authors
stringlengths
11
2.41k
title
stringlengths
38
184
journal-ref
stringclasses
115 values
doi
stringlengths
17
34
report-no
stringclasses
3 values
categories
stringlengths
17
83
abstract
stringlengths
124
1.92k
versions
stringlengths
62
689
update_date
stringdate
2007-09-13 00:00:00
2025-05-15 00:00:00
Siyu Liu, Tongqi Wen, A. S. L. Subrahmanyam Pattamatta, and David J. Srolovitz
A Prompt-Engineered Large Language Model, Deep Learning Workflow for Materials Classification
null
10.1016/j.mattod.2024.08.028
null
cond-mat.mtrl-sci
Large language models (LLMs) have demonstrated rapid progress across a wide array of domains. Owing to the very large number of parameters and training data in LLMs, these models inherently encompass an expansive and comprehensive materials knowledge database, far exceeding the capabilities of individual researcher. Nonetheless, devising methods to harness the knowledge embedded within LLMs for the design and discovery of novel materials remains a formidable challenge. We introduce a general approach for addressing materials classification problems, which incorporates LLMs, prompt engineering, and deep learning. Utilizing a dataset of metallic glasses as a case study, our methodology achieved an improvement of up to 463% in prediction accuracy compared to conventional classification models. These findings underscore the potential of leveraging textual knowledge generated by LLMs for materials especially in the common situation where datasets are sparse, thereby promoting innovation in materials discovery and design.
[{'version': 'v1', 'created': 'Wed, 31 Jan 2024 12:31:52 GMT'}, {'version': 'v2', 'created': 'Wed, 27 Mar 2024 13:22:22 GMT'}]
2024-11-20
Hoang-Giang Nguyen, Thanh-Dung Le
Predictive Models based on Deep Learning Algorithms for Tensile Deformation of AlCoCuCrFeNi High-entropy alloy
null
null
null
cond-mat.mtrl-sci eess.SP
High-entropy alloys (HEAs) stand out between multi-component alloys due to their attractive microstructures and mechanical properties. In this investigation, molecular dynamics (MD) simulation and machine learning were used to ascertain the deformation mechanism of AlCoCuCrFeNi HEAs under the influence of temperature, strain rate, and grain sizes. First, the MD simulation shows that the yield stress decreases significantly as the strain and temperature increase. In other cases, changes in strain rate and grain size have less effect on mechanical properties than changes in strain and temperature. The alloys exhibited superplastic behavior under all test conditions. The deformity mechanism discloses that strain and temperature are the main sources of beginning strain, and the shear bands move along the uniaxial tensile axis inside the workpiece. Furthermore, the fast phase shift of inclusion under mild strain indicates the relative instability of the inclusion phase of HCP. Ultimately, the dislocation evolution mechanism shows that the dislocations are transported to free surfaces under increased strain when they nucleate around the grain boundary. Surprisingly, the ML prediction results also confirm the same characteristics as those confirmed from the MD simulation. Hence, the combination of MD and ML reinforces the confidence in the findings of mechanical characteristics of HEA. Consequently, this combination fills the gaps between MD and ML, which can significantly save time human power and cost to conduct real experiments for testing HEA deformation in practice.
