-
Epistemic and Aleatoric Uncertainty Quantification and Surrogate Modelling in High-Performance Multiscale Plasma Physics Simulations
Authors:
Yehor Yudin,
David Coster,
Udo von Toussaint,
Frank Jenko
Abstract:
This work suggests several methods of uncertainty treatment in multiscale modelling and describes their application to a system of coupled turbulent transport simulations of a tokamak plasma. We propose a method to quantify the usually aleatoric uncertainty of a system in a quasi-stationary state, estimating the mean values and their errors for quantities of interest, which is average heat fluxes…
▽ More
This work suggests several methods of uncertainty treatment in multiscale modelling and describes their application to a system of coupled turbulent transport simulations of a tokamak plasma. We propose a method to quantify the usually aleatoric uncertainty of a system in a quasi-stationary state, estimating the mean values and their errors for quantities of interest, which is average heat fluxes in the case of turbulence simulations. The method defines the stationarity of the system and suggests a way to balance the computational cost of simulation and the accuracy of estimation. This allows, contrary to many approaches, to incorporate aleatoric uncertainties in the analysis of the model and to have a quantifiable decision for simulation runtime. Furthermore, the paper describes methods for quantifying the epistemic uncertainty of a model and the results of such a procedure for turbulence simulations, identifying the model's sensitivity to particular input parameters and sensitivity to uncertainties in total. Finally, we introduce a surrogate model approach based on Gaussian Process Regression and present a preliminary result of training and analysing the performance of such a model based on turbulence simulation data. Such an approach shows a potential to significantly decrease the computational cost of the uncertainty propagation for the given model, making it feasible on current HPC systems.
△ Less
Submitted 13 June, 2023;
originally announced June 2023.
-
Bias and synergy in the self-consistent approach of data analysis of ion beam techniques
Authors:
Tiago F. Silva,
Cleber L. Rodrigues,
Manfredo H. Tabacniks,
Udo von Toussaint,
Matej Mayer
Abstract:
Using multiple ion beam analysis measurements, or techniques, combined with self-consistent data processing, generally allows extracting more (or more accurate) information from the measurements than processing separately data from single measurements. Solving ambiguities, improving the final depth resolution, defining constraints and extending applicability are the main strengths of the data-fusi…
▽ More
Using multiple ion beam analysis measurements, or techniques, combined with self-consistent data processing, generally allows extracting more (or more accurate) information from the measurements than processing separately data from single measurements. Solving ambiguities, improving the final depth resolution, defining constraints and extending applicability are the main strengths of the data-fusion approach. It basically consists in formulating a multi-objective minimization problem that can be tackled by the adoption of the weighted-sum method. A simulation study is reported in order to evaluate the systematic error inserted in the analysis by the choice of a specific objective function, or even by the weights or normalization adopted in the weighted-sum method. We demonstrate that the bias of the analyzed objective functions asymptotically converges to the true value for better statistics. We also demonstrate that the joint analysis inherits the accuracy of the most accurate measurement, establishing a rank of information content, where some combinations of measurements are more valuable than others, i.e. when processed together they provide more information by means of a better constraint for the multi-objective optimization.
△ Less
Submitted 14 October, 2022; v1 submitted 25 November, 2021;
originally announced November 2021.
-
Self-consistent ion beam analysis: an approach by multi-objective optimization
Authors:
Tiago F. Silva,
Cleber L. Rodrigues,
Nemitala Added,
Marcia A. Rizzutto,
Manfredo H. Tabacniks,
Till Höschen,
Udo von Toussaint,
Matej Mayer
Abstract:
Ion Beam Analysis (IBA) comprises a set of analytical techniques suited for material analysis, many of which are rather closely related. Self-consistent analysis of several IBA techniques takes advantage of this close relationship to combine different Ion Beam measurements in a unique model to obtain an improved characterization of the sample. This approach provides a powerful tool to obtain an un…
▽ More
Ion Beam Analysis (IBA) comprises a set of analytical techniques suited for material analysis, many of which are rather closely related. Self-consistent analysis of several IBA techniques takes advantage of this close relationship to combine different Ion Beam measurements in a unique model to obtain an improved characterization of the sample. This approach provides a powerful tool to obtain an unequivocal and reliable model of the sample, increasing confidence and reducing ambiguities. Taking advantage of the recognized reliability and quality of the simulations provided by SIMNRA, we developed a multi-process program for a self-consistent analysis based on SIMNRA calculations. MultiSIMNRA uses computational algorithms to minimize an objective function running multiple instances of SIMNRA. With four different optimization algorithms, the code can handle sample and setup parameters (including correlations and constraints), to find the set of parameters that best fits simultaneously all experimental data.
