—from about an hour per simulation to millisecond-scale inference—albeit with a slight trade-off in physical accuracy. This advancement establishes a solid foundation for real-time ELM control. Collectively, RMP-NET offers a scalable deep learning framework tailored for real-time ELM forecasting and RMP optimization.
- 9 February 2026
- Published by IOP Publishing in Plasma Science and Technology
- Vol. 28 (4)
Two-fluid effects on resistive wall mode stability in ITER
A systematic numerical investigation is carried out to examine the two-fluid effects on the stability of the n=1 (n is the toroidal mode number) resistive wall mode (RWM) in an ITER 10 MA/5.3 T/Q = 5 steady-state scenario. Results are compared with the single-fluid prediction while varying (i) the plasma fluid flow, (ii) the plasma pressure, and (iii) the strength of the parallel sound wave damping (PSWD) mimicking the ion-Landau kinetic resonant effect. Inclusion of two-fluid effects opens a stability window for the RWM in the slow plasma regime (with the flow speed well below 1% of the toroidal Alfvén speed) in ITER. This stability window is robust against variation of the plasma pressure (within the RWM regime) when combined with strong PSWD. A second stability window for the mode is also identified at faster plasma (fluid) flow according to the two-fluid model, similar to that obtained with the single-fluid model. Partial cancellation of the fluid flow by the thermal ion diamagnetic flow, within the two-fluid model, however results in a larger rotation threshold (in the fast-flow regime) for the mode stabilization than the single-fluid prediction. A two-fluid model with instantaneous thermal equilibration, on the other hand, finds comparable rotation threshold to the single-fluid counterpart in the fast-flow regime.
- 9 February 2026
- Published by IOP Publishing in Plasma Science and Technology
- Vol. 28 (4)
Two-fluid effects on resistive wall mode stability in ITER
A systematic numerical investigation is carried out to examine the two-fluid effects on the stability of the n=1 (n is the toroidal mode number) resistive wall mode (RWM) in an ITER 10 MA/5.3 T/Q = 5 steady-state scenario. Results are compared with the single-fluid prediction while varying (i) the plasma fluid flow, (ii) the plasma pressure, and (iii) the strength of the parallel sound wave damping (PSWD) mimicking the ion-Landau kinetic resonant effect. Inclusion of two-fluid effects opens a stability window for the RWM in the slow plasma regime (with the flow speed well below 1% of the toroidal Alfvén speed) in ITER. This stability window is robust against variation of the plasma pressure (within the RWM regime) when combined with strong PSWD. A second stability window for the mode is also identified at faster plasma (fluid) flow according to the two-fluid model, similar to that obtained with the single-fluid model. Partial cancellation of the fluid flow by the thermal ion diamagnetic flow, within the two-fluid model, however results in a larger rotation threshold (in the fast-flow regime) for the mode stabilization than the single-fluid prediction. A two-fluid model with instantaneous thermal equilibration, on the other hand, finds comparable rotation threshold to the single-fluid counterpart in the fast-flow regime.
- Open Access
- 1 September 2025
- Published by AIP Publishing in Physics of Plasmas
- Vol. 32 (9)
Effects of parallel and poloidal flows on tearing mode stability in tokamak plasma
Effects of the plasma parallel and poloidal flows, as well as the flow shear, on the tearing mode stability are investigated in conjunction with the toroidal flow and the favorable average curvature stabilization, utilizing an updated version of the MARS-F code [Liu et al., Phys. Plasmas7, 3681–3690 (2000)]. With vanishing plasma equilibrium pressure, both parallel and poloidal flows destabilize the mode although the dominant effect still comes from the toroidal flow. At finite equilibrium pressure and in the presence of favorable average curvature effect, the parallel and toroidal flows synergistically stabilize the tearing mode. This synergy does not occur for the combination of the toroidal and poloidal flows, since the latter tends to weaken the favorable curvature stabilization. The parallel flow on the other hand enhances the curvature stabilization. Finally, although the tearing mode may not be completely stabilized, the growth rate of the mode is reduced with finite equilibrium pressure for all values of flow shear, with the suppression being more pronounced at low or moderate flow shear.