[{'version': 'v1', 'created': 'Fri, 2 Feb 2024 17:17:30 GMT'}]
2024-02-05
Yutack Park, Jaesun Kim, Seungwoo Hwang, and Seungwu Han
Scalable Parallel Algorithm for Graph Neural Network Interatomic Potentials in Molecular Dynamics Simulations
Journal of Chemical Theory and Computation 20 (2024) 4857-4868
10.1021/acs.jctc.4c00190
null
cond-mat.mtrl-sci
Message-passing graph neural network interatomic potentials (GNN-IPs), particularly those with equivariant representations such as NequIP, are attracting significant attention due to their data efficiency and high accuracy. However, parallelizing GNN-IPs poses challenges because multiple message-passing layers complicate data communication within the spatial decomposition method, which is preferred by many molecular dynamics (MD) packages. In this article, we propose an efficient parallelization scheme compatible with GNN-IPs and develop a package, SevenNet (Scalable EquiVariance-Enabled Neural NETwork), based on the NequIP architecture. For MD simulations, SevenNet interfaces with the LAMMPS package. Through benchmark tests on a 32-GPU cluster with examples of SiO$_2$, SevenNet achieves over 80% parallel efficiency in weak-scaling scenarios and exhibits nearly ideal strong-scaling performance as long as GPUs are fully utilized. However, the strong-scaling performance significantly declines with suboptimal GPU utilization, particularly affecting parallel efficiency in cases involving lightweight models or simulations with small numbers of atoms. We also pre-train SevenNet with a vast dataset from the Materials Project (dubbed `SevenNet-0') and assess its performance on generating amorphous Si$_3$N$_4$ containing more than 100,000 atoms. By developing scalable GNN-IPs, this work aims to bridge the gap between advanced machine learning models and large-scale MD simulations, offering researchers a powerful tool to explore complex material systems with high accuracy and efficiency.
[{'version': 'v1', 'created': 'Tue, 6 Feb 2024 08:10:02 GMT'}]
2024-06-13
Zilong Yuan, Zhiming Xu, He Li, Xinle Cheng, Honggeng Tao, Zechen Tang, Zhiyuan Zhou, Wenhui Duan, Yong Xu
Equivariant Neural Network Force Fields for Magnetic Materials
null
null
null
cond-mat.mtrl-sci
Neural network force fields have significantly advanced ab initio atomistic simulations across diverse fields. However, their application in the realm of magnetic materials is still in its early stage due to challenges posed by the subtle magnetic energy landscape and the difficulty of obtaining training data. Here we introduce a data-efficient neural network architecture to represent density functional theory total energy, atomic forces, and magnetic forces as functions of atomic and magnetic structures. Our approach incorporates the principle of equivariance under the three-dimensional Euclidean group into the neural network model. Through systematic experiments on various systems, including monolayer magnets, curved nanotube magnets, and moir\'e-twisted bilayer magnets of $\text{CrI}_{3}$, we showcase the method's high efficiency and accuracy, as well as exceptional generalization ability. The work creates opportunities for exploring magnetic phenomena in large-scale materials systems.
[{'version': 'v1', 'created': 'Wed, 7 Feb 2024 13:59:47 GMT'}]
2024-02-08
Elena Stellino, Beatrice D'Al\`o, Elena Blundo, Paolo Postorino, Antonio Polimeni
Fine-Tuning of the Excitonic Response in Monolayer WS2 Domes via Coupled Pressure and Strain Variation
null
null
null
cond-mat.mtrl-sci
We present a spectroscopic investigation into the vibrational and optoelectronic properties of WS2 domes in the 0-0.65 GPa range. The pressure evolution of the system morphology, deduced by the combined analysis of Raman and photoluminescence spectra, revealed a significant variation in the dome's aspect ratio. The modification of the dome shape caused major changes in the mechanical properties of the system resulting in a sizable increase of the out-of-plane compressive strain while keeping the in-plane tensile strain unchanged. The variation of the strain gradients drives a non-linear behavior in both the exciton energy and radiative recombination intensity, interpreted as the consequence of a hybridization mechanism between the electronic states of two distinct minima in the conduction band. Our results indicate that pressure and strain can be efficiently combined in low dimensional systems with unconventional morphology to obtain modulations of the electronic band structure not achievable in planar crystals.
[{'version': 'v1', 'created': 'Wed, 7 Feb 2024 14:09:44 GMT'}]
2024-02-08
Miao Liu, Sheng Meng
Recent Breakthrough in AI-Driven Materials Science: Tech Giants Introduce Groundbreaking Models
Mater. Futures 3 027501 (2024)
10.1088/2752-5724/ad2e0c
null
cond-mat.mtrl-sci
A close look of Google's GNoME inorganic materials dataset [Nature 624, 80 (2023)], and 11 things you would like to know.
