arXiv subjects

Kangming Li

Publications and source records attributed to Kangming Li.

At least 19 recordsLinked to original sources

arXiv2026

Element priors and target support shape chemical transfer in materials graph networks

Materials graph neural networks must often transfer to chemical regions weakly represented in training data. Such transfer can rely on predefined relations among elements or supervised evidence from target-containing structures, but these pathways are usually entangled. Here, held-out-element splits and incremental target support separate their roles. Without target-containing training structures, formation-energy errors depend strongly on the element representation, particularly for H, O and F. Matched perturbations show that representation-induced sharing matters beyond input dimension or numerical form, while a label-free similarity-graph prior reduces selected zero-shot errors. Adding a few target-containing structures sharply lowers errors and contracts differences among one-hot, k-hot and continuous inputs across ALIGNN and CGCNN. Calibration explains only part of this recovery, and freezing the initial element projection preserves most gains in five of six ALIGNN splits. Target support therefore shifts chemical transfer from reliance on static element relations toward learning from target-containing environments.

cond-mat.mtrl-sci↗

arXiv2026

Building informative materials datasets beyond targeted objectives

Materials science data collection can be expensive, making the reuse and long-term utility of datasets critical important for future discovery campaigns. In practice, researchers prioritize a subset of properties due to research interests. However, ignoring a subset of outcomes in data collection campaigns potentially generate datasets poorly suited for future learning tasks. Here, we present a framework for dataset construction that maximizes informativeness for target properties of interest while preserving performance on untargeted ones. Our approach uses diversity-aware selection to ensure broad coverage of the materials space. In noisy experimental dataset construction, we find that without our diversity-aware framework, prediction performance on untargeted properties can degrade by up to 40% relative to random sampling, whereas applying our framework yields improvements of up to 10% . For targeted properties, performance can degrade with respect to random sampling by up to 12.5% without diversity, while our framework achieves gains of up to 25%. Incorporating diversity into dataset construction not only preserves informativeness for the targeted properties, but also improves materials coverage for potential future objectives. As a result, the constructed datasets remain broadly informative across considered and unconsidered outcomes, ensuring unbiased quality entries and mitigating cold-start limitations in subsequent modeling and discovery campaigns.

cond-mat.mtrl-sci↗

arXiv2025

High-throughput validation of phase formability and simulation accuracy of Cantor alloys

High-throughput methods enable accelerated discovery of novel materials in complex systems such as high-entropy alloys, which exhibit intricate phase stability across vast compositional spaces. Computational approaches, including Density Functional Theory (DFT) and calculation of phase diagrams (CALPHAD), facilitate screening of phase formability as a function of composition and temperature. However, the integration of computational predictions with experimental validation remains challenging in high-throughput studies. In this work, we introduce a quantitative confidence metric to assess the agreement between predictions and experimental observations, providing a quantitative measure of the confidence of machine learning models trained on either DFT or CALPHAD input in accounting for experimental evidence. The experimental dataset was generated via high-throughput in-situ synchrotron X-ray diffraction on compositionally varied FeNiMnCr alloy libraries, heated from room temperature to ~1000 {\deg}C. Agreement between the observed and predicted phases was evaluated using either temperature-independent phase classification or a model that incorporates a temperature-dependent probability of phase formation. This integrated approach demonstrates where strong overall agreement between computation and experiment exists, while also identifying key discrepancies, particularly in FCC/BCC predictions at Mn-rich regions to inform future model refinement.

cond-mat.mtrl-sci↗

arXiv2025

When Active Learning Fails, Uncalibrated Out of Distribution Uncertainty Quantification Might Be the Problem

