arXiv subjects

Noel Jakse

Publications and source records attributed to Noel Jakse.

11 recordsLinked to original sources

arXiv2026

Predicting Spin-Crossover Behavior in Metal-Organic Frameworks from Limited and Noisy Data Using Quantile Active Learning

Spin-crossover (SCO) metal-organic frameworks (MOFs) hold great promise for sensing, spintronics, and gas-related applications, however, only a small number of SCO-active examples are known among the thousands of MOFs already synthesized. Computational screening enhanced by machine learning offers a powerful route to uncover these hidden candidates much more rapidly than trial-and-error experiments. However, progress is limited by the computational complexity of obtaining accurate adiabatic energy differences, as these typically require separate geometry optimizations for both spin states, a process that is technically challenging, prone to convergence failures, and difficult to automate at scale. To mitigate these issues, we introduce a data-efficient strategy based on Quantile Regression Tree-based Active Learning, designed to navigate large chemical spaces while remaining robust to noisy and scarce labels obtained from unrelaxed geometries. After actively selecting a 200-sized subset of representative MOFs for electronic-structure evaluation, a Random Forest regressor trained on this data accurately identifies SCO-relevant candidates despite label noise, recovering 82% of true positives with only two false negatives. Applying the model to the unlabeled dataset yields a new collection of high-confidence SCO MOFs, which we denote pSCO-105. This work shows that spin crossover can be reliably identified from limited and imperfect data through smart training-set selection, enabling accelerated screening of SCO MOFs.

cond-mat.mtrl-sci↗

arXiv2025

Crystal Nucleation in Eutectic Al-Si Alloys by Machine-Learned Molecular Dynamics

Solidification control is crucial in manufacturing technologies, as it determines the microstructure and, consequently, the performance of the final product. Investigating the mechanisms occurring during the early stages of nucleation remains experimentally challenging as it initiates on nanometer length and sub-picoseconds time scales. Large scale molecular dynamics simulations using machine learning interatomic potential with quantum accuracy appears the dedicated approach to complex, atomic level, multidimensional mechanisms with local symmetry breaking. A potential trained on a high-dimensional neural network on density functional theory-based ab initio molecular dynamics (AIMD) trajectories for liquid and undercooled states for Al-Si binary alloys enables us to study the nucleation mechanisms occurring at the early stages from the liquid phase near the eutectic composition. Our results indicate that nucleation starts with Al in hypoeutectic conditions and with Si in hypereutectic conditions. Whereas Al nuclei grow in a globular shape, Si ones grow with polygonal faceting, whose underlying mechanisms are further discussed.

cond-mat.mtrl-sci↗

arXiv2024

Impact of local structure on melt dynamics in Cu-Ti alloys: Insights from ab-initio molecular dynamics simulations

First-principle based molecular-dynamics simulations have been performed for binary Cu$_x$Ti$_{1-x}$ (x = 0.31, 0.50, and 0.76) alloys to investigate the relationship between local structure and dynamical properties in the liquid and undercooled melt. The undercooled melts show a pronounced short-range order, majorly a five-fold symmetry (FFS) around the Cu atoms, which competes with bcc ordering. This complex SRO is also reflected in the partial coordination numbers, where mainly a Z12 coordination is present around Cu, which corresponds to an icosahedral ordering. Higher coordination numbers were obtained for Ti compatible with Frank-Kasper polyhedra. The increasing Frank-Kasper polyhedra coordination scenario around Ti impacts on the interatomic distances of Ti atoms, which increase with increasing Ti content. The Cu$_{50}$Ti$_{50}$ composition exhibits the highest FFS ordering and amount of Frank-Kasper polyhedra, which explains the slowest melt dynamics, found experimentally and in simulations for this composition. Thus, our results suggest that the high undercooling degree and glass-forming ability of binary CuTi alloys, originates from the high complexity of the local structure rather than due to the preferred formation of Cu-Ti pairs, as Cu-Ti interactions were found to be weak.

cond-mat.mtrl-sci↗

arXiv2024

Diffusion in liquid metals is directed by competing collective modes

The self-diffusion process in a dense liquid is influenced by collective particle movements. Extensive molecular dynamics simulations for liquid aluminium and rubidium evidence a crossover in the diffusion coefficient at about $1.4$ times the melting temperature $T_m$, indicating a profound change in the diffusion mechanism. The corresponding velocity auto-correlation functions demonstrate a decrease of the cage effect with a gradual set-in of a power-law decay, the celebrate {\it long time tail}. This behavior is caused by a competition of density fluctuations near the melting point with vortex-type particle patterns from transverse currents in the hot fluid. The investigation of the velocity autocorrelation function evidences a gradual transition in dynamics with rising temperature. The competition between these two collective particle movements, one hindering and one enhancing the diffusion process, leads to a non-Arrhenius-type behavior of the diffusion coefficient around $1.4~T_m$, which signals the transition from a dense to a fluid-like liquid dynamics in the potential energy landscape picture.

cond-mat.dis-nn↗

arXiv2024

Homogeneous Nucleation of Undercooled Al-Ni melts via a Machine-Learned Interaction Potential

Homogeneous nucleation processes are important for understanding solidification and the resulting microstructure of materials. Simulating this process requires accurately describing the interactions between atoms, hich is further complicated by chemical order through cross-species interactions. The large scales needed to observe rare nucleation events are far beyond the capabilities of ab initio simulations. Machine-learning is used for overcoming these limitations in terms of both accuracy and speed, by building a high-dimensional neural network potential for binary Al-Ni alloys, which serve as a model system relevant to many industrial applications. The potential is validated against experimental diffusion, viscosity, and scattering data, and is applied to large-scale molecular dynamics simulations of homogeneous nucleation at equiatomic composition, as well as for pure Ni. Pure Ni nucleates in a single-step into an fcc crystal phase, in contrast to previous results obtained with a classical empirical potential. This highlights the sensitivity of nucleation pathways to the underlying atomic interactions. Our findings suggest that the nucleation pathway for AlNi proceeds in a single step toward a B2 structure, which is discussed in relation to the pure elements counterparts.

cond-mat.mtrl-sci↗

arXiv2024

Classification Tree-based Active Learning: A Wrapper Approach

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.

cs.LG↗

arXiv2023

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

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.

cond-mat.dis-nn↗

arXiv2023

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

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.

cond-mat.other↗

arXiv2022

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

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.

cond-mat.mtrl-sci↗

arXiv2022

Interatomic machine learning potentials for aluminium: application to solidification phenomena

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.

cond-mat.mtrl-sci↗

arXiv2021

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

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.

cond-mat.stat-mech↗