△ Less
Submitted 30 August, 2021; v1 submitted 29 July, 2021;
originally announced August 2021.
-
Computational study of crystal defects formation in Mo by machine learned molecular dynamics simulations
Authors:
F. J. Dominguez-Gutierrez,
J. Byggmästar,
K. Nordlund,
F. Djurabekova,
U. von Toussaint
Abstract:
In this work, we study the damage in crystalline molybdenum material samples due to neutron bombardment in a primary knock-on atom range of 0.5-10 keV at room temperature. We perform machine learned molecular dynamics (MD) simulations with a previously developed interatomic potential based on the Gaussian Approximation Potential (GAP) framework. We utilize a recently developed software workflow fo…
▽ More
In this work, we study the damage in crystalline molybdenum material samples due to neutron bombardment in a primary knock-on atom range of 0.5-10 keV at room temperature. We perform machine learned molecular dynamics (MD) simulations with a previously developed interatomic potential based on the Gaussian Approximation Potential (GAP) framework. We utilize a recently developed software workflow for fingerprinting and visualizing defects in damage crystal structures to analyze the damaged Mo samples by computing the formation of point defects during and after a collision cascade. As a benchmark, we report results for the total number of Frenkel pairs (a self-interstitial atom and a single vacancy) formed and atom displacement as a function of the PKA energy. A comparison to results obtained by using an Embedded Atom Method (EAM) potential is presented to discuss the advantages and limits of the machine learned MD simulations. The formation of Frenkel pairs follows a sublinear scaling law related to the PKA energy with $E^{0.54}_\mathrm{PKA}$ to the GAP MD results and $E^{0.667}_\mathrm{PKA}$ for the EAM simulations. Although the average number total defects is similar for both methods, we notice that MD potentials model different atomic geometries for the complex point defects, where the formation of crowdions is more favorable for the GAP potential. Finally, ion beam mixing results for GAP MD simulations are reported and discussed.
△ Less
Submitted 5 October, 2020;
originally announced October 2020.
-
Symplectic Gaussian Process Regression of Hamiltonian Flow Maps
Authors:
Katharina Rath,
Christopher G. Albert,
Bernd Bischl,
Udo von Toussaint
Abstract:
We present an approach to construct appropriate and efficient emulators for Hamiltonian flow maps. Intended future applications are long-term tracing of fast charged particles in accelerators and magnetic plasma confinement configurations. The method is based on multi-output Gaussian process regression on scattered training data. To obtain long-term stability the symplectic property is enforced vi…
▽ More
We present an approach to construct appropriate and efficient emulators for Hamiltonian flow maps. Intended future applications are long-term tracing of fast charged particles in accelerators and magnetic plasma confinement configurations. The method is based on multi-output Gaussian process regression on scattered training data. To obtain long-term stability the symplectic property is enforced via the choice of the matrix-valued covariance function. Based on earlier work on spline interpolation we observe derivatives of the generating function of a canonical transformation. A product kernel produces an accurate implicit method, whereas a sum kernel results in a fast explicit method from this approach. Both correspond to a symplectic Euler method in terms of numerical integration. These methods are applied to the pendulum and the Hénon-Heiles system and results compared to an symmetric regression with orthogonal polynomials. In the limit of small map** times, the Hamiltonian function can be identified with a part of the generating function and thereby learned from observed time-series data of the system's evolution. Besides comparable performance of implicit kernel and spectral regression for symplectic maps, we demonstrate a substantial increase in performance for learning the Hamiltonian function compared to existing approaches.