[{'version': 'v1', 'created': 'Thu, 8 Feb 2024 16:39:26 GMT'}]
2024-03-13
Francis G. VanGessel, Efrem Perry, Salil Mohan, Oliver M. Barham, Mark Cavolowsky
NLP for Knowledge Discovery and Information Extraction from Energetics Corpora
null
null
null
cs.CL cond-mat.mtrl-sci
We present a demonstration of the utility of NLP for aiding research into energetic materials and associated systems. The NLP method enables machine understanding of textual data, offering an automated route to knowledge discovery and information extraction from energetics text. We apply three established unsupervised NLP models: Latent Dirichlet Allocation, Word2Vec, and the Transformer to a large curated dataset of energetics-related scientific articles. We demonstrate that each NLP algorithm is capable of identifying energetic topics and concepts, generating a language model which aligns with Subject Matter Expert knowledge. Furthermore, we present a document classification pipeline for energetics text. Our classification pipeline achieves 59-76\% accuracy depending on the NLP model used, with the highest performing Transformer model rivaling inter-annotator agreement metrics. The NLP approaches studied in this work can identify concepts germane to energetics and therefore hold promise as a tool for accelerating energetics research efforts and energetics material development.
[{'version': 'v1', 'created': 'Sat, 10 Feb 2024 14:43:08 GMT'}]
2024-02-13
Xiang Huang, C. Y. Zhao, Hong Wang, Shenghong Ju
AI-assisted inverse design of sequence-ordered high intrinsic thermal conductivity polymers
Materials Today Physics 44, 101438, 2024
10.1016/j.mtphys.2024.101438
null
cond-mat.soft cond-mat.mtrl-sci physics.app-ph physics.comp-ph
Artificial intelligence (AI) promotes the polymer design paradigm from a traditional trial-and-error approach to a data-driven style. Achieving high thermal conductivity (TC) for intrinsic polymers is urgent because of their importance in the thermal management of many industrial applications such as microelectronic devices and integrated circuits. In this work, we have proposed a robust AI-assisted workflow for the inverse design of high TC polymers. By using 1144 polymers with known computational TCs, we construct a surrogate deep neural network model for TC prediction and extract a polymer-unit library with 32 sequences. Two state-of-the-art multi-objective optimization algorithms of unified non-dominated sorting genetic algorithm III (U-NSGA-III) and q-noisy expected hypervolume improvement (qNEHVI) are employed for sequence-ordered polymer design with both high TC and synthetic possibility. For triblock polymer design, the result indicates that qNHEVI is capable of exploring a diversity of optimal polymers at the Pareto front, but the uncertainty in Quasi-Monte Carlo sampling makes the trials costly. The performance of U-NSGA-III is affected by the initial random structures and usually falls into a locally optimal solution, but it takes fewer attempts with lower costs. 20 parallel U-NSGA-III runs are conducted to design the pentablock polymers with high TC, and half of the candidates among 1921 generated polymers achieve the targets (TC > 0.4 W/(mK) and SA < 3.0). Ultimately, we check the TC of 50 promising polymers through molecular dynamics simulations and reveal the intrinsic connections between microstructures and TCs. Our developed AI-assisted inverse design approach for polymers is flexible and universal, and can be extended to the design of polymers with other target properties.
[{'version': 'v1', 'created': 'Sun, 18 Feb 2024 14:34:57 GMT'}]
2024-05-01
Binh Duong Nguyen, Johannes Steiner, Peter Wellmann, Stefan Sandfeld
Combining unsupervised and supervised learning in microscopy enables defect analysis of a full 4H-SiC wafer
null
null
null
cs.CV cond-mat.mtrl-sci cs.LG
Detecting and analyzing various defect types in semiconductor materials is an important prerequisite for understanding the underlying mechanisms as well as tailoring the production processes. Analysis of microscopy images that reveal defects typically requires image analysis tasks such as segmentation and object detection. With the permanently increasing amount of data that is produced by experiments, handling these tasks manually becomes more and more impossible. In this work, we combine various image analysis and data mining techniques for creating a robust and accurate, automated image analysis pipeline. This allows for extracting the type and position of all defects in a microscopy image of a KOH-etched 4H-SiC wafer that was stitched together from approximately 40,000 individual images.
