Showing 1–18 of 18 results for author: Jakse, N

Classification Tree-based Active Learning: A Wrapper Approach

Authors: Ashna Jose, Emilie Devijver, Massih-Reza Amini, Noel Jakse, Roberta Poloni

Abstract: Supervised machine learning often requires large training sets to train accurate models, yet obtaining large amounts of labeled data is not always feasible. Hence, it becomes crucial to explore active learning methods for reducing the size of training sets while maintaining high accuracy. The aim is to select the optimal subset of data for labeling from an initial unlabeled set, ensuring precise p… ▽ More Supervised machine learning often requires large training sets to train accurate models, yet obtaining large amounts of labeled data is not always feasible. Hence, it becomes crucial to explore active learning methods for reducing the size of training sets while maintaining high accuracy. The aim is to select the optimal subset of data for labeling from an initial unlabeled set, ensuring precise prediction of outcomes. However, conventional active learning approaches are comparable to classical random sampling. This paper proposes a wrapper active learning method for classification, organizing the sampling process into a tree structure, that improves state-of-the-art algorithms. A classification tree constructed on an initial set of labeled samples is considered to decompose the space into low-entropy regions. Input-space based criteria are used thereafter to sub-sample from these regions, the total number of points to be labeled being decomposed into each region. This adaptation proves to be a significant enhancement over existing active learning methods. Through experiments conducted on various benchmark data sets, the paper demonstrates the efficacy of the proposed framework by being effective in constructing accurate classification models, even when provided with a severely restricted labeled data set. △ Less

Submitted 15 April, 2024; originally announced April 2024.

Feature Selection for High-Dimensional Neural Network Potentials with the Adaptive Group Lasso

Authors: Johannes Sandberg, Thomas Voigtmann, Emilie Devijver, Noel Jakse

Abstract: Neural network potentials are a powerful tool for atomistic simulations, allowing to accurately reproduce \textit{ab initio} potential energy surfaces with computational performance approaching classical force fields. A central component of such potentials is the transformation of atomic positions into a set of atomic features in a most efficient and informative way.In this work, a feature selecti… ▽ More Neural network potentials are a powerful tool for atomistic simulations, allowing to accurately reproduce \textit{ab initio} potential energy surfaces with computational performance approaching classical force fields. A central component of such potentials is the transformation of atomic positions into a set of atomic features in a most efficient and informative way.In this work, a feature selection method is introduced for high dimensional neural network potentials, based on the Adaptive Group Lasso (AGL) approach. It is shown that the use of an embedded method, taking into account the interplay between features and their action in the estimator, is necessary to optimize the number of features. The method's efficiency is tested on three different monoatomic systems, including Lennard-Jones as a simple test case, Aluminium as a system characterized by predominantly radial interactions, and Boron as representative of a system with strongly directional interactions. The AGL is compared with unsupervised filter methods and found to perform consistently better in reducing the number of features needed to reproduce the reference simulation data. {In particular, our results show the importance of taking into account model predictions in feature selection for interatomic potentials. △ Less

Submitted 26 December, 2023; originally announced December 2023.

Comments: 30 pages, 9 figures

On the existence of a second branch of transverse collective excitations in liquid metals

Authors: J. -F. Wax, N. Jakse

Abstract: It was found recently that the liquid dynamics of several metals (Li, Zn, Ni, Fe, Tl, Pb) under pressure is characterized by transverse spectral functions containing an additional high-frequency peak. To rationalize the pressure dependence of the contributions from different propagating processes to transverse spectral functions in liquid metals, ab initio molecular dynamics simulations were perfo… ▽ More It was found recently that the liquid dynamics of several metals (Li, Zn, Ni, Fe, Tl, Pb) under pressure is characterized by transverse spectral functions containing an additional high-frequency peak. To rationalize the pressure dependence of the contributions from different propagating processes to transverse spectral functions in liquid metals, ab initio molecular dynamics simulations were performed for two typical liquid metals (Na and Al) in a wide range of pressures. The influence of density/pressure is investigated for Na by considering four pressures ranging from 15 to 147 GPa, while the temperature influence is considered for Al between 600 K in the deep supercooled liquid up to 1700 K well above the melting point at ambient pressure. Both temperature and density dependence of the spectra of collective excitations are analyzed with a focus on the appearance of a second high-frequency mode in the transverse spectra. A correspondence between spectra of transverse collective excitations and the peak positions of the Fourier-spectra of velocity autocorrelation functions (vibrational density of states) is found. △ Less

Submitted 11 August, 2023; originally announced August 2023.

