Showing 1–5 of 5 results for author: Bourasseau, E

Targeting the partition function of chemically disordered materials with a generative approach based on inverse variational autoencoders

Authors: Maciej J. Karcz, Luca Messina, Eiji Kawasaki, Emeric Bourasseau

Abstract: Computing atomic-scale properties of chemically disordered materials requires an efficient exploration of their vast configuration space. Traditional approaches such as Monte Carlo or Special Quasirandom Structures either entail sampling an excessive amount of configurations or do not ensure that the configuration space has been properly covered. In this work, we propose a novel approach where gen… ▽ More Computing atomic-scale properties of chemically disordered materials requires an efficient exploration of their vast configuration space. Traditional approaches such as Monte Carlo or Special Quasirandom Structures either entail sampling an excessive amount of configurations or do not ensure that the configuration space has been properly covered. In this work, we propose a novel approach where generative machine learning is used to yield a representative set of configurations for accurate property evaluation and provide accurate estimations of atomic-scale properties with minimal computational cost. Our method employs a specific type of variational autoencoder with inverse roles for the encoder and decoder, enabling the application of an unsupervised active learning scheme that does not require any initial training database. The model iteratively generates configuration batches, whose properties are computed with conventional atomic-scale methods. These results are then fed back into the model to estimate the partition function, repeating the process until convergence. We illustrate our approach by computing point-defect formation energies and concentrations in (U, Pu)O2 mixed-oxide fuels. In addition, the ML model provides valuable insights into the physical factors influencing the target property. Our method is generally applicable to explore other properties, such as atomic-scale diffusion coefficients, in ideally or non-ideally disordered materials like high-entropy alloys. △ Less

Submitted 10 September, 2024; v1 submitted 27 August, 2024; originally announced August 2024.

Semi-supervised generative approach to point-defect formation in chemically disordered compounds: application to uranium-plutonium mixed oxides

Authors: Maciej J. Karcz, Luca Messina, Eiji Kawasaki, Serenah Rajaonson, Didier Bathellier, Emeric Bourasseau

Abstract: Machine-learning methods are nowadays of common use in the field of material science. For example, they can aid in optimizing the physicochemical properties of new materials, or help in the characterization of highly complex chemical compounds. An especially challenging problem arises in the modeling of chemically disordered solid solutions, for which some properties depend on the distribution of… ▽ More Machine-learning methods are nowadays of common use in the field of material science. For example, they can aid in optimizing the physicochemical properties of new materials, or help in the characterization of highly complex chemical compounds. An especially challenging problem arises in the modeling of chemically disordered solid solutions, for which some properties depend on the distribution of chemical species in the crystal lattice. This is the case of defect properties of uranium-plutonium mixed oxides nuclear fuels. The number of possible configurations is so high that the problem becomes intractable if treated with direct sampling. We thus propose a machine learning approach, based on generative modeling, to optimize the exploration of this large configuration space. A probabilistic, semi-supervised approach using Mixture Density Network is applied to estimate the concentration of thermal defects in (U, Pu)O2. We show that this method, based on the prediction of the density of states of formation energy of a defect, is computationally much more cost-efficient compared to other approaches available in the literature. △ Less

Submitted 28 November, 2022; v1 submitted 22 November, 2022; originally announced November 2022.

Comments: The performed calculations require re-verification

Effect of cationic chemical disorder on defect formation energies in uranium-plutonium mixed oxides

Authors: Didier Bathellier, Luca Messina, Michel Freyss, Marjorie Bertolus, Thomas Schuler, Maylise Nastar, Pär Olsson, Emeric Bourasseau