[{'version': 'v1', 'created': 'Tue, 20 Feb 2024 20:04:23 GMT'}]
2024-02-22
Bashir Kazimi and Karina Ruzaeva and Stefan Sandfeld
Self-Supervised Learning with Generative Adversarial Networks for Electron Microscopy
null
null
null
cs.CV cond-mat.mtrl-sci cs.AI cs.LG
In this work, we explore the potential of self-supervised learning with Generative Adversarial Networks (GANs) for electron microscopy datasets. We show how self-supervised pretraining facilitates efficient fine-tuning for a spectrum of downstream tasks, including semantic segmentation, denoising, noise \& background removal, and super-resolution. Experimentation with varying model complexities and receptive field sizes reveals the remarkable phenomenon that fine-tuned models of lower complexity consistently outperform more complex models with random weight initialization. We demonstrate the versatility of self-supervised pretraining across various downstream tasks in the context of electron microscopy, allowing faster convergence and better performance. We conclude that self-supervised pretraining serves as a powerful catalyst, being especially advantageous when limited annotated data are available and efficient scaling of computational cost is important.
[{'version': 'v1', 'created': 'Wed, 28 Feb 2024 12:25:01 GMT'}, {'version': 'v2', 'created': 'Thu, 18 Jul 2024 09:58:03 GMT'}]
2024-07-19
Dongchen Huang, Junde Liu, Tian Qian, and Hongming Weng
Training-set-free two-stage deep learning for spectroscopic data de-noising
null
null
null
cond-mat.mtrl-sci cs.LG physics.data-an
De-noising is a prominent step in the spectra post-processing procedure. Previous machine learning-based methods are fast but mostly based on supervised learning and require a training set that may be typically expensive in real experimental measurements. Unsupervised learning-based algorithms are slow and require many iterations to achieve convergence. Here, we bridge this gap by proposing a training-set-free two-stage deep learning method. We show that the fuzzy fixed input in previous methods can be improved by introducing an adaptive prior. Combined with more advanced optimization techniques, our approach can achieve five times acceleration compared to previous work. Theoretically, we study the landscape of a corresponding non-convex linear problem, and our results indicates that this problem has benign geometry for first-order algorithms to converge.
[{'version': 'v1', 'created': 'Thu, 29 Feb 2024 03:31:41 GMT'}, {'version': 'v2', 'created': 'Tue, 5 Mar 2024 12:39:23 GMT'}]
2024-03-06
Fankai Xie, Tenglong Lu, Sheng Meng, Miao Liu
GPTFF: A high-accuracy out-of-the-box universal AI force field for arbitrary inorganic materials
Science Bulletin, 10.1016/j.scib.2024.08.039
10.1016/j.scib.2024.08.039
null
cond-mat.mtrl-sci
This study introduces a novel AI force field, namely graph-based pre-trained transformer force field (GPTFF), which can simulate arbitrary inorganic systems with good precision and generalizability. Harnessing a large trove of the data and the attention mechanism of transformer algorithms, the model can accurately predict energy, atomic forces, and stress with Mean Absolute Error (MAE) values of 32 meV/atom, 71 meV/{\AA}, and 0.365 GPa, respectively. The dataset used to train the model includes 37.8 million single-point energies, 11.7 billion force pairs, and 340.2 million stresses. We also demonstrated that GPTFF can be universally used to simulate various physical systems, such as crystal structure optimization, phase transition simulations, and mass transport.