- 1 September 2025
- Published by AIP Publishing in Physics of Plasmas
- Vol. 32 (9)
Open AccessEffects of parallel and poloidal flows on tearing mode stability in tokamak plasma
Effects of the plasma parallel and poloidal flows, as well as the flow shear, on the tearing mode stability are investigated in conjunction with the toroidal flow and the favorable average curvature stabilization, utilizing an updated version of the MARS-F code [Liu et al., Phys. Plasmas7, 3681–3690 (2000)]. With vanishing plasma equilibrium pressure, both parallel and poloidal flows destabilize the mode although the dominant effect still comes from the toroidal flow. At finite equilibrium pressure and in the presence of favorable average curvature effect, the parallel and toroidal flows synergistically stabilize the tearing mode. This synergy does not occur for the combination of the toroidal and poloidal flows, since the latter tends to weaken the favorable curvature stabilization. The parallel flow on the other hand enhances the curvature stabilization. Finally, although the tearing mode may not be completely stabilized, the growth rate of the mode is reduced with finite equilibrium pressure for all values of flow shear, with the suppression being more pronounced at low or moderate flow shear.
- Open Access
- 29 August 2025
- Published by IOP Publishing in Plasma Physics and Controlled Fusion
- Vol. 67 (8) , 085040
Effects of equilibrium pressure on plasma response to RMPs in a spherical tokamak
This study presents a comprehensive analysis of the equilibrium pressure on the plasma response to resonant magnetic perturbations (RMPs) in the spherical tokamak (ST) MAST-U, employing both single-fluid and MHD-kinetic hybrid models (implemented via the MARS-F/K codes). As a key finding, the study identifies two different pressure-driven eigenmodes, exhibiting Sturmian property, that affect the Troyon no-wall limits for the onset of the n = 1 and n = 2 (n is the toroidal mode number) ideal external kink instabilities as well as the corresponding plasma response to the applied RMP. With increasing equilibrium pressure, the plasma response to RMPs is significantly enhanced in the ST plasma, particularly in the high-pressure regime where kinetic effects strongly stabilize the external kink instability. The Troyon no-wall limit divides the plasma response into two regions: well below the limit, the response amplitudes and trends (versus pressure) are similar between the fluid and kinetic models; as the equilibrium pressure approaches the Troyon limit, the kinetic model predicts significant amplification of the RMP field, up to 30 times for cases considered. A relatively weak dependence of the optimal coil phasing on the equilibrium pressure is computed in this ST plasma, similar to the trend obtained for the conventional aspect ratio devices. These findings underscore the importance of incorporating kinetic effects in accurate prediction of the plasma response to RMPs in high-pressure ST tokamak plasmas and provide a theoretical basis for optimizing RMP-based control of the edge-localized modes in future ST devices.
- 29 August 2025
- Published by IOP Publishing in Plasma Physics and Controlled Fusion
- Vol. 67 (8) , 085040
Open AccessEffects of equilibrium pressure on plasma response to RMPs in a spherical tokamak
This study presents a comprehensive analysis of the equilibrium pressure on the plasma response to resonant magnetic perturbations (RMPs) in the spherical tokamak (ST) MAST-U, employing both single-fluid and MHD-kinetic hybrid models (implemented via the MARS-F/K codes). As a key finding, the study identifies two different pressure-driven eigenmodes, exhibiting Sturmian property, that affect the Troyon no-wall limits for the onset of the n = 1 and n = 2 (n is the toroidal mode number) ideal external kink instabilities as well as the corresponding plasma response to the applied RMP. With increasing equilibrium pressure, the plasma response to RMPs is significantly enhanced in the ST plasma, particularly in the high-pressure regime where kinetic effects strongly stabilize the external kink instability. The Troyon no-wall limit divides the plasma response into two regions: well below the limit, the response amplitudes and trends (versus pressure) are similar between the fluid and kinetic models; as the equilibrium pressure approaches the Troyon limit, the kinetic model predicts significant amplification of the RMP field, up to 30 times for cases considered. A relatively weak dependence of the optimal coil phasing on the equilibrium pressure is computed in this ST plasma, similar to the trend obtained for the conventional aspect ratio devices. These findings underscore the importance of incorporating kinetic effects in accurate prediction of the plasma response to RMPs in high-pressure ST tokamak plasmas and provide a theoretical basis for optimizing RMP-based control of the edge-localized modes in future ST devices.
