Estimation of plasma equilibrium parameters via a neural network approach*

Zhu, Zijian; Guo, Youming; Yang, Fei; Xiao, Bingjia; Li, Jiangang

doi:10.1088/1674-1056/ab55d1

Cited by 4 publications

(3 citation statements)

References 12 publications

(14 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Similarly the reconstruction and reconstruction-control modes predict the same outputs, but use diagnostic information such as magnetic probes as inputs to the NN. Previous neural net equilibrium solvers [15][16][17][18][19][20][21][22] were developed in the spirit of reconstruction-control mode-mapping diagnostics to shaping parameters-with some exceptions such as [23] which reconstructs flux surfaces and [24] which is a fixedboundary solver. Similar to previous efforts on other machines, we obtain good predictions using the magnetic diagnostics.…”

Section: Introductionmentioning

confidence: 99%

Neural net modeling of equilibria in NSTX-U

Wai,

Boyer,

Kolemen

2022

Preprint

View full text Add to dashboard Cite

Neural networks (NNs) offer a path towards synthesizing and interpreting data on faster timescales than traditional physics-informed computational models. In this work we develop two neural networks relevant to equilibrium and shape control modeling, which are part of a suite of tools being developed for the National Spherical Torus Experiment-Upgrade (NSTX-U) for fast prediction, optimization, and visualization of plasma scenarios. The networks include Eqnet, a freeboundary equilibrium solver trained on the EFIT01 reconstruction algorithm, and Pertnet, which is trained on the Gspert code and predicts the non-rigid plasma response, a nonlinear term that arises in shape control modeling. The equilibrium neural network is trained with different combinations of inputs and outputs in order to offer flexibility in use cases. In particular, the NN can use magnetic diagnostics as inputs for equilibrium prediction thus acting as a reconstruction code, or can use profiles and external currents as inputs to act as a traditional free-boundary Grad-Shafranov solver. We report strong performance for both networks indicating that these models could reliably be used within closed-loop simulations. Some limitations regarding generalizability and noise are discussed.

show abstract

Section: Introductionmentioning

confidence: 99%

Neural net modeling of equilibria in NSTX-U

Wai,

Boyer,

Kolemen

2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Similarly the reconstruction and reconstruction-control modes predict the same outputs, but use diagnostic information such as magnetic probes as inputs to the NN. Previous NN equilibrium solvers [15][16][17][18][19][20][21][22] were developed in the spirit of reconstruction-control mode-mapping diagnostics to shaping parameters-with some exceptions such as [23] which reconstructs flux surfaces and [24] which is a fixed-boundary solver.…”

Section: Introductionmentioning

confidence: 99%

Neural net modeling of equilibria in NSTX-U

Wai¹,

Boyer²,

Kolemen³

2022

Nucl. Fusion

View full text Add to dashboard Cite

Neural networks (NNs) offer a path towards synthesizing and interpreting data on faster timescales than traditional physics-informed computational models. In this work we develop two neural networks relevant to equilibrium and shape control modeling, which are part of a suite of tools being developed for the National Spherical Torus Experiment-Upgrade (NSTX-U) for fast prediction, optimization, and visualization of plasma scenarios. The networks include Eqnet, a free-boundary equilibrium solver trained on the EFIT01 reconstruction algorithm, and Pertnet, which is trained on the Gspert code and predicts the non-rigid plasma response, a nonlinear term that arises in shape control modeling. The NNs are trained with different combinations of inputs and outputs in order to offer flexibility in use cases. In particular, Eqnet can use magnetic diagnostics as inputs and act as an EFIT-like reconstruction algorithm, or, by using pressure and current profile information the NN can act as a forward Grad-Shafranov equilibrium solver. This forward-mode version is envisioned to be implemented in the suite of tools for simulation of plasma scenarios. The reconstruction-mode version gives some performance improvements compared to the online reconstruction code real-time EFIT (RTEFIT), especially when vessel eddy currents are significant. We report strong performance for all NNs indicating that the models could reliably be used within closed-loop simulations or other applications. Some limitations are discussed.

show abstract

“…[1][2][3][4][5][6][7][8][9][10][11][12][13] topological nature of quantum states, [14][15][16][17][18] many-body correlated effects, [19][20][21][22][23][24][25][26] optimization of numerical simulations, [27][28][29] quantum error correction codes, [30][31][32][33] and so on. [34][35][36][37] Recently, machine learning is also applied to the open quantum systems which have many applications in various fields such as solid state physics, quantum chemistry, quantum sensing, quantum information transport, and quantum computing. Luchnikov et al applied the supervised learning to the open quantum system.…”

Section: Introductionmentioning

confidence: 99%

Dynamical learning of non-Markovian quantum dynamics

Yang¹,

Cao²,

Yang³

2022

Chinese Phys. B

View full text Add to dashboard Cite

We study the non-Markovian dynamics of an open quantum system with machine learning. The observable physical quantities and their evolutions are generated by using the neural network. After the pre-training is completed, we fix the weights in the subsequent processes thus do not need the further gradient feedback. We find that the dynamical properties of physical quantities obtained by the dynamical learning are better than those obtained by the learning of Hamiltonian and time evolution operator. The dynamical learning can be applied to other quantum many-body systems, non-equilibrium statistics and random processes.

show abstract

Estimation of plasma equilibrium parameters via a neural network approach*

Cited by 4 publications

References 12 publications

Neural net modeling of equilibria in NSTX-U

Neural net modeling of equilibria in NSTX-U

Neural net modeling of equilibria in NSTX-U

Dynamical learning of non-Markovian quantum dynamics

Contact Info

Product

Resources

About