[{'version': 'v1', 'created': 'Thu, 29 Feb 2024 16:30:07 GMT'}]
2024-09-04
Vahe Gharakhanyan, Luke J. Wirth, Jose A. Garrido Torres, Ethan Eisenberg, Ting Wang, Dallas R. Trinkle, Snigdhansu Chatterjee and Alexander Urban
Discovering Melting Temperature Prediction Models of Inorganic Solids by Combining Supervised and Unsupervised Learning
null
null
null
cond-mat.mtrl-sci
The melting temperature is important for materials design because of its relationship with thermal stability, synthesis, and processing conditions. Current empirical and computational melting point estimation techniques are limited in scope, computational feasibility, or interpretability. We report the development of a machine learning methodology for predicting melting temperatures of binary ionic solid materials. We evaluated different machine-learning models trained on a data set of the melting points of 476 non-metallic crystalline binary compounds, using materials embeddings constructed from elemental properties and density-functional theory calculations as model inputs. A direct supervised-learning approach yields a mean absolute error of around 180~K but suffers from low interpretability. We find that the fidelity of predictions can further be improved by introducing an additional unsupervised-learning step that first classifies the materials before the melting-point regression. Not only does this two-step model exhibit improved accuracy, but the approach also provides a level of interpretability with insights into feature importance and different types of melting that depend on the specific atomic bonding inside a material. Motivated by this finding, we used a symbolic learning approach to find interpretable physical models for the melting temperature, which recovered the best-performing features from both prior models and provided additional interpretability.
[{'version': 'v1', 'created': 'Tue, 5 Mar 2024 16:23:37 GMT'}]
2024-03-06
Yingjie Zhao and Hongbo Zhou and Zian Zhang and Zhenxing Bo and Baoan Sun and Minqiang Jiang and Zhiping Xu
Discovering High-Strength Alloys via Physics-Transfer Learning
null
null
null
cond-mat.mtrl-sci cs.LG physics.comp-ph
Predicting the strength of materials requires considering various length and time scales, striking a balance between accuracy and efficiency. Peierls stress measures material strength by evaluating dislocation resistance to plastic flow, reliant on elastic lattice responses and crystal slip energy landscape. Computational challenges due to the non-local and non-equilibrium nature of dislocations prohibit Peierls stress evaluation from state-of-the-art material databases. We propose a data-driven framework that leverages neural networks trained on force field simulations to understand crystal plasticity physics, predicting Peierls stress from material parameters derived via density functional theory computations, which are otherwise computationally intensive for direct dislocation modeling. This physics transfer approach successfully screen the strength of metallic alloys from a limited number of single-point calculations with chemical accuracy. Guided by these predictions, we fabricate high-strength binary alloys previously unexplored, utilizing high-throughput ion beam deposition techniques. The framework extends to problems facing the accuracy-performance dilemma in general by harnessing the hierarchy of physics of multiscale models in materials sciences.
[{'version': 'v1', 'created': 'Tue, 12 Mar 2024 11:05:05 GMT'}, {'version': 'v2', 'created': 'Sun, 26 Jan 2025 07:32:07 GMT'}]
2025-01-28
Matteo Masto, Vincent Favre-Nicolin, Steven Leake, Tobias Sch\"ulli, Marie-Ingrid Richard, Ewen Bellec
Patching-based Deep Learning model for the Inpainting of Bragg Coherent Diffraction patterns affected by detectors' gaps
null
null
null
cond-mat.mtrl-sci
We propose a deep learning algorithm for the inpainting of Bragg Coherent Diffraction Imaging (BCDI) patterns affected by detector gaps. These regions of missing intensity can compromise the accuracy of reconstruction algorithms, inducing artifacts in the final result. It is thus desirable to restore the intensity in these regions in order to ensure more reliable reconstructions. The key aspect of our method lies in the choice of training the neural network with cropped sections of both experimental diffraction data and simulated data and subsequently patching the predictions generated by the model along the gap, thus completing the full diffraction peak. This provides us with more experimental training data and allows for a faster model training due to the limited size, while the neural network can be applied to arbitrarily larger BCDI datasets. Moreover, our method not only broadens the scope of application but also ensures the preservation of data integrity and reliability in the face of challenging experimental conditions.