Efficiently and meaningfully estimating prediction uncertainty is important for exploration in active learning campaigns in materials discovery, where samples with high uncertainty are interpreted as containing information missing from the model. In this work, the effect of different uncertainty estimation and calibration methods are evaluated for active learning when using ensembles of ALIGNN, eXtreme Gradient Boost, Random Forest, and Neural Network model architectures. We compare uncertainty estimates from ALIGNN deep ensembles to loss landscape uncertainty estimates obtained for solubility, bandgap, and formation energy prediction tasks. We then evaluate how the quality of the uncertainty estimate impacts an active learning campaign that seeks model generalization to out-of-distribution data. Uncertainty calibration methods were found to variably generalize from in-domain data to out-of-domain data. Furthermore, calibrated uncertainties were generally unsuccessful in reducing the amount of data required by a model to improve during an active learning campaign on out-of-distribution data when compared to random sampling and uncalibrated uncertainties. The impact of poor-quality uncertainty persists for random forest and eXtreme Gradient Boosting models trained on the same data for the same tasks, indicating that this is at least partially intrinsic to the data and not due to model capacity alone. Analysis of the target, in-distribution uncertainty, out-of-distribution uncertainty, and training residual distributions suggest that future work focus on understanding empirical uncertainties in the feature input space for cases where ensemble prediction variances do not accurately capture the missing information required for the model to generalize.

cond-mat.mtrl-sci↗

arXiv2025

Exploring the Limitations of kNN Noisy Feature Detection and Recovery for Self-Driving Labs

Self-driving laboratories (SDLs) have shown promise to accelerate materials discovery by integrating machine learning with automated experimental platforms. However, errors in the capture of input parameters may corrupt the features used to model system performance, compromising current and future campaigns. This study develops an automated workflow to systematically detect noisy features, determine sample-feature pairings that can be corrected, and finally recover the correct feature values. A systematic study is then performed to examine how dataset size, noise intensity, noise type, and feature value distribution affect both the detectability and recoverability of noisy features on both Density Functional Theory (DFT) and SDL datasets. In general, high-intensity noise and large training datasets are conducive to the detection and correction of noisy features. Low-intensity noise reduces detection and recovery but can be compensated for by larger clean training data sets. Detection and correction results vary between features, with continuous and dispersed feature distributions showing greater recoverability compared to features with discrete or narrow distributions. This systematic study not only demonstrates a model agnostic framework for rational data recovery in the presence of noise, limited data, and differing feature distributions but also provides a tangible benchmark of kNN imputation in materials datasets. Ultimately, it aims to enhance data quality and experimental precision in automated materials discovery.

cs.LG↗

arXiv2024

LLM4Mat-Bench: Benchmarking Large Language Models for Materials Property Prediction

Large language models (LLMs) are increasingly being used in materials science. However, little attention has been given to benchmarking and standardized evaluation for LLM-based materials property prediction, which hinders progress. We present LLM4Mat-Bench, the largest benchmark to date for evaluating the performance of LLMs in predicting the properties of crystalline materials. LLM4Mat-Bench contains about 1.9M crystal structures in total, collected from 10 publicly available materials data sources, and 45 distinct properties. LLM4Mat-Bench features different input modalities: crystal composition, CIF, and crystal text description, with 4.7M, 615.5M, and 3.1B tokens in total for each modality, respectively. We use LLM4Mat-Bench to fine-tune models with different sizes, including LLM-Prop and MatBERT, and provide zero-shot and few-shot prompts to evaluate the property prediction capabilities of LLM-chat-like models, including Llama, Gemma, and Mistral. The results highlight the challenges of general-purpose LLMs in materials science and the need for task-specific predictive models and task-specific instruction-tuned LLMs in materials property prediction.

cond-mat.mtrl-sci↗

arXiv2024

Evaluating the Performance and Robustness of LLMs in Materials Science Q&A and Property Predictions

