Multiscale simulation of plasma flows using active learning

Diaw, Abdourahmane; Barros, Kipton; Haack, J.; Junghans, Christoph; Keenan, Brett; Li, Y W; Livescu, Daniel; Lubbers, Nicholas; McKerns, Michael; Pavel, Robert S.; Rosenberger, David; Sagert, Irina; Germann, Timothy C.

doi:10.1103/physreve.102.023310

Cited by 19 publications

(18 citation statements)

References 52 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For instance, the complex transient interplay between reactive plasma and surface dynamics inherent to plasma catalysis or atmospheric pressure plasma in contact with surfaces/liquids may only be resolved with data-driven PSI models. In this context, data-driven chemical reaction pathway analysis [611] and active/transfer learning strategies for computationally costly atomic scale simulations [612] should be considered. Data-driven PSI models may ultimately permit a continuous and high fidelity physical description of technological plasmas, providing guidelines for future research and exploration.…”

Section: Data Management In Manufacturingmentioning

confidence: 99%

2022 Review of Data-Driven Plasma Science

Archibald¹,

Asif²,

Becker³

et al. 2022

Preprint

View full text Add to dashboard Cite

Data science and technology offer transformative tools and methods to science. This review article highlights latest development and progress in the interdisciplinary field of data-driven plasma science (DDPS). A large amount of data and machine learning algorithms go hand in hand. Most plasma data, whether experimental, observational or computational, are generated or collected by machines today. It is now becoming impractical for humans to analyze all the data manually. Therefore, it is imperative to train machines to analyze and interpret (eventually) such data as intelligently as humans but far more efficiently in quantity. Despite the recent impressive progress in applications of data science to plasma science and technology, the emerging field of DDPS is still in its infancy. Fueled by some of the most challenging problems such as fusion energy, plasmaprocessing of materials, and fundamental understanding of the universe through observable plasma phenomena, it is expected that DDPS continues to benefit significantly from the interdisciplinary marriage between plasma science and data science into the foreseeable future. CONTENTSI. Introduction 6 II. Fundamental Data Science 6 A. Introduction 6 B. Data Reduction/Compression 7 C. Dimensional reduction and sparse modeling 8 D. ML-enhanced modeling and simulation 9 E. ML Hardware and integration with models 10 F. Workflow Automation 11 G. Uncertainty Quantification 12 H. Visualization and Data Understanding 13 I. ML Control Theory 15 III. Basic Plasma Physics and Laboratory Experiments A. Introduction B. Spectroscopy, imaging and tomography C. Sparse measurement and noise D. Synthetic instruments and data E. Experimental data visualization F. High-rep rate laser experiments G. Charged particle beams H. Control and Optimisation of Plasma Accelerator Experiments I. Dusty and complex plasmas J. Physics and machine learning K. Challenges and outlook 5 IV. Magnetic Confinement Fusion 36 A. Introduction 36 B. Data-Driven Physics Models 36 C. Optimizing experimental workflows with data-driven methods 37 D. Diagnostics and Fusion Data Streams 38 E. Prediction of Tokamak Disruption 39 F. Surrogate models of fusion plasma 40 G. Magnetic Fusion Energy Data Challenges and Solutions 41 H. Data Science for extreme scale simulation 43 I. Challenges and outlook 44 V. Inertial confinement fusion and high-energy-density physics 45 A. Introduction 45 B. Representation learning for multimodal data 45 C. Transfer learning for simulation and experimen 47 D. Uncertainty quantification and Bayesian inference 47 E. High-performance computing and simulation acceleration 49 F. Design exploration and optimization 49 G. Self-driving experimental facilities 51 H. Challenges and outlook 52 VI. Space and astronomical plasmas 52 A. Introduction 52 B. Space and ground instruments 53 C. Space weather prediction 55 D. Transfer learning to improve historic data 56 E. Surrogate models of fluid closures using machine learning 56 F. Magnetic reconnection 58 G. Challenges and outlook 60 VII. Plasma tec...

show abstract

Section: Data Management In Manufacturingmentioning

confidence: 99%

2022 Review of Data-Driven Plasma Science

Archibald¹,

Asif²,

Becker³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Notably, the MSE standard deviation σ MSE from the ensemble of ANNs with identical HPs (indicated by the error bars in Figure 10) provides a metric for the uncertainty of the prediction. This is, for instance, utilized in the context of active learning to initiate the calculation of additional data samples when the uncertainty is above a predefined threshold [25].…”