Comments: 8 pages, 6 figures, 1 table

Journal ref: Condensed Matter Physics, 2023, vol. 26, No. 3, 33603

Structure-Dynamics Relationship in Al-Mg-Si Liquid Alloys

Authors: Alaa Fahs, Philippe Jarry, Noël Jakse

Abstract: Enhancing properties and performances of aluminium alloys by a control of their solidification is pivotal in automotive and aerospace industries. The fundamental role of the structure-diffusion relationship is investigated for Al-Mg-Si liquid alloys taken as a prototype of Al-6xxx. For this purpose, first principles-based molecular dynamics simulations were performed for various Si and Mg content… ▽ More Enhancing properties and performances of aluminium alloys by a control of their solidification is pivotal in automotive and aerospace industries. The fundamental role of the structure-diffusion relationship is investigated for Al-Mg-Si liquid alloys taken as a prototype of Al-6xxx. For this purpose, first principles-based molecular dynamics simulations were performed for various Si and Mg content for Al-rich compositions, including the binary alloy counterparts. Results indicate that Mg and/or Si in alloys create a more compact ordering around Al than in pure Al, lowering diffusion. Mg promotes icosahedral short-range order, while Si displays a preference towards cubic local ordering, impacting diffusion based on their respective content. It suggests a mechanism whereby an increase in Mg content generally lowers the diffusion of each species, whereas an increase in Si content enhances their diffusion, providing insights for future alloy design. △ Less

Submitted 29 June, 2023; originally announced June 2023.

Comments: 5 figues, 18 pages

An Artificial Neural Network-based Density Functional Approach for Adiabatic Energy Differences in Transition Metal Complexes

Authors: Joao Paulo Almeida de Mendonca, Lorenzo A. Mariano, Emilie Devijver, Noel Jakse, Roberta Poloni

Abstract: During the past decades, approximate Kohn-Sham density-functional theory schemes garnered many successes in computational chemistry and physics; yet the performance in the prediction of spin state energetics is often unsatisfactory. By means of a machine-learning approach, an enhanced exchange and correlation functional is developed to describe adiabatic energy differences in transition metal comp… ▽ More During the past decades, approximate Kohn-Sham density-functional theory schemes garnered many successes in computational chemistry and physics; yet the performance in the prediction of spin state energetics is often unsatisfactory. By means of a machine-learning approach, an enhanced exchange and correlation functional is developed to describe adiabatic energy differences in transition metal complexes. The functional is based on the computationally efficient revision of the regularized strongly constrained and appropriately normed (R2SCAN) functional and improved by an artificial neural-network correction trained over a small dataset of electronic densities, atomization energies and/or spin state energetics. The training process, performed using a bio-inspired non gradient-based approach adapted for this work from Particle Swarm Optimization, is discussed in detail. The meta-GGA functional is finally shown to outperform most known density functionals in the prediction of adiabatic energy differences for both the validation test and the generality test. △ Less

Submitted 4 May, 2023; v1 submitted 16 April, 2023; originally announced April 2023.

Crystal Nucleation in Al-Ni Alloys: an Unsupervised Chemical and Topological Learning Approach

Authors: Sébastien Becker, Emilie Devijver, Rémi Molinier, Noël Jakse

Abstract: Crystallization represents a fundamental process engendering solidification of a material and determines its microstructure. Driven by complex phenomena at the atomic scale, its understanding for alloys still remains elusive. The present work proposes a large scale molecular dynamics simulation study of the homogeneous crystal nucleation pathways of prototypical undercooled Al-Ni binary alloys. An… ▽ More Crystallization represents a fundamental process engendering solidification of a material and determines its microstructure. Driven by complex phenomena at the atomic scale, its understanding for alloys still remains elusive. The present work proposes a large scale molecular dynamics simulation study of the homogeneous crystal nucleation pathways of prototypical undercooled Al-Ni binary alloys. An unsupervised topological learning analysis shows that the nucleation sets in first from a chemical ordering, followed by a bond-orientational ordering of the underlying crystal phase. Our results indicate also a different polymorph selection that depends on composition. While the nucleation pathway of Al50 Ni50 displays a single step with the emergence of B2 short-range order, a step-wise nucleation toward the L12 phase is seen for Al25 Ni75 . The influence of the nucleation of pure Al and Ni counterparts is further discussed. △ Less