Large Language Models (LLMs) have the potential to revolutionize scientific research, yet their robustness and reliability in domain-specific applications remain insufficiently explored. In this study, we evaluate the performance and robustness of LLMs for materials science, focusing on domain-specific question answering and materials property prediction across diverse real-world and adversarial conditions. Three distinct datasets are used in this study: 1) a set of multiple-choice questions from undergraduate-level materials science courses, 2) a dataset including various steel compositions and yield strengths, and 3) a band gap dataset, containing textual descriptions of material crystal structures and band gap values. The performance of LLMs is assessed using various prompting strategies, including zero-shot chain-of-thought, expert prompting, and few-shot in-context learning. The robustness of these models is tested against various forms of 'noise', ranging from realistic disturbances to intentionally adversarial manipulations, to evaluate their resilience and reliability under real-world conditions. Additionally, the study showcases unique phenomena of LLMs during predictive tasks, such as mode collapse behavior when the proximity of prompt examples is altered and performance recovery from train/test mismatch. The findings aim to provide informed skepticism for the broad use of LLMs in materials science and to inspire advancements that enhance their robustness and reliability for practical applications.

cs.CL↗

arXiv2024

An Assessment of Commonly Used Equivalent Circuit Models for Corrosion Analysis: A Bayesian Approach to Electrochemical Impedance Spectroscopy

Electrochemical Impedance Spectroscopy (EIS) is a crucial technique for assessing corrosion of a metallic materials. The analysis of EIS hinges on the selection of an appropriate equivalent circuit model (ECM) that accurately characterizes the system under study. In this work, we systematically examined the applicability of three commonly used ECMs across several typical material degradation scenarios. By applying Bayesian Inference to simulated corrosion EIS data, we assessed the suitability of these ECMs under different corrosion conditions and identified regions where the EIS data lacks sufficient information to statistically substantiate the ECM structure. Additionally, we posit that the traditional approach to EIS analysis, which often requires measurements to very low frequencies, might not be always necessary to correctly model the appropriate ECM. Our study assesses the impact of omitting data from low to medium-frequency ranges on inference results and reveals that a significant portion of low-frequency measurements can be excluded without substantially compromising the accuracy of extracting system parameters. Further, we propose simple checks to the posterior distributions of the ECM components and posterior predictions, which can be used to quantitatively evaluate the suitability of a particular ECM and the minimum frequency required to be measured. This framework points to a pathway for expediting EIS acquisition by intelligently reducing low-frequency data collection and permitting on-the-fly EIS measurements

cond-mat.mtrl-sci↗

arXiv2024

Probing out-of-distribution generalization in machine learning for materials

Scientific machine learning (ML) endeavors to develop generalizable models with broad applicability. However, the assessment of generalizability is often based on heuristics. Here, we demonstrate in the materials science setting that heuristics based evaluations lead to substantially biased conclusions of ML generalizability and benefits of neural scaling. We evaluate generalization performance in over 700 out-of-distribution tasks that features new chemistry or structural symmetry not present in the training data. Surprisingly, good performance is found in most tasks and across various ML models including simple boosted trees. Analysis of the materials representation space reveals that most tasks contain test data that lie in regions well covered by training data, while poorly-performing tasks contain mainly test data outside the training domain. For the latter case, increasing training set size or training time has marginal or even adverse effects on the generalization performance, contrary to what the neural scaling paradigm assumes. Our findings show that most heuristically-defined out-of-distribution tests are not genuinely difficult and evaluate only the ability to interpolate. Evaluating on such tasks rather than the truly challenging ones can lead to an overestimation of generalizability and benefits of scaling.

cond-mat.mtrl-sci↗

arXiv2024

Efficient first principles based modeling via machine learning: from simple representations to high entropy materials