△ Less
Submitted 11 September, 2020;
originally announced September 2020.
-
FaVAD: A software workflow for characterisation and visualizing of defects in crystalline structures
Authors:
U. von Toussaint,
F. J. Dominguez-Gutierrez,
M. Compostella,
M. Rampp
Abstract:
The analysis of defects and defect dynamics in crystalline materials is important for fundamental science and for a wide range of applied engineering. With increasing system size the analysis of molecular-dynamics simulation data becomes non-trivial. Here, we present a workflow for semi-automatic identification and classification of defects in crystalline structures, combining a new approach for d…
▽ More
The analysis of defects and defect dynamics in crystalline materials is important for fundamental science and for a wide range of applied engineering. With increasing system size the analysis of molecular-dynamics simulation data becomes non-trivial. Here, we present a workflow for semi-automatic identification and classification of defects in crystalline structures, combining a new approach for defect description with several already existing open-source software packages. Our approach addresses the key challenges posed by the often relatively tiny volume fraction of the modified parts of the sample, thermal motion and the presence of potentially unforeseen atomic configurations (defect types) after irradiation. The local environment of any atom is converted into a rotation-invariant descriptive vector ('fingerprint'), which can be compared to known defect types and also yields a distance metric suited for classification. Vectors which cannot be associated to known structures indicate new types of defects. As proof-of-concept we apply our method on an iron sample to analyze the defects caused by a collision cascade induced by a 10 keV primary-knock-on-atom. The obtained results are in good agreement with reported literature values.
△ Less
Submitted 17 April, 2020;
originally announced April 2020.
-
Formation of C$_{18}$H and C$_{18}$H$_2$ molecules by low energy irradiation with atomic and molecular hydrogen
Authors:
F. J. Domínguez-Gutiérrez,
C. Martínez-Flores,
P. S. Krstic,
R. Cabrera-Trujillo,
U. von Toussaint
Abstract:
We study the formation of C$_{18}$H and C$_{18}$H$_2$ by irradiating a cyclo[$18$]carbon molecule with atomic and molecular hydrogen at impact energy, $E$, in the range of 0.5-25 eV. We utilize the density-functional tight-binding method to perform molecular dynamics simulations to emulate the interaction of a carbon ring when colliding with atomic or molecular hydrogen. From our results, the form…
▽ More
We study the formation of C$_{18}$H and C$_{18}$H$_2$ by irradiating a cyclo[$18$]carbon molecule with atomic and molecular hydrogen at impact energy, $E$, in the range of 0.5-25 eV. We utilize the density-functional tight-binding method to perform molecular dynamics simulations to emulate the interaction of a carbon ring when colliding with atomic or molecular hydrogen. From our results, the formation of the C$_{18}$H molecules is likely to occur upon irradiating by H atoms at $E < 10$ eV and by H$_2$ molecules at $2 < E < 15$ eV center of mass energy. Formation of C$_{18}$H$_2$ molecules is only observed at around $E = 2$ eV. Our results show that the absorption of hydrogen is more prone in atomic than in molecular hydrogen atmosphere. Thus, we find that the probability of physio-absorption reaches up to 80 \% for atomic projectiles with $E < 5$ eV but only up to 10 \% for the molecular ones. Our analysis shows that the deformation of the carbon ring due to the hydrogen bonding produces transition from $sp$ to $sp^2$ hybridization. The angle between the carbon atoms at the locations near to the H bond in the resulting ring is not 120$^o$ but instead 110$^o$ degrees. No molecular fragmentation of the C$_{18}$ ring is observed.
△ Less
Submitted 7 April, 2020;
originally announced April 2020.