Abstract: At the atomic scale, uranium-plutonium mixed oxides (U,Pu)O_2 are characterized by cationic chemical disorder, which entails that U and Pu cations are randomly distributed on the cation sublattice. In the present work, we study the impact of disorder on point-defect formation energies in (U,Pu)O_2 using interatomic-potential and Density Functional Theory (DFT+U) calculations. We focus on bound Sch… ▽ More At the atomic scale, uranium-plutonium mixed oxides (U,Pu)O_2 are characterized by cationic chemical disorder, which entails that U and Pu cations are randomly distributed on the cation sublattice. In the present work, we study the impact of disorder on point-defect formation energies in (U,Pu)O_2 using interatomic-potential and Density Functional Theory (DFT+U) calculations. We focus on bound Schottky defects (BSD) that are among the most stable defects in these oxides. As a first step, we estimate the distance R_D around the BSD up to which the local chemical environment significantly affects their formation energy. To this end, we propose an original procedure in which the formation energy is computed for several supercells at varying levels of disorder. We conclude that the first three cation shells around the BSD have a non-negligible influence on their formation energy (R_{D} = 7.0 \{AA}). We apply then a systematic approach to compute the BSD formation energies for all the possible cation configurations on the first and second nearest neighbor shells around the BSD. We show that the formation energy can range in an interval of 0.97 eV, depending on the relative amount of U and Pu neighboring cations. Based on these results, we propose an interaction model that describes the effect of nominal and local composition on the BSD formation energy. Finally, the DFT+U benchmark calculations show a satisfactory agreement for configurations characterized by a U-rich local environment, and a larger mismatch in the case of a Pu-rich one. In summary, this work provides valuable insights on the properties of BSD defects in (U,Pu)O_2, and can represent a valid strategy to study point defect properties in disordered compounds. △ Less

Submitted 27 June, 2022; originally announced June 2022.

Comments: 33 pages, 20 figures

Microscopic calculations of Hugoniot curves of neat TATB and of its detonation products

Authors: Emeric Bourasseau, Jean-Bernard Maillet, Nicolas Desbiens, Gabriel Stoltz

Abstract: We compute the Hugoniot curves of both neat TATB and its detonation products mixture using atomistic simulation tools. To compute the Hugoniot states, we adapted our "Sampling Constraints in Average" (SCA) method (Maillet et al., Applied Math. Research eXpress 2008, 2009) to Monte-Carlo simulations. For neat TATB, we show that the potential proposed by Rai (Rai et al., J. Chem. Phys. 129, 2008) is… ▽ More We compute the Hugoniot curves of both neat TATB and its detonation products mixture using atomistic simulation tools. To compute the Hugoniot states, we adapted our "Sampling Constraints in Average" (SCA) method (Maillet et al., Applied Math. Research eXpress 2008, 2009) to Monte-Carlo simulations. For neat TATB, we show that the potential proposed by Rai (Rai et al., J. Chem. Phys. 129, 2008) is not accurate enough to predict the Hugoniot curve and requires some optimization of its parameters. Concerning detonation products, thermodynamic properties at chemical equilibrium are computed using a specific RxMC method (Bourasseau et al., Phys. Chem. Chem. Phys. 13, 2011) taking into account the presence of carbon clusters in the fluid mixture. We show that this explicit description of the solid phase immersed in the fluid phase modifies the chemical equilibrium. △ Less

Submitted 23 May, 2011; originally announced May 2011.

Constant entropy sampling and release waves of shock compressions

Authors: Jean-Bernard Maillet, Emeric Bourasseau, Laurent Soulard, Jean Clerouin, Gabriel Stoltz

Abstract: We present several equilibrium methods that allow to compute isentropic processes, either during the compression or the release of the material. These methods are applied to compute the isentropic release of a shocked monoatomic liquid at high pressure and temperature. Moreover, equilibrium results of isentropic release are compared to the direct nonequilibrium simulation of the same process. We… ▽ More We present several equilibrium methods that allow to compute isentropic processes, either during the compression or the release of the material. These methods are applied to compute the isentropic release of a shocked monoatomic liquid at high pressure and temperature. Moreover, equilibrium results of isentropic release are compared to the direct nonequilibrium simulation of the same process. We show that due to the viscosity of the liquid but also to nonequilibrium effects, the release of the system is not strictly isentropic. △ Less

Submitted 27 March, 2009; originally announced March 2009.