Submitted 4 October, 2022; originally announced October 2022.

Influence of oscillatory shear on nucleation in metallic glasses: a molecular dynamics study

Authors: Baoshuang Shang, Noel Jakse, Pengfei Guan, Weihua Wang, Jean-louis Barrat

Abstract: The process of crystal nucleation can be accelerated or retarded by ultrasonic vibration, which is particularly attractive for the addictive manufacture and thermoplastic forming of metallic glasses, however, the effect and mechanism of oscillatory loading on the nucleation process are still elusive. By using molecular dynamics simulation, the changes in the time-temperature-transformation (TTT) c… ▽ More The process of crystal nucleation can be accelerated or retarded by ultrasonic vibration, which is particularly attractive for the addictive manufacture and thermoplastic forming of metallic glasses, however, the effect and mechanism of oscillatory loading on the nucleation process are still elusive. By using molecular dynamics simulation, the changes in the time-temperature-transformation (TTT) curve under oscillatory external loading are systematically investigated in two typical binary alloys. A glass forming ability dependent response to the external loading is found, and the shortest incubation time is insensitive to the external loading, while the corresponding temperature can be significantly shifted. Within the framework of classical nucleation theory, a fitting formula is proposed to describe the simulation data quantitatively. In contrast to stationary shear, the elastic stress, rather than the strain rate, is the key parameter to control the evolution of TTT curve under oscillatory loading. Furthermore, the model shows that oscillatory loading can decouple the mobility and nucleation in the deeply supercooled liquid, hence the formation ability can be enhanced while the nucleation is suppressed, which is particularly helpful for the forming and manufacturing of metallic glasses. △ Less

Submitted 5 December, 2022; v1 submitted 20 September, 2022; originally announced September 2022.

Unsupervised topological learning approach of crystal nucleation in pure Tantalum

Authors: Sébastien Becker, Emilie Devijver, Rémi Molinier, Noël Jakse

Abstract: Nucleation phenomena commonly observed in our every day life are of fundamental, technological and societal importance in many areas, but some of their most intimate mechanisms remain however to be unraveled. Crystal nucleation, the early stages where the liquid-to-solid transition occurs upon undercooling, initiates at the atomic level on nanometer length and sub-picoseconds time scales and invol… ▽ More Nucleation phenomena commonly observed in our every day life are of fundamental, technological and societal importance in many areas, but some of their most intimate mechanisms remain however to be unraveled. Crystal nucleation, the early stages where the liquid-to-solid transition occurs upon undercooling, initiates at the atomic level on nanometer length and sub-picoseconds time scales and involves complex multidimensional mechanisms with local symmetry breaking that can hardly be observed experimentally in the very details. To reveal their structural features in simulations without a priori, an unsupervised learning approach founded on topological descriptors loaned from persistent homology concepts is proposed. Applied here to a monatomic metal, namely Tantalum (Ta), it shows that both translational and orientational ordering always come into play simultaneously when homogeneous nucleation starts in regions with low five-fold symmetry. △ Less

Submitted 8 February, 2022; originally announced February 2022.

