-
Predictive Scale-Bridging Simulations through Active Learning
Authors:
Satish Karra,
Mohamed Mehana,
Nicholas Lubbers,
Yu Chen,
Abdourahmane Diaw,
Javier E. Santos,
Aleksandra Pachalieva,
Robert S. Pavel,
Jeffrey R. Haack,
Michael McKerns,
Christoph Junghans,
Qinjun Kang,
Daniel Livescu,
Timothy C. Germann,
Hari S. Viswanathan
Abstract:
Throughout computational science, there is a growing need to utilize the continual improvements in raw computational horsepower to achieve greater physical fidelity through scale-bridging over brute-force increases in the number of mesh elements. For instance, quantitative predictions of transport in nanoporous media, critical to hydrocarbon extraction from tight shale formations, are impossible w…
▽ More
Throughout computational science, there is a growing need to utilize the continual improvements in raw computational horsepower to achieve greater physical fidelity through scale-bridging over brute-force increases in the number of mesh elements. For instance, quantitative predictions of transport in nanoporous media, critical to hydrocarbon extraction from tight shale formations, are impossible without accounting for molecular-level interactions. Similarly, inertial confinement fusion simulations rely on numerical diffusion to simulate molecular effects such as non-local transport and mixing without truly accounting for molecular interactions. With these two disparate applications in mind, we develop a novel capability which uses an active learning approach to optimize the use of local fine-scale simulations for informing coarse-scale hydrodynamics. Our approach addresses three challenges: forecasting continuum coarse-scale trajectory to speculatively execute new fine-scale molecular dynamics calculations, dynamically updating coarse-scale from fine-scale calculations, and quantifying uncertainty in neural network models.
△ Less
Submitted 20 September, 2022;
originally announced September 2022.
-
Efficient Learning of Accurate Surrogates for Simulations of Complex Systems
Authors:
A. Diaw,
M. McKerns,
I. Sagert,
L. G. Stanton,
M. S. Murillo
Abstract:
Machine learning methods are increasingly used to build computationally inexpensive surrogates for complex physical models. The predictive capability of these surrogates suffers when data are noisy, sparse, or time-dependent. As we are interested in finding a surrogate that provides valid predictions of any potential future model evaluations, we introduce an online learning method empowered by opt…
▽ More
Machine learning methods are increasingly used to build computationally inexpensive surrogates for complex physical models. The predictive capability of these surrogates suffers when data are noisy, sparse, or time-dependent. As we are interested in finding a surrogate that provides valid predictions of any potential future model evaluations, we introduce an online learning method empowered by optimizer-driven sampling. The method has two advantages over current approaches. First, it ensures that all turning points on the model response surface are included in the training data. Second, after any new model evaluations, surrogates are tested and "retrained" (updated) if the "score" drops below a validity threshold. Tests on benchmark functions reveal that optimizer-directed sampling generally outperforms traditional sampling methods in terms of accuracy around local extrema, even when the scoring metric favors overall accuracy. We apply our method to simulations of nuclear matter to demonstrate that highly accurate surrogates for the nuclear equation of state can be reliably auto-generated from expensive calculations using a few model evaluations.
△ Less
Submitted 17 May, 2024; v1 submitted 11 July, 2022;
originally announced July 2022.
-
Impact of Electron Transport Models on Capillary Discharge Plasmas
Authors:
A. Diaw,
S. J. Coleman,
N. M. Cook,
J. Edelen,
E. C. Hansen,
P. Tzeferacos
Abstract:
Magnetohydrodynamics (MHD) can be used to model capillary discharge waveguides in laser-wakefield accelerators. However, the predictive capability of MHD can suffer due to poor microscopic closure models. Here, we study the impact of electron heating and thermal conduction on capillary waveguide performance as part of an effort to understand and quantify uncertainties in modeling and designing nex…
▽ More
Magnetohydrodynamics (MHD) can be used to model capillary discharge waveguides in laser-wakefield accelerators. However, the predictive capability of MHD can suffer due to poor microscopic closure models. Here, we study the impact of electron heating and thermal conduction on capillary waveguide performance as part of an effort to understand and quantify uncertainties in modeling and designing next-generation plasma accelerators. To do so, we perform two-dimensional high-resolution MHD simulations using an argon-filled capillary discharge waveguide with three different electron transport coefficients models. The models tested include (i) Davies et al. (ii) Spitzer, and (iii) Epperlein-Haines (EH). We found that the EH model overestimates the electron temperature inside the channel by over $20\%$ while predicting a lower azimuthal magnetic field. Moreover, the Spitzer model, often used in MHD simulations for plasma-based accelerators, predicts a significantly higher electron temperature than the other models suggest.
△ Less
Submitted 17 March, 2022;
originally announced March 2022.
-
Pairs of foliations and Mattei-Moussu's theorem
Authors:
Adjaratou Arame Diaw,
Frank Loray
Abstract:
We prove a reduction of singularities for pairs of foliations by blowing-up, and then investigate the analytic classification of the reduced models. Those reduced pairs of regular foliations are well understood. The case of a regular and a singular foliation is dealt with Mattei-Moussu's Theorem for which we provide a new proof, avoiding Gronwall's inequality. We end-up announcing results recently…
▽ More
We prove a reduction of singularities for pairs of foliations by blowing-up, and then investigate the analytic classification of the reduced models. Those reduced pairs of regular foliations are well understood. The case of a regular and a singular foliation is dealt with Mattei-Moussu's Theorem for which we provide a new proof, avoiding Gronwall's inequality. We end-up announcing results recently obtained by the first author in the case of a pair of reduced foliations sharing the same separatrices.