High-entropy materials (HEMs) have recently emerged as a significant category of materials, offering highly tunable properties. However, the scarcity of HEM data in existing density functional theory (DFT) databases, primarily due to computational expense, hinders the development of effective modeling strategies for computational materials discovery. In this study, we introduce an open DFT dataset of alloys and employ machine learning (ML) methods to investigate the material representations needed for HEM modeling. Utilizing high-throughput DFT calculations, we generate a comprehensive dataset of 84k structures, encompassing both ordered and disordered alloys across a spectrum of up to seven components and the entire compositional range. We apply descriptor-based models and graph neural networks to assess how material information is captured across diverse chemical-structural representations. We first evaluate the in-distribution performance of ML models to confirm their predictive accuracy. Subsequently, we demonstrate the capability of ML models to generalize between ordered and disordered structures, between low-order and high-order alloys, and between equimolar and non-equimolar compositions. Our findings suggest that ML models can generalize from cost-effective calculations of simpler systems to more complex scenarios. Additionally, we discuss the influence of dataset size and reveal that the information loss associated with the use of unrelaxed structures could significantly degrade the generalization performance. Overall, this research sheds light on several critical aspects of HEM modeling and offers insights for data-driven atomistic modeling of HEMs.

cond-mat.mtrl-sci↗

arXiv2024

Accurate and efficient predictions of keyhole dynamics in laser materials processing using machine learning-aided simulations

The keyhole phenomenon has been widely observed in laser materials processing, including laser welding, remelting, cladding, drilling, and additive manufacturing. Keyhole-induced defects, primarily pores, dramatically affect the performance of final products, impeding the broad use of these laser-based technologies. The formation of these pores is typically associated with the dynamic behavior of the keyhole. So far, the accurate characterization and prediction of keyhole features, particularly keyhole depth, as a function of time, has been a challenging task. In situ characterization of keyhole dynamic behavior using the synchrotron X-ray technique is informative but complicated and expensive. Current simulations are generally hindered by their poor accuracy and generalization abilities in predicting keyhole depths due to the lack of accurate laser absorptance data. In this study, we develop a machine learning-aided simulation method that accurately predicts keyhole dynamics, especially in keyhole depth fluctuations, over a wide range of processing parameters. In two case studies involving titanium and aluminum alloys, we achieve keyhole depth prediction with a mean absolute percentage error of 10%, surpassing those simulated using the ray-tracing method with an error margin of 30%, while also reducing computational time. This exceptional fidelity and efficiency empower our model to serve as a cost-effective alternative to synchrotron experiments. Our machine learning-aided simulation method is affordable and readily deployable for a large variety of materials, opening new doors to eliminate or reduce defects for a wide range of laser materials processing techniques.

cond-mat.mtrl-sci↗

arXiv2023

JARVIS-Leaderboard: A Large Scale Benchmark of Materials Design Methods

Lack of rigorous reproducibility and validation are major hurdles for scientific development across many fields. Materials science in particular encompasses a variety of experimental and theoretical approaches that require careful benchmarking. Leaderboard efforts have been developed previously to mitigate these issues. However, a comprehensive comparison and benchmarking on an integrated platform with multiple data modalities with both perfect and defect materials data is still lacking. This work introduces JARVIS-Leaderboard, an open-source and community-driven platform that facilitates benchmarking and enhances reproducibility. The platform allows users to set up benchmarks with custom tasks and enables contributions in the form of dataset, code, and meta-data submissions. We cover the following materials design categories: Artificial Intelligence (AI), Electronic Structure (ES), Force-fields (FF), Quantum Computation (QC) and Experiments (EXP). For AI, we cover several types of input data, including atomic structures, atomistic images, spectra, and text. For ES, we consider multiple ES approaches, software packages, pseudopotentials, materials, and properties, comparing results to experiment. For FF, we compare multiple approaches for material property predictions. For QC, we benchmark Hamiltonian simulations using various quantum algorithms and circuits. Finally, for experiments, we use the inter-laboratory approach to establish benchmarks. There are 1281 contributions to 274 benchmarks using 152 methods with more than 8 million data-points, and the leaderboard is continuously expanding. The JARVIS-Leaderboard is available at the website: https://pages.nist.gov/jarvis_leaderboard

cond-mat.mtrl-sci↗

arXiv2023

AutoEIS: automated Bayesian model selection and analysis for electrochemical impedance spectroscopy

Electrochemical Impedance Spectroscopy (EIS) is a powerful tool for electrochemical analysis; however, its data can be challenging to interpret. Here, we introduce a new open-source tool named AutoEIS that assists EIS analysis by automatically proposing statistically plausible equivalent circuit models (ECMs). AutoEIS does this without requiring an exhaustive mechanistic understanding of the electrochemical systems. We demonstrate the generalizability of AutoEIS by using it to analyze EIS datasets from three distinct electrochemical systems, including thin-film oxygen evolution reaction (OER) electrocatalysis, corrosion of self-healing multi-principal components alloys, and a carbon dioxide reduction electrolyzer device. In each case, AutoEIS identified competitive or in some cases superior ECMs to those recommended by experts and provided statistical indicators of the preferred solution. The results demonstrated AutoEIS's capability to facilitate EIS analysis without expert labels while diminishing user bias in a high-throughput manner. AutoEIS provides a generalized automated approach to facilitate EIS analysis spanning a broad suite of electrochemical applications with minimal prior knowledge of the system required. This tool holds great potential in improving the efficiency, accuracy, and ease of EIS analysis and thus creates an avenue to the widespread use of EIS in accelerating the development of new electrochemical materials and devices.

cond-mat.mtrl-sci↗

arXiv2023

On the redundancy in large material datasets: efficient and robust learning with less data

Extensive efforts to gather materials data have largely overlooked potential data redundancy. In this study, we present evidence of a significant degree of redundancy across multiple large datasets for various material properties, by revealing that up to 95 % of data can be safely removed from machine learning training with little impact on in-distribution prediction performance. The redundant data is related to over-represented material types and does not mitigate the severe performance degradation on out-of-distribution samples. In addition, we show that uncertainty-based active learning algorithms can construct much smaller but equally informative datasets. We discuss the effectiveness of informative data in improving prediction performance and robustness and provide insights into efficient data acquisition and machine learning training. This work challenges the "bigger is better" mentality and calls for attention to the information richness of materials data rather than a narrow emphasis on data volume.

cond-mat.mtrl-sci↗

arXiv2023

Magnetochemical coupling effects on thermodynamics, point-defect formation and diffusion in Fe-Ni alloys: a theoretical study

This thesis is a theoretical study of thermodynamic, point-defect formation and diffusion properties in Fe-Ni alloys with a focus on the magnetochemical effects. The results are derived from density functional theory (DFT) calculations and Monte Carlo (MC) simulations using a DFT-parametrized effective interaction model (EIM) with explicit atomic and spin variables. The first part of this work is focused on thermodynamics. We compute via DFT energetic, magnetic and vibrational properties and the bcc-fcc phase diagram, revealing the relative importance between magnetic and vibrational entropies. Combining MC simulations with the EIM, we obtain an fcc phase diagram across the Curie points. We also discuss Mn and Cr effects on phase stability. The second part of the work is dedicated to point-defect properties. We develop MC schemes to compute vacancy formation free energy in alloys. We show that vacancy formation in fcc Fe and Ni exhibits features that are well distinct from those in bcc Fe. The results in fcc Fe-Ni alloys reveal that magnetic disorder tends to increase vacancy formation free energy, while chemical disorder shows an opposite effect. We also study magnetic effects on the properties of self-interstitials in fcc Fe and Ni. The final part of the work is devoted to vacancy-mediated diffusion. We evaluate diffusion properties over the whole concentration range, probe into the magnetochemical effects on diffusion. This work fully takes into account the impacts of transversal and longitudinal spin fluctuations and the magnetochemical interplay. It provides an accurate and consistent prediction of thermodynamic, defect formation and diffusion properties in the Fe-Ni system, and contributes to a better understanding of effects of magnetism in austenitic steels. The applied approach is also transferable to the investigation of other magnetic alloys.