Comments: 6 pages, 3 figures, Conference Neurip 2021. arXiv admin note: substantial text overlap with arXiv:2109.06797

Interatomic machine learning potentials for aluminium: application to solidification phenomena

Authors: Noel Jakse, Johannes Sandberg, Leon F. Granz, Anthony Saliou, Philippe Jarry, Emilie Devijver, Thomas Voigtmann, Jürgen Horbach, Andreas Meyer

Abstract: In studying solidification process by simulations on the atomic scale, the modeling of crystal nucleation or amorphisation requires the construction of interatomic interactions that are able to reproduce the properties of both the solid and the liquid states. Taking into account rare nucleation events or structural relaxation under deep undercooling conditions requires much larger length scales an… ▽ More In studying solidification process by simulations on the atomic scale, the modeling of crystal nucleation or amorphisation requires the construction of interatomic interactions that are able to reproduce the properties of both the solid and the liquid states. Taking into account rare nucleation events or structural relaxation under deep undercooling conditions requires much larger length scales and longer time scales than those achievable by \textit{ab initio} molecular dynamics (AIMD). This problem is addressed by means of classical MD simulations using a well established high dimensional neural network potential trained on a relevant set of configurations generated by AIMD. Our dataset contains various crystalline structures and liquid states at different pressures, including their time fluctuations in a wide range of temperatures considering only their energy labels. Applied to elemental aluminium, the resulting potential is shown to be efficient to reproduce the basic structural, dynamics and thermodynamic quantities in the liquid and undercooled states without the need to include neither explicitly the forces nor all kind of configurations in the training procedure. The early stage of crystallization is further investigated on a much larger scale with one million atoms, allowing us to unravel features of the homogeneous nucleation mechanisms in the fcc phase at ambient pressure as well as in the bcc phase at high pressure with unprecedented accuracy close to the \textit{ab initio} one. In both case, a single step nucleation process is observed. △ Less

Submitted 5 August, 2022; v1 submitted 4 January, 2022; originally announced January 2022.

Comments: 40 pages and 10 figures + supplement file

Unsupervised topological learning for identification of atomic structures

Authors: Sébastien Becker, Emilie Devijver, Rémi Molinier, Noël Jakse

Abstract: We propose an unsupervised learning methodology with descriptors based on Topological Data Analysis (TDA) concepts to describe the local structural properties of materials at the atomic scale. Based only on atomic positions and without a priori knowledge, our method allows for an autonomous identification of clusters of atomic structures through a Gaussian mixture model. We apply successfully this… ▽ More We propose an unsupervised learning methodology with descriptors based on Topological Data Analysis (TDA) concepts to describe the local structural properties of materials at the atomic scale. Based only on atomic positions and without a priori knowledge, our method allows for an autonomous identification of clusters of atomic structures through a Gaussian mixture model. We apply successfully this approach to the analysis of elemental Zr in the crystalline and liquid states as well as homogeneous nucleation events under deep undercooling conditions. This opens the way to deeper and autonomous study of complex phenomena in materials at the atomic scale. △ Less

Submitted 9 December, 2021; v1 submitted 16 September, 2021; originally announced September 2021.

Unsupervised topological learning approach of crystal nucleation

Authors: Sébastien Becker, Emilie Devijver, Rémi Molinier, Noël Jakse

Abstract: Nucleation phenomena commonly observed in our every day life are of fundamental, technological and societal importance in many areas, but some of their most intimate mechanisms remain however to be unravelled. Crystal nucleation, the early stages where the liquid-to-solid transition occurs upon undercooling, initiates at the atomic level on nanometre length and sub-picoseconds time scales and invo… ▽ More Nucleation phenomena commonly observed in our every day life are of fundamental, technological and societal importance in many areas, but some of their most intimate mechanisms remain however to be unravelled. Crystal nucleation, the early stages where the liquid-to-solid transition occurs upon undercooling, initiates at the atomic level on nanometre length and sub-picoseconds time scales and involves complex multidimensional mechanisms with local symmetry breaking that can hardly be observed experimentally in the very details. To reveal their structural features in simulations without a priori, an unsupervised learning approach founded on topological descriptors loaned from persistent homology concepts is proposed. Applied here to monatomic metals, it shows that both translational and orientational ordering always come into play simultaneously when homogeneous nucleation starts in regions with low five-fold symmetry. It also reveals the specificity of the nucleation pathways depending on the element considered, with features beyond the hypothesis of Classical Nucleation Theory. △ Less

Submitted 16 September, 2021; v1 submitted 14 September, 2021; originally announced September 2021.