[{'version': 'v1', 'created': 'Wed, 13 Mar 2024 15:03:13 GMT'}]
2024-03-14
Zhiqiang Zhao, Wanlin Guo, and Zhuhua Zhang
A general-purpose neural network potential for Ti-Al-Nb alloys towards large-scale molecular dynamics with ab initio accuracy
null
null
null
cond-mat.mtrl-sci physics.comp-ph
High Nb-containing TiAl alloys exhibit exceptional high-temperature strength and room-temperature ductility, making them widely used in hot-section components of automotive and aerospace engines. However, the lack of accurate interatomic interaction potentials for large-scale modeling severely hampers a comprehensive understanding of the failure mechanism of Ti-Al-Nb alloys and the development of strategies to enhance the mechanical properties. Here, we develop a general-purpose machine-learned potential (MLP) for the Ti-Al-Nb ternary system by combining the neural evolution potentials framework with an active learning scheme. The developed MLP, trained on extensive first-principles datasets, demonstrates remarkable accuracy in predicting various lattice and defect properties, as well as high-temperature characteristics such as thermal expansion and melting point for TiAl systems. Notably, this potential can effectively describe the key effect of Nb doping on stacking fault energies and formation energies. Of practical importance is that our MLP enables large-scale molecular dynamics simulations involving tens of millions of atoms with ab initio accuracy, achieving an outstanding balance between computational speed and accuracy. These results pave the way for studying micro-mechanical behaviors in TiAl lamellar structures and developing high-performance TiAl alloys towards applications at elevated temperatures.
[{'version': 'v1', 'created': 'Thu, 14 Mar 2024 16:11:14 GMT'}]
2024-03-15
Vincent Bl\"umer, Celal Soyarslan, Ton van den Boogaard
Generative reconstruction of 3D volume elements for Ti-6Al-4V basketweave microstructure by optimization of CNN-based microstructural descriptors
null
null
null
cond-mat.mtrl-sci
We present a methodology for the generative reconstruction of 3D Volume Elements (VE) for numerical multiscale analysis of Ti-6Al-4V processed by Additive Manufacturing (AM). The basketweave morphology, which is typically dominant in AM-processed Ti-6Al-4V, is analyzed in conventional Electron Backscatter Diffusion (EBSD) micrographs. Prior \b{eta}-grain reconstruction is performed to obtain the out-of-plane orientation of the observed grains leveraging Burgers orientation relationship. Convolutional Neural Network (CNN) - based microstructure descriptors are extracted from the 2D data, and used for cross-section-based optimization of pixel values on orthogonal planes in 3D, using the Microstructure Characterization and Reconstruction (MCR) implementation MCRpy [16]. In order to utilize MCRpy, which performs best for binary systems, the basketweave microstructure, which consists of up to twelve distinct grain orientations, is decomposed into several separate two-phase systems. Our reconstructions capture key characteristics of the titanium basketweave morphology and show qualitative resemblance to experimentally obtained 3D data. The preservation of volume fraction during assembly of the reconstruction remains an unadressed challenge at this stage.
[{'version': 'v1', 'created': 'Thu, 14 Mar 2024 17:50:24 GMT'}]
2024-03-15
Ryo Murakami, Taisuke T. Sasaki, Hideki Yoshikawa, Yoshitaka Matsushita, Keitaro Sodeyama, Tadakatsu Ohkubo, Hiroshi Shinotsuka, Kenji Nagata
Rapid and Robust construction of an ML-ready peak feature table from X-ray diffraction data using Bayesian peak-top fitting
null
null
null
cond-mat.mtrl-sci stat.AP
To advance the development of materials through data-driven scientific methods, appropriate methods for building machine learning (ML)-ready feature tables from measured and computed data must be established. In materials development, X-ray diffraction (XRD) is an effective technique for analysing crystal structures and other microstructural features that have information that can explain material properties. Therefore, the fully automated extraction of peak features from XRD data without the bias of an analyst is a significant challenge. This study aimed to establish an efficient and robust approach for constructing peak feature tables that follow ML standards (ML-ready) from XRD data. We challenge peak feature extraction in the situation where only the peak function profile is known a priori, without knowledge of the measurement material or crystal structure factor. We utilized Bayesian estimation to extract peak features from XRD data and subsequently performed Bayesian regression analysis with feature selection to predict the material property. The proposed method focused only on the tops of peaks within localized regions of interest (ROIs) and extracted peak features quickly and accurately. This process facilitated the rapid extracting of major peak features from the XRD data and the construction of an ML-ready feature table. We then applied Bayesian linear regression to the maximum energy product $(BH)_{max}$, using the extracted peak features as the explanatory variable. The outcomes yielded reasonable and robust regression results. Thus, the findings of this study indicated that \textit{004} peak height and area were important features for predicting $(BH)_{max}$.