-
On the classification and quantification of crystal defects after energetic bombardment by machine learned molecular dynamics simulations
Authors:
F. J. Domínguez-Gutiérrez,
J Byggmästar,
K. Nordlund,
F. Djurabekova,
U von Toussaint
Abstract:
The analysis of the damage on plasma facing materials (PFM), due to its direct interaction with the plasma environment, is needed to build the next generation of nuclear machines, where tungsten has been proposed as a candidate. In this work, we perform molecular dynamics (MD) simulations using a machine learned inter-atomic potential, based on the Gaussian Approximation Potential framework, to mo…
▽ More
The analysis of the damage on plasma facing materials (PFM), due to its direct interaction with the plasma environment, is needed to build the next generation of nuclear machines, where tungsten has been proposed as a candidate. In this work, we perform molecular dynamics (MD) simulations using a machine learned inter-atomic potential, based on the Gaussian Approximation Potential framework, to model better neutron bombardment mechanisms in pristine W lattices. The MD potential is trained to reproduce realistic short-range dynamics, the liquid phase, and the material recrystallization, which are important for collision cascades. The formation of point defects is quantified and classified by a descriptor vector (DV) based method, which is independent of the sample temperature and its constituents, requiring only modest computational resources. The locations of vacancies are calculated by the k-d-tree algorithm. The analysis of the damage in the W samples is compared to results obtained by EAM Finnis-Sinclair and Tersoff-ZBL potentials, at a sample temperature of 300 K and a primary knock-on atom (PKA) energy range of 0.5-10 keV, where a good agreement with the reported number of Frenkel pair is observed. Our results provide information about the advantages and limits of the machine learned MD simulations with respect to the standard ones. The formation of dumbbell and crowdion defects as a function of PKA is discussed.
△ Less
Submitted 26 October, 2019;
originally announced October 2019.
-
On the detection and classification of material defects in crystalline solids after energetic particle impact simulations
Authors:
F. J. Dominguez-Gutierrez,
U. von Toussaint
Abstract:
We present a fingerprint-like method to analyze material defects after energetic particle irradiation by computing a rotation invariant descriptor vector for each atom of a given sample. For ordered solids this new method is easy to use, does not require extreme computational resources, and is largely independent of the sample material and sample temperature. As illustration we applied the method…
▽ More
We present a fingerprint-like method to analyze material defects after energetic particle irradiation by computing a rotation invariant descriptor vector for each atom of a given sample. For ordered solids this new method is easy to use, does not require extreme computational resources, and is largely independent of the sample material and sample temperature. As illustration we applied the method to molecular dynamics simulations of deuterated and pristine tungsten lattices at 300 K using a primary knock-on atom (PKA) of 1 keV with different velocity directions to emulate a neutron bombardment process. The number of W atoms, that are affected after the collision cascade, have been quantified with the presented approach. At first atoms at regular lattice positions as well as common defect types like interstitials and vacancies have been identified using precomputed descriptor vectors. A principal component analysis (PCA) is used to identify previously overlooked defect types and to derive the corresponding local atomic structure. A comparison of the irradiation effects for deuterated and pristine tungsten samples revealed that deuterated samples exhibit consistently more defects than pristine ones.
△ Less
Submitted 2 September, 2019;
originally announced September 2019.
-
Optimizing Nuclear Reaction Analysis (NRA) using Bayesian Experimental Design
Authors:
U. von Toussaint,
T. Schwarz-Selinger,
S. Gori
Abstract:
Nuclear Reaction Analysis with ${}^{3}$He holds the promise to measure Deuterium depth profiles up to large depths. However, the extraction of the depth profile from the measured data is an ill-posed inversion problem. Here we demonstrate how Bayesian Experimental Design can be used to optimize the number of measurements as well as the measurement energies to maximize the information gain. Compa…
▽ More
Nuclear Reaction Analysis with ${}^{3}$He holds the promise to measure Deuterium depth profiles up to large depths. However, the extraction of the depth profile from the measured data is an ill-posed inversion problem. Here we demonstrate how Bayesian Experimental Design can be used to optimize the number of measurements as well as the measurement energies to maximize the information gain. Comparison of the inversion properties of the optimized design with standard settings reveals huge possible gains. Application of the posterior sampling method allows to optimize the experimental settings interactively during the measurement process.
△ Less
Submitted 19 December, 2008;
originally announced December 2008.