- Open Access
- 10 June 2025
- Published by IOP Publishing in Nuclear Fusion
- Vol. 65 (7) , 076019
Two-fluid effects on resistive wall modes in plasmas with negative versus positive triangularity
The MARS-F code (Liu et al 2000 Phys. Plasmas 7 3681), has been updated to include two-fluid (2F) effects for linear stability analysis of magneto-hydrodynamic modes. The updated code, referred to as MARS-2F, is then applied to study flow stabilization of the resistive wall mode (RWM) in a reactor-scale negative-triangularity (NT) plasma, with the results also compared to that obtained in a positive-triangularity counterpart. The RWM is computed to be stable in plasmas with slow or even vanishing toroidal fluid flow according to the 2F-model, with much wider stability window for the NT plasma. This stability regime is attributed to the E × B flow stabilization, which is finite even at vanishing (single) fluid flow since the diamagnetic flow remains finite. This finding provides an alternative interpretation of the RWM stability in the slow-flow regime in tokamak plasmas as observed in experiments. Varying the fluid flow shear and a parallel sound wave damping model confirms the robustness of the 2F-model in predicting the stable window for the RWM in the slow-fluid-flow regime.
- 10 June 2025
- Published by IOP Publishing in Nuclear Fusion
- Vol. 65 (7) , 076019
Open AccessTwo-fluid effects on resistive wall modes in plasmas with negative versus positive triangularity
The MARS-F code (Liu et al 2000 Phys. Plasmas 7 3681), has been updated to include two-fluid (2F) effects for linear stability analysis of magneto-hydrodynamic modes. The updated code, referred to as MARS-2F, is then applied to study flow stabilization of the resistive wall mode (RWM) in a reactor-scale negative-triangularity (NT) plasma, with the results also compared to that obtained in a positive-triangularity counterpart. The RWM is computed to be stable in plasmas with slow or even vanishing toroidal fluid flow according to the 2F-model, with much wider stability window for the NT plasma. This stability regime is attributed to the E × B flow stabilization, which is finite even at vanishing (single) fluid flow since the diamagnetic flow remains finite. This finding provides an alternative interpretation of the RWM stability in the slow-flow regime in tokamak plasmas as observed in experiments. Varying the fluid flow shear and a parallel sound wave damping model confirms the robustness of the 2F-model in predicting the stable window for the RWM in the slow-fluid-flow regime.
- 9 January 2025
- Published by Springer Nature in The European Physical Journal Special Topics
- Vol. 234 (13) , 3321-3343
Influence of pedestal pressure on plasma response to resonant magnetic perturbation field
- 9 January 2025
- Published by Springer Nature in The European Physical Journal Special Topics
- Vol. 234 (13) , 3321-3343
Influence of pedestal pressure on plasma response to resonant magnetic perturbation field
- 30 September 2024
- Published by IOP Publishing in Plasma Physics and Controlled Fusion
- Vol. 66 (11) , 115010
Helical resonant magnetic perturbation coils for controlling edge localized modes: a robustness study
Plasma response to helical resonant magnetic perturbation coil current is numerically computed for tokamak plasmas, with optimization results compared with that for conventional window-frame coils. The key aspect of study is the robustness of the proposed new concept against variation of plasma equilibrium parameters including (i) the plasma resistivity, (ii) the toroidal rotation and (iii) the plasma shaping (both elongation and triangularity). Toroidal modeling results yield several important conclusions. First, assuming the same coil current, the optimal helical coils robustly outperform the optimal window-frame coils against variation of the aforementioned plasma equilibrium parameters. Secondly, for a chosen toroidal spectrum, the optimal helical coil geometry including the poloidal location, poloidal coverage and the overall shape, is robust against variation of plasma parameters except the safety factor. Finally, in all cases, optimization based on the plasma response naturally yields a single row of helical coils located near the outboard mid-plane of the torus, ensuring a relatively simple design of the coil geometry.
- 30 September 2024
- Published by IOP Publishing in Plasma Physics and Controlled Fusion
- Vol. 66 (11) , 115010
Helical resonant magnetic perturbation coils for controlling edge localized modes: a robustness study
Plasma response to helical resonant magnetic perturbation coil current is numerically computed for tokamak plasmas, with optimization results compared with that for conventional window-frame coils. The key aspect of study is the robustness of the proposed new concept against variation of plasma equilibrium parameters including (i) the plasma resistivity, (ii) the toroidal rotation and (iii) the plasma shaping (both elongation and triangularity). Toroidal modeling results yield several important conclusions. First, assuming the same coil current, the optimal helical coils robustly outperform the optimal window-frame coils against variation of the aforementioned plasma equilibrium parameters. Secondly, for a chosen toroidal spectrum, the optimal helical coil geometry including the poloidal location, poloidal coverage and the overall shape, is robust against variation of plasma parameters except the safety factor. Finally, in all cases, optimization based on the plasma response naturally yields a single row of helical coils located near the outboard mid-plane of the torus, ensuring a relatively simple design of the coil geometry.