[{'version': 'v1', 'created': 'Wed, 7 Feb 2024 01:24:39 GMT'}]
2024-03-18
Xiaoshan Luo, Zhenyu Wang, Pengyue Gao, Jian Lv, Yanchao Wang, Changfeng Chen and Yanming Ma
Deep learning generative model for crystal structure prediction
npj Comput. Mater., 10, 254 (2024)
10.1038/s41524-024-01443-y
null
cond-mat.mtrl-sci physics.comp-ph
Recent advances in deep learning generative models (GMs) have created high capabilities in accessing and assessing complex high-dimensional data, allowing superior efficiency in navigating vast material configuration space in search of viable structures. Coupling such capabilities with physically significant data to construct trained models for materials discovery is crucial to moving this emerging field forward. Here, we present a universal GM for crystal structure prediction (CSP) via a conditional crystal diffusion variational autoencoder (Cond-CDVAE) approach, which is tailored to allow user-defined material and physical parameters such as composition and pressure. This model is trained on an expansive dataset containing over 670,000 local minimum structures, including a rich spectrum of high-pressure structures, along with ambient-pressure structures in Materials Project database. We demonstrate that the Cond-CDVAE model can generate physically plausible structures with high fidelity under diverse pressure conditions without necessitating local optimization, accurately predicting 59.3% of the 3,547 unseen ambient-pressure experimental structures within 800 structure samplings, with the accuracy rate climbing to 83.2% for structures comprising fewer than 20 atoms per unit cell. These results meet or exceed those achieved via conventional CSP methods based on global optimization. The present findings showcase substantial potential of GMs in the realm of CSP.
[{'version': 'v1', 'created': 'Sat, 16 Mar 2024 07:54:19 GMT'}, {'version': 'v2', 'created': 'Sat, 10 Aug 2024 07:02:27 GMT'}]
2024-11-13
An Chen, Zhilong Wang, Karl Luigi Loza Vidaurre, Yanqiang Han, Simin Ye, Kehao Tao, Shiwei Wang, Jing Gao, and Jinjin Li
Knowledge-Reuse Transfer Learning Methods in Molecular and Material Science
null
null
null
cond-mat.mtrl-sci cs.LG physics.chem-ph
Molecules and materials are the foundation for the development of modern advanced industries such as energy storage systems and semiconductor devices. However, traditional trial-and-error methods or theoretical calculations are highly resource-intensive, and extremely long R&D (Research and Development) periods cannot meet the urgent need for molecules/materials in industrial development. Machine learning (ML) methods based on big data are expected to break this dilemma. However, the difficulty in constructing large-scale datasets of new molecules/materials due to the high cost of data acquisition and annotation limits the development of machine learning. The application of transfer learning lowers the data requirements for model training, which makes transfer learning stand out in researches addressing data quality issues. In this review, we summarize recent advances in transfer learning related to molecular and materials science. We focus on the application of transfer learning methods for the discovery of advanced molecules/materials, particularly, the construction of transfer learning frameworks for different systems, and how transfer learning can enhance the performance of models. In addition, the challenges of transfer learning are also discussed.
[{'version': 'v1', 'created': 'Sat, 2 Mar 2024 12:41:25 GMT'}]
2024-03-21
Yubo Qi, Weiyi Gong, Qimin Yan
Bridging deep learning force fields and electronic structures with a physics-informed approach
null
null
null
cond-mat.mtrl-sci
This work presents a physics-informed neural network approach bridging deep-learning force field and electronic structure simulations, illustrated through twisted two-dimensional large-scale material systems. The deep potential molecular dynamics model is adopted as the backbone, and electronic structure simulation is integrated. Using Wannier functions as the basis, we categorize Wannier Hamiltonian elements based on physical principles to incorporate diverse information from a deep-learning force field model. This information-sharing mechanism streamