arXiv subjects

Thomas A. R. Purcell

Publications and source records attributed to Thomas A. R. Purcell.

11 recordsLinked to original sources

arXiv2025

A high-performance and portable implementation of the SISSO method for CPUs and GPUs

SISSO (sure-independence screening and sparsifying operator) is an artificial intelligence (AI) method based on symbolic regression and compressed sensing widely used in materials science research. SISSO++ is its C++ implementation that employs MPI and OpenMP for parallelization, rendering it well-suited for high-performance computing (HPC) environments. As heterogeneous hardware becomes mainstream in the HPC and AI fields, we chose to port the SISSO++ code to GPUs using the Kokkos performance-portable library. Kokkos allows us to maintain a single codebase for both Nvidia and AMD GPUs, significantly reducing the maintenance effort. In this work, we summarize the necessary code changes we did to achieve hardware and performance portability. This is accompanied by performance benchmarks on Nvidia and AMD GPUs. We demonstrate the speedups obtained from using GPUs across the three most time-consuming parts of our code.

cs.PF↗

arXiv2024

Prediction of the Infrared Absorbance Intensities and Frequencies of Hydrocarbons:A Message Passing Neural Network Approach

Accurately and efficiently predicting the infrared (IR) spectra of a molecule can provide insights into the structure-properties relationships of molecular species, which has led to a proliferation of machine learning tools designed for this purpose. However, earlier studies have focused primarily on obtaining normalized IR spectra, which limits their potential for a comprehensive analysis of molecular behavior in the IR range. For instance, to fully understand and predict the optical properties, such as the transparency characteristics, it is necessary to predict the molar absorptivity IR spectra instead. Here, we propose a graph-based communicative message passing neural network (CMPNN) algorithm that can predict both the peak positions and absolute intensities corresponding to density functional theory (DFT) calculated molar absorptivities in the IR domain. By modifying existing spectral loss functions, we show that our method is able to predict with DFT-accuracy level the IR molar absorptivities of a series of hydrocarbons containing up to ten carbon atoms and apply the model to a set of larger molecules. We also compare the predicted spectra with those generated by the direct message passing neural network (DMPNN). The results suggest that both algorithms demonstrate similar predictive capabilities for hydrocarbons, indicating that either model could be effectively used in future research on spectral prediction for such systems.

physics.chem-ph↗

arXiv2024

Accelerating the Training and Improving the Reliability of Machine-Learned Interatomic Potentials for Strongly Anharmonic Materials through Active Learning

Molecular dynamics (MD) employing machine-learned interatomic potentials (MLIPs) serve as an efficient, urgently needed complement to ab initio molecular dynamics (aiMD). By training these potentials on data generated from ab initio methods, their averaged predictions can exhibit comparable performance to ab initio methods at a fraction of the cost. However, insufficient training sets might lead to an improper description of the dynamics in strongly anharmonic materials, because critical effects might be overlooked in relevant cases, or only incorrectly captured, or hallucinated by the MLIP when they are not actually present. In this work, we show that an active learning scheme that combines MD with MLIPs (MLIP-MD) and uncertainty estimates can avoid such problematic predictions. In short, efficient MLIP-MD is used to explore configuration space quickly, whereby an acquisition function based on uncertainty estimates and on energetic viability is employed to maximize the value of the newly generated data and to focus on the most unfamiliar but reasonably accessible regions of phase space. To verify our methodology, we screen over 112 materials and identify 10 examples experiencing the aforementioned problems. Using CuI and AgGaSe$_2$ as archetypes for these problematic materials, we discuss the physical implications for strongly anharmonic effects and demonstrate how the developed active learning scheme can address these issues.

cond-mat.mtrl-sci↗

arXiv2024

Roadmap on Data-Centric Materials Science

Science is and always has been based on data, but the terms "data-centric" and the "4th paradigm of" materials research indicate a radical change in how information is retrieved, handled and research is performed. It signifies a transformative shift towards managing vast data collections, digital repositories, and innovative data analytics methods. The integration of Artificial Intelligence (AI) and its subset Machine Learning (ML), has become pivotal in addressing all these challenges. This Roadmap on Data-Centric Materials Science explores fundamental concepts and methodologies, illustrating diverse applications in electronic-structure theory, soft matter theory, microstructure research, and experimental techniques like photoemission, atom probe tomography, and electron microscopy. While the roadmap delves into specific areas within the broad interdisciplinary field of materials science, the provided examples elucidate key concepts applicable to a wider range of topics. The discussed instances offer insights into addressing the multifaceted challenges encountered in contemporary materials research.

cond-mat.mtrl-sci↗

arXiv2023

Accelerating Materials-Space Exploration for Thermal Insulators by Mapping Materials Properties via Artificial Intelligence

Reliable artificial-intelligence models have the potential to accelerate the discovery of materials with optimal properties for various applications, including superconductivity, catalysis, and thermoelectricity. Advancements in this field are often hindered by the scarcity and quality of available data and the significant effort required to acquire new data. For such applications, reliable surrogate models that help guide materials space exploration using easily accessible materials properties are urgently needed. Here, we present a general, data-driven framework that provides quantitative predictions as well as qualitative rules for steering data creation for all datasets via a combination of symbolic regression and sensitivity analysis. We demonstrate the power of the framework by generating an accurate analytic model for the lattice thermal conductivity using only 75 experimentally measured values. By extracting the most influential material properties from this model, we are then able to hierarchically screen 732 materials and find 80 ultra-insulating materials.