△ Less
Submitted 3 August, 2020;
originally announced August 2020.
-
Designing Radiation Transport Tests: Simulation-Driven Uncertainty-Quantification of the COAX Temperature Diagnostic
Authors:
Chris L. Fryer,
Abdourahmane Diaw,
Chris J. Fontes,
Aimee L. Hungerford,
John Kline,
Heather Johns,
Nick Lanier,
Suzannah Wood,
Todd Urbatsch
Abstract:
One of the difficulties in develo** accurate numerical models of radiation flow in a coupled radiation-hydrodynamics setting is accurately modeling the transmission across a boundary layer. The COAX experiment is a platform design to test this transmission including standard radiograph and flux diagnostics as well as a temperature diagnostic measuring the population of excitation levels and ioni…
▽ More
One of the difficulties in develo** accurate numerical models of radiation flow in a coupled radiation-hydrodynamics setting is accurately modeling the transmission across a boundary layer. The COAX experiment is a platform design to test this transmission including standard radiograph and flux diagnostics as well as a temperature diagnostic measuring the population of excitation levels and ionization states of a dopant embedded within the target material. Using a broad range of simulations, we study the experimental errors in this temperature diagnostic. We conclude with proposed physics experiments that show features that are much stronger than the experimental errors and provide the means to study transport models.
△ Less
Submitted 22 December, 2019;
originally announced December 2019.
-
Ion friction at small values of the Coulomb logarithm
Authors:
Tucker Sprenkle,
Adam Dodson,
Quinton McKnight,
Ross Spencer,
Scott Bergeson,
Abdourahmane Diaw,
Michael S. Murillo
Abstract:
Transport properties of high-energy-density plasmas are influenced by the ion collision rate. Traditionally, this rate involves the Coulomb logarithm, $\lnΛ$. Typical values of $\lnΛ$ are $\approx 10~\mbox{to}~20$ in kinetic theories where transport properties are dominated by weak-scattering events caused by long-range forces. The validity of these theories breaks down for strongly-coupled plasma…
▽ More
Transport properties of high-energy-density plasmas are influenced by the ion collision rate. Traditionally, this rate involves the Coulomb logarithm, $\lnΛ$. Typical values of $\lnΛ$ are $\approx 10~\mbox{to}~20$ in kinetic theories where transport properties are dominated by weak-scattering events caused by long-range forces. The validity of these theories breaks down for strongly-coupled plasmas, when $\lnΛ$ is of order one. We present measurements and simulations of collision data in strongly-coupled plasmas when $\lnΛ$ is small. Experiments are carried out in the first dual-species ultracold neutral plasma (UNP), using Ca$^+$ and Yb$^+$ ions. We find strong collisional coupling between the different ion species in the bulk of the plasma. We simulate the plasma using a two-species fluid code that includes Coulomb logarithms derived from either a screened Coulomb potential or a the potential of mean force. We find generally good agreement between the experimental measurements and the simulations. With some improvements, the mixed Ca$^+$ and Yb$^+$ dual-species UNP will be a promising platform for testing theoretical expressions for $\lnΛ$ and collision cross-sections from kinetic theories through measurements of energy relaxation, stop** power, two-stream instabilities, and the evolution of sculpted distribution functions in an idealized environment in which the initial temperatures, densities, and charge states are accurately known.
△ Less
Submitted 21 May, 2019;
originally announced May 2019.
-
Geometry of stable ruled surface over an elliptic curve
Authors:
Adjaratou Arame Diaw
Abstract:
We consider the stable ruled surface $S_1$ over an elliptic curve. There is a unique foliation on $S_1$ transverse to the fibration. The minimal self-intersection sections also define a 2-web. We prove that the 4-web defined by the fibration, the foliation and the 2-web is locally parallelizable.
We consider the stable ruled surface $S_1$ over an elliptic curve. There is a unique foliation on $S_1$ transverse to the fibration. The minimal self-intersection sections also define a 2-web. We prove that the 4-web defined by the fibration, the foliation and the 2-web is locally parallelizable.
△ Less
Submitted 1 March, 2019;
originally announced March 2019.
-
Energy dispersion in radiation pressure accelerated ion beams
Authors:
M. Grech,
S. Skupin,
A. Diaw,
T. Schlegel,
V. T. Tikhonchuk
Abstract:
We address the problem of energy dispersion of radiation pressure accelerated (RPA) ion beams emerging from a thin (solid) target. Two different acceleration schemes, namely phase-stable acceleration and multi-stage acceleration, are considered by means of analytical modelling and one-dimensional particle-in-cell simulations. Our investigations offer a deeper understanding of RPA and allow us to d…
▽ More
We address the problem of energy dispersion of radiation pressure accelerated (RPA) ion beams emerging from a thin (solid) target. Two different acceleration schemes, namely phase-stable acceleration and multi-stage acceleration, are considered by means of analytical modelling and one-dimensional particle-in-cell simulations. Our investigations offer a deeper understanding of RPA and allow us to derive some guidelines for generating monoenergetic ion beams.
△ Less
Submitted 24 May, 2011;
originally announced May 2011.