Section: Comparison With the Estimated Ground Truthmentioning

confidence: 99%

“…In addition, establishing an interface model by manually prioritizing and implementing all relevant interactions becomes a tedious task [23,24]. In contrast, machine learning models (e.g., artificial neural networks, ANNs), have been used to generalize complex correlations and create surrogate models in the frame of plasma or gas-phase interactions with surfaces [25][26][27]. In particular, the feasibility and accuracy of this concept related to the prediction of energy-angular distributions (EADs) of sputtered particles as a function of the impinging projectile ion energy distribution (IED) has been demonstrated [1].…”

Section: Introductionmentioning

confidence: 99%

An efficient plasma-surface interaction surrogate model for sputtering processes based on autoencoder neural networks

Gergs,

Borislavov,

Trieschmann

2021

Preprint

View full text Add to dashboard Cite

Simulations of thin film sputter deposition require the separation of the plasma and material transport in the gas-phase from the growth/sputtering processes at the bounding surfaces (e.g., substrate and target). Interface models based on analytic expressions or look-up tables inherently restrict this complex interaction to a bare minimum. A machine learning model has recently been shown to overcome this remedy for Ar ions bombarding a Ti-Al composite target [1]. However, the chosen network structure (i.e., a multilayer perceptron, MLP) provides approximately 4 million degrees of freedom, which bears the risk of overfitting the relevant dynamics and complicating the model to an unreliable extend. This work proposes a conceptually more sophisticated but parameterwise simplified regression artificial neural network for an extended scenario, considering a variable instead of a single fixed Ti-Al stoichiometry. A convolutional β-variational autoencoder is trained to reduce the high-dimensional energy-angular distribution of sputtered particles to a latent space representation of only two components. In addition to a primary decoder which is trained to reconstruct the input energy-angular distribution, a secondary decoder is employed to reconstruct the mean energy of incident Ar ions as well as the present Ti-Al composition. The mutual latent space is hence conditioned on these quantities. The trained primary decoder of the variational autoencoder network is subsequently transferred to a regression network, for which only the mapping to the particular latent space has to be learned. While obtaining a competitive performance, the number of degrees of freedom is drastically reduced to 15,111 (0.378 % of the MLP) and 486 (0.012 % of the MLP) parameters for the primary decoder and the remaining regression network, respectively. The underlying methodology is very general and can easily be extended to more complex physical descriptions (e.g., taking into account dynamical surface properties) with a minimal amount of data required.

show abstract

“…Berg model [21,26], Sigmund-Thompson theory [11][12][13]), look-up tables). This issue has been addressed in previous works by proposing data-driven machine learning plasma-surface interaction (PSI) surrogate models that are intended to complement plasma simulations [22,[27][28][29][30][31][32][33]. In particular, a multi layer perceptron (MLP) was trained to generalize on data describing the sputtering of a Ti 0.5 Al 0.5 composite target, simulated with TRIDYN [7,22].…”

Section: Introductionmentioning

confidence: 99%

Physics-separating artificial neural networks for predicting initial stages of Al sputtering and thin film deposition in Ar plasma discharges

Gergs¹,

Mussenbrock²,

Trieschmann³

2022

Preprint

View full text Add to dashboard Cite

Simulations of Al thin film sputter depositions rely on accurate plasma and surface interaction models. Establishing the latter commonly requires a higher level of abstraction and means to dismiss the fundamental atomic fidelity. Previous works on sputtering processes addressed this issue by establishing machine learning surrogate models, which include a basic surface state (i.e., stoichiometry) as static input. In this work, an evolving surface state and defect structure are introduced to jointly describe sputtering and growth with physics-separating artificial neural networks. The data describing the plasma-surface interactions stem from hybrid reactive molecular dynamics/time-stamped force bias Monte Carlo simulations of Al neutrals and Ar + ions impinging onto Al(001) surfaces. It is demonstrated that the fundamental processes are comprehensively described by taking the surface state as well as defect structure into account. Hence, a machine learning plasma-surface interaction surrogate model is established that resolves the inherent kinetics with high physical fidelity. The resulting model is not restricted to input from modeling and simulation, but may similarly be applied to experimental input data.

show abstract

Multiscale simulation of plasma flows using active learning

Cited by 19 publications

References 52 publications

2022 Review of Data-Driven Plasma Science

2022 Review of Data-Driven Plasma Science

An efficient plasma-surface interaction surrogate model for sputtering processes based on autoencoder neural networks

Physics-separating artificial neural networks for predicting initial stages of Al sputtering and thin film deposition in Ar plasma discharges

Contact Info

Product

Resources

About