cond-mat.mtrl-sci↗

arXiv2022

A critical examination of robustness and generalizability of machine learning prediction of materials properties

Recent advances in machine learning (ML) methods have led to substantial improvement in materials property prediction against community benchmarks, but an excellent benchmark score may not imply good generalization of performance. Here we show that ML models trained on the Materials Project 2018 (MP18) dataset can have severely degraded prediction performance on new compounds in the Materials Project 2021 (MP21) dataset. We document performance degradation in graph neural networks and traditional descriptor-based ML models for both quantitative and qualitative predictions. We find the source of the predictive degradation is due to the distribution shift between the MP18 and MP21 versions. This is revealed by the uniform manifold approximation and projection (UMAP) of the feature space. We then show that the performance degradation issue can be foreseen using a few simple tools. Firstly, the UMAP can be used to investigate the connectivity and relative proximity of the training and test data within feature space. Secondly, the disagreement between multiple ML models on the test data can illuminate out-of-distribution samples. We demonstrate that the simple yet efficient UMAP-guided and query-by-committee acquisition strategies can greatly improve prediction accuracy through adding only 1~\% of the test data. We believe this work provides valuable insights for building materials databases and ML models that enable better prediction robustness and generalizability.

cond-mat.mtrl-sci↗

arXiv2022

Magnetochemical effects on phase stability and vacancy formation in fcc Fe-Ni alloys

We investigate phase stability and vacancy formation in fcc Fe-Ni alloys over a broad composition-temperature range, via a density functional theory parametrized effective interaction model, which includes explicitly spin and chemical variables. On-lattice Monte Carlo simulations based on this model are used to predict the temperature evolution of the magnetochemical phase. The experimental composition-dependent Curie and chemical order-disorder transition temperatures are successfully predicted. We point out a significant effect of chemical and magnetic orders on the magnetic and chemical transitions, respectively. The resulting phase diagram shows a magnetically driven phase separation around 10-40% Ni and 570-700 K, between ferromagnetic and paramagnetic solid solutions, in agreement with experimental observations. We compute vacancy formation magnetic free energy as a function of temperature and alloy composition. We identify opposite magnetic and chemical disordering effects on vacancy formation in the alloys with 50% and 75% Ni. We find that thermal magnetic effects on vacancy formation are much larger in concentrated Fe-Ni alloys than in fcc Fe and Ni due to a stronger magnetic interaction.

cond-mat.mtrl-sci↗

arXiv2021

Effects of magnetic excitations and transitions on vacancy formation: cases of fcc Fe and Ni compared to bcc Fe

Vacancy is one of the most frequent defects in metals. We study the impacts of magnetism on vacancy formation properties in fcc Ni, and in bcc and fcc Fe, via density functional theory (DFT) and effective interaction models combined with Monte Carlo simulations. Overall, the predicted vacancy formation energies and equilibrium vacancy concentrations are in good agreement with experimental data, available only at the high-temperature paramagnetic regime. Effects of magnetic transitions on vacancy formation energies are found to be more important in bcc Fe than in fcc Fe and Ni. The distinct behaviour is correlated to the relative roles of longitudinal and transversal spin excitations. At variance with the bcc-Fe case, we note a clear effect of longitudinal spin excitations on the magnetic free energy of vacancy formation in fcc Fe and Ni, leading to its steady variation above the respective magnetic transition temperature. Below the N\'{e}el point, such effect in fcc Fe is comparable but opposite to the one of the transversal excitations. Regarding fcc Ni, although neglecting the longitudinal spin excitations induces an overestimation of the Curie temperature by 220 K, no additional effect is visible below the Curie point. The distinct effects on the three systems are closely linked to DFT predictions of the dependence of vacancy formation energy on the variation of local magnetic-moment magnitudes and orientations.

cond-mat.mtrl-sci↗