On the Excess Entropy Scaling Law: a Potential Energy Landscape View

Authors: Anthony Saliou, Philippe Jarry, Noel Jakse

Abstract: The relationship between excess entropy and diffusion is revisited by means of large-scale computer simulation combined to supervised learning approach to determine the excess entropy for the Lennard-Jones potential. Results reveal that it finds its roots in the properties of the potential energy landscape (PEL). In particular the exponential law holding in the liquid is seen to be correlated with… ▽ More The relationship between excess entropy and diffusion is revisited by means of large-scale computer simulation combined to supervised learning approach to determine the excess entropy for the Lennard-Jones potential. Results reveal that it finds its roots in the properties of the potential energy landscape (PEL). In particular the exponential law holding in the liquid is seen to be correlated with the landscape-influenced regime of the PEL while the fluid-like power-law corresponds to the free diffusion regime. △ Less

Submitted 1 June, 2021; v1 submitted 27 May, 2021; originally announced May 2021.

Comment on "Universal effect of excitation dispersion on the heat capacity and gapped states in fluids"

Authors: T. Bryk, N. Jakse, I. Mryglod, G. Ruocco, J. -F. Wax

Abstract: We discuss the validity of recent results in [Phys. Rev. Lett. 125, 125501 (2020)] on an universal relation between the heat capacity and dispersions of collective excitations in liquids. We discuss the validity of recent results in [Phys. Rev. Lett. 125, 125501 (2020)] on an universal relation between the heat capacity and dispersions of collective excitations in liquids. △ Less

Submitted 13 April, 2021; originally announced April 2021.

Comments: 7 pages, 4 figures

Fast dynamics perspective on the breakdown of the Stokes-Einstein law in fragile glassformers

Authors: F. Puosi, A. Pasturel, N. Jakse, D. Leporini

Abstract: The breakdown of the Stokes-Einstein (SE) law in fragile glassformers is examined by Molecular-Dynamics simulations of atomic liquids and polymers and consideration of the experimental data concerning the archetypical OTP glassformer. All the four systems comply with the universal scaling between the viscosity (or the structural relaxation) and the Debye-Waller factor $\langle u^2\rangle$, the mea… ▽ More The breakdown of the Stokes-Einstein (SE) law in fragile glassformers is examined by Molecular-Dynamics simulations of atomic liquids and polymers and consideration of the experimental data concerning the archetypical OTP glassformer. All the four systems comply with the universal scaling between the viscosity (or the structural relaxation) and the Debye-Waller factor $\langle u^2\rangle$, the mean square amplitude of the particle rattling in the cage formed by the surrounding neighbors. It is found that the SE breakdown is scaled in a master curve by a reduced $\langle u^2\rangle$. Two approximated expressions of the latter, with no and one adjustable parameters respectively, are derived. △ Less

Submitted 27 March, 2018; v1 submitted 14 February, 2018; originally announced February 2018.

Comments: accepted for publication in J Chem Phys

Dynamical, structural and chemical heterogeneities in a binary metallic glass-forming liquid

Authors: F. Puosi, N. Jakse, A. Pasturel

Abstract: As approaching the glass transition, particle motion in liquids becomes highly heterogeneous and regions with virtually no mobility coexist with liquid-like domains. This complex dynamics is believed to be responsible for different phenomena including non-exponential relaxation and the breakdown of Stokes-Einstein relation. Understanding the relationships between dynamical heterogeneities and loca… ▽ More As approaching the glass transition, particle motion in liquids becomes highly heterogeneous and regions with virtually no mobility coexist with liquid-like domains. This complex dynamics is believed to be responsible for different phenomena including non-exponential relaxation and the breakdown of Stokes-Einstein relation. Understanding the relationships between dynamical heterogeneities and local structure in metallic liquids and glasses is a major scientific challenge. Here we use classical molecular dynamics simulations to study the atomic dynamics and microscopic structure of $\mbox{Cu}_{50}\mbox{Zr}_{50}$ alloy in the supercooling regime. Dynamical heterogeneities are identified via an isoconfigurational analysis. As deeper supercooling is achieved a transition from isolated to clustering low mobility atoms is reported. These slow clusters, whose size grow upon cooling, are also associated to concentration fluctuations, characterized by a Zr-enriched phase, with a composition $\mbox{CuZr}_2$. In addition, a structural analysis of slow clusters based on Voronoi tessellation evidences an increase with respect of the bulk system of the fraction of Cu atoms having a local icosahedral order. These results are in agreement with the consolidated scenario of the relevant role played by icosahedral order in the dynamic slowing-down in supercooled metal alloys. △ Less

Submitted 7 November, 2017; originally announced November 2017.

