arXiv subjects

Conrard Giresse Tetsassi Feugmo

Publications and source records attributed to Conrard Giresse Tetsassi Feugmo.

6 recordsLinked to original sources

arXiv2026

Sparse data limit what a mechanistic corrosion model can predict

Mechanistic corrosion models are routinely fitted with four to six parameters to a few measurements and extrapolated across service lifetimes, yet whether the data determine them is rarely tested. We identify a reduced point defect model for the duplex oxide on Nb-stabilized AISI 347 in simulated boiling-water-reactor water from published depth profiles at three exposure times, by differentiable inversion graded against a closed-form reference. Profile likelihood, a 1000-member bootstrap and a prior-relaxation test agree that one of the five parameters is undetermined and a second only weakly so, and that 95% of the fit statistic rests on three chromium points. The kinetics track an empirical power law over the calibration window but diverge from it by a factor of 3.5 to 6.3 at ten years, outside the propagated uncertainty band. That separation, not the thickness it brackets, is what these data determine; an exposure near 3000 hours would resolve it.

cond-mat.mtrl-sci↗

arXiv2026

Differentiable inverse design of short-range order in high-entropy alloys: from target sro to target property

Short-range order (SRO) governs the mechanical response of multi-principal-element alloys, but designing an alloy for a target property usually means solving two disconnected problems: building a structure matching a desired SRO pattern, then separately checking its property, with no shared optimization. This work replaces the standard random-swap search (reverse Monte Carlo) with a gradient-based approach: atom occupancy is treated as continuous rather than fixed, so the whole process can be tuned using gradient descent, the same method used to train neural networks. This builder matches random-swap accuracy on small systems, but is six times faster and eight times more accurate on large 4000-atom systems, and scales smoothly to alloys with many elements without extra bookkeeping. A physics-based correction term, adapted from prior two-element work and extended here to many elements, keeps designed structures thermodynamically realistic rather than just numerically matching the target SRO pattern. A small neural network then predicts mechanical properties directly from composition and SRO statistics, closing the loop from target property back to structure. Tested on nine face-centered-cubic and body-centered-cubic alloys, the pipeline captured SRO-driven stiffness changes from -20% to +57%, and cell-size checks showed at least 864 atoms are needed to get the direction and size of these changes right, since the commonly used 108-atom cells can mislead. Against real simulations for a cobalt-chromium-nickel alloy, the method matched three of four target stiffness values within 6%. The method is released as an open-source Python package, anisro, offering a practical route to gradient-based, property-driven alloy design.

cond-mat.mtrl-sci↗

arXiv2026

A Systematic Benchmark of Physics-Informed Neural Network Architectures for the Stiff Poisson-Nernst-Planck System: Adaptive LossWeighting and Multi-Scale Resolution

The Poisson Nernst Planck PNP system constitutes a canonical stiff coupled PDE problem where the charge density prefactor produces extreme coefficient ratios and the electric double layer imposes sharp boundary layers. Physics informed neural networks PINNs are appealing here because they require no mesh and differentiate through the physics automatically. Spectral bias and multi task loss imbalance however have limited their accuracy on stiff PNP systems. We present the first systematic data free benchmark of eleven PINN configurations organised into four strategy groups on a physically parametrised one dimensional PNP model for a lithium symmetric cell implemented within NVIDIA PhysicsNeMo Sym and validated against a finite volume method FVM reference. Root mean square errors RMSE span across architectures. The balanced residual decay rate BRDR scheme matches Neural Tangent Kernel NTK performance for concentration fields while reducing mean wall clock time making it the preferable strategy under compute constraints. Loss landscape geometry corroborates the RMSE ranking. We release an open source PhysicsNeMo Sym implementation for reuse on stiff coupled PDE problems in computational mechanics.

physics.app-ph↗

arXiv2026

Neural Spectral Element Methods for stiff multiphysics PDEs with electrochemical transport benchmarks

The Neural Spectral Element Method (NSEM) evaluates each network only at fixed Legendre-Gauss-Lobatto quadrature nodes and replaces all derivative calls with precomputed spectral differentiation matrices. The resulting deterministic loss enables limited-memory BFGS (L-BFGS) to reach residuals of 10^-9 to 10^-10. A Kosloff-Tal-Ezer coordinate map resolves electrochemical boundary layers, while a mesh-free neural mortar framework couples multi-element domains. On the four-example Poisson-Nernst-Planck (PNP) benchmark of Huang and co-workers, NSEM attains 10^-4 to 10^-7 relative pointwise error with two orders of magnitude fewer collocation points than the adaptive-resampling PINN baseline. Both a tanh multilayer perceptron (MLP) and a basis-aligned Legendre Kolmogorov-Arnold Network (KAN) backbone attain spectral accuracy within the same NSEM infrastructure, with the KAN requiring roughly half the Adam steps to enter the L-BFGS basin of attraction on the 1D PNP benchmark.

cond-mat.mtrl-sci↗

arXiv2023

Automatic graph representation algorithm for heterogeneous catalysis

One of the most appealing aspects of machine learning for material design is its high throughput exploration of chemical spaces, but to reach the ceiling of ML-aided exploration, more than current model architectures and processing algorithms are required. New architectures such as Graph Neural Networks (GNNs) have seen significant research investments recently. For heterogeneous catalysis, defining substrate intramolecular bonds and adsorbate/substrate intermolecular bonds is a time-consuming and challenging process. Before applying a model, dataset pre-processing, node/bond descriptor design, and specific model constraints have to be considered. In this work, a framework designed to solve these issues is presented in the form of an automatic graph representation algorithm (AGRA) tool to extract the local chemical environment of metallic surface adsorption sites is presented. This tool is able to gather multiple adsorption geometry datasets composed of different systems and combine them into a single model. To show AGRA's excellent transferability and reduced computational cost compared to other graph representation methods, it was applied to 5 different catalytic reaction datasets and benchmarked against the Open Catalyst Projects (OCP) graph representation method. The two ORR datasets with O/OH adsorbates obtained 0.053 eV RMSD when combined together, whereas the three CO2RR datasets with CHO/CO/COOH obtained an average performance of 0.088 eV RMSD. To further display the algorithm's versatility and extrapolation ability, a model was trained on a subset combination of all 5 datasets with an RMSD of 0.105 eV. This universal model was then used to predict a wide range of adsorption energies and an entirely new ORR catalyst system and then verified through Density Functional Theory calculations

cond-mat.mtrl-sci↗

arXiv2021

Neural evolution structure generation: High Entropy Alloys

We propose a method of neural evolution structures (NESs) combining artificial neural networks (ANNs) and evolutionary algorithms (EAs) to generate High Entropy Alloys (HEAs) structures. Our inverse design approach is based on pair distribution functions and atomic properties and allows one to train a model on smaller unit cells and then generate a larger cell. With a speed-up factor of approximately 1000 with respect to the SQSs, the NESs dramatically reduces computational costs and time, making possible the generation of very large structures (over 40,000 atoms) in few hours. Additionally, unlike the SQSs, the same model can be used to generate multiple structures with the same fractional composition.

cond-mat.dis-nn↗