cond-mat.mtrl-sci↗

arXiv2023

Recent advances in the SISSO method and their implementation in the SISSO++ code

Accurate and explainable artificial-intelligence (AI) models are promising tools for the acceleration of the discovery of new materials, ore new applications for existing materials. Recently, symbolic regression has become an increasingly popular tool for explainable AI because it yields models that are relatively simple analytical descriptions of target properties. Due to its deterministic nature, the sure-independence screening and sparsifying operator (SISSO) method is a particularly promising approach for this application. Here we describe the new advancements of the SISSO algorithm, as implemented into SISSO++, a C++ code with Python bindings. We introduce a new representation of the mathematical expressions found by SISSO. This is a first step towards introducing ``grammar'' rules into the feature creation step. Importantly, by introducing a controlled non-linear optimization to the feature creation step we expand the range of possible descriptors found by the methodology. Finally, we introduce refinements to the solver algorithms for both regression and classification, that drastically increase the reliability and efficiency of SISSO. For all of these improvements to the basic SISSO algorithm, we not only illustrate their potential impact, but also fully detail how they operate both mathematically and computationally.

physics.data-an↗

arXiv2023

Anharmonicity in Thermal Insulators: An Analysis from First Principles

The anharmonicity of atomic motion limits the thermal conductivity in crystalline solids. However, a microscopic understanding of the mechanisms active in strong thermal insulators is lacking. In this letter, we classify 465 experimentally known materials with respect to their anharmonicity and perform fully anharmonic ab initio Green-Kubo calculations for 58 of them, finding 28 thermal insulators with $κ< 10$ W/mK including 6 with ultralow $κ\lesssim 1$ W/mK. Our analysis reveals that the underlying strong anharmonic dynamics is driven by the exploration of meta-stable intrinsic defect geometries. This is at variance with the frequently applied perturbative approach, in which the dynamics is assumed to evolve around a single stable geometry.

cond-mat.mtrl-sci↗

arXiv2022

Hierarchical symbolic regression for identifying key physical parameters correlated with bulk properties of perovskites

Symbolic regression identifies key physical parameters describing materials properties by uncovering correlations as nonlinear analytical expressions. However, the pool of expressions grows rapidly with complexity, compromising its efficiency. We tackle this challenge by a hierarchical approach: identified expressions are used as input parameters for obtaining more complex expressions. Crucially, this framework can transfer knowledge among properties, highlighting physical relationships. We demonstrate this strategy by using the Sure-Independence-Screening-and-Sparsifying-Operator (SISSO) approach to identify expressions correlated with the lattice constant and cohesive energy, which are then used to model the bulk modulus of ABO3 perovskites.

cond-mat.mtrl-sci↗

arXiv2021

Interpretable Machine Learning for Materials Design

Fueled by the widespread adoption of Machine Learning (ML) and the high-throughput screening of materials, the data-centric approach to materials design has asserted itself as a robust and powerful tool for the in-silico prediction of materials properties. When training models to predict material properties, researchers often face a difficult choice between a model's interpretability or its performance. We study this trade-off by leveraging four different state-of-the-art ML techniques: XGBoost, SISSO, Roost, and TPOT for the prediction of structural and electronic properties of perovskites and 2D materials. We then assess the future outlook of the continued integration of ML into materials discovery and identify key problems that will continue to challenge researchers as the size of the literature's datasets and complexity of models increases. Finally, we offer several possible solutions to these challenges with a focus on retaining interpretability and share our thoughts on magnifying the impact of ML on materials design.

cond-mat.mtrl-sci↗

arXiv2020

Anharmonicity Measure for Materials

Theoretical frameworks used to qualitatively and quantitatively describe nuclear dynamics in solids are often based on the harmonic approximation. However, this approximation is known to become inaccurate or to break down completely in many modern functional materials. Interestingly, there is no reliable measure to quantify anharmonicity so far. Thus, a systematic classification of materials in terms of anharmonicity and a benchmark of methodologies that may be appropriate for different strengths of anharmonicity is currently impossible. In this work, we derive and discuss a statistical measure that reliably classifies compounds across temperature regimes and material classes by their "degree of anharmonicity". This enables us to distinguish "harmonic" materials, for which anharmonic effects constitute a small perturbation on top of the harmonic approximation, from strongly "anharmonic" materials, for which anharmonic effects become significant or even dominant and the treatment of anharmonicity in terms of perturbation theory is more than questionable. We show that the analysis of this measure in real and reciprocal space is able to shed light on the underlying microscopic mechanisms, even at conditions close to, e.g., phase transitions or defect formation. Eventually, we demonstrate that the developed approach is computationally efficient and enables rapid high-throughput searches by scanning over a set of several hundred binary solids. The results show that strong anharmonic effects beyond the perturbative limit are not only active in complex materials or close to phase transitions, but already at moderate temperatures in simple binary compounds.

cond-mat.mtrl-sci↗

arXiv2019

Parametrically Constrained Geometry Relaxations for High-Throughput Materials Science

Reducing parameter spaces via exploiting symmetries has greatly accelerated and increased the quality of electronic-structure calculations. Unfortunately, many of the traditional methods fail when the global crystal symmetry is broken, even when the distortion is only a slight perturbation (e.g. Jahn-Teller like distortions). Here we introduce a flexible and generalizable parametric relaxation scheme, and implement it in the all-electron code FHI-aims. This approach utilizes parametric constraints to maintain symmetry at any level. After demonstrating the method's ability to relax metastable structures, we highlight its adaptability and performance over a test set of 359 materials, across thirteen lattice prototypes. Finally we show how these constraints can reduce the number of steps needed to relax local lattice distortions by an order of magnitude. The flexibility of these constraints enables a significant acceleration of the high-throughput searches for novel materials for numerous applications.

cond-mat.mtrl-sci↗