Journal ref: J. Phys.: Condens. Matter 30 145701 (2018)

Stokes-Einstein relation and excess entropy scaling law in liquid Copper

Authors: N. Jakse, A. Pasturel

Abstract: We report an ab initio study of structural and dynamic properties of liquid copper as a function of temperature. In particular, we have evaluated the temperature dependence of the self-diffusion coefficient from the velocity autocorrelation function as well the temperature dependence of the viscosity from the transverse current correlation function. We show that LDA based results are in close agre… ▽ More We report an ab initio study of structural and dynamic properties of liquid copper as a function of temperature. In particular, we have evaluated the temperature dependence of the self-diffusion coefficient from the velocity autocorrelation function as well the temperature dependence of the viscosity from the transverse current correlation function. We show that LDA based results are in close agreement with experimental data for both the self-diffusion coefficient and the viscosity over the temperature range investigated. Our findings are then used to test empirical approaches like the Stokes-Einstein relation and the excess entropy scaling law widely used in the literature. We show that the Stokes-Einstein relation is valid for the liquid phase and that the excess entropy scaling law proposed by Dzugutov is legitimate only if a self-consistent method for determining the packing fraction of the hard sphere reference liquid is used within the Carnahan-Starling approach to express the excess entropy. △ Less

Submitted 24 December, 2015; originally announced December 2015.

Comments: 11 pages, 6 figures

Journal ref: Condens. Matter Phys., 2015, vol. 18, No. 4, 43603

Growth of graphene on 6H-SiC by molecular dynamics simulation

Authors: N. Jakse, R. Arifin, S. K. Lai

Abstract: Classical molecular-dynamics simulations were carried out to study epitaxial growth of graphene on 6H-SiC(0001) substrate. It was found that there exists a threshold annealing temperature above which we observe formation of graphitic structure on the substrate. To check the sensitivity of the simulation results, we tested two empirical potentials and evaluated their reliability by the calculated c… ▽ More Classical molecular-dynamics simulations were carried out to study epitaxial growth of graphene on 6H-SiC(0001) substrate. It was found that there exists a threshold annealing temperature above which we observe formation of graphitic structure on the substrate. To check the sensitivity of the simulation results, we tested two empirical potentials and evaluated their reliability by the calculated characteristics of graphene, its carbon-carbon bond-length, pair correlation function, and binding energy. △ Less

Submitted 22 February, 2012; originally announced February 2012.

Comments: 7 pages, 5 figures

Journal ref: Condens. Matter Phys., 2011, vol. 14, No. 4, p. 43802:1-7

Effects of three-body interactions on the structure and thermodynamics of liquid krypton

Authors: N. Jakse, J. M. Bomont, J. L. Bretonnet

Abstract: Large-scale molecular dynamics simulations are performed to predict the structural and thermodynamic properties of liquid krypton using a potential energy function based on the two-body potential of Aziz and Slaman plus the triple-dipole Axilrod-Teller (AT) potential. By varying the strength of the AT potential we study the influence of three-body contribution beyond the triple-dipole dispersion… ▽ More Large-scale molecular dynamics simulations are performed to predict the structural and thermodynamic properties of liquid krypton using a potential energy function based on the two-body potential of Aziz and Slaman plus the triple-dipole Axilrod-Teller (AT) potential. By varying the strength of the AT potential we study the influence of three-body contribution beyond the triple-dipole dispersion. It is seen that the AT potential gives an overall good description of liquid Kr, though other contributions such as higher order three-body dispersion and exchange terms cannot be ignored. △ Less

Submitted 7 March, 2002; originally announced March 2002.

Comments: 11 pages, 3 figures, LaTeX, to appear in J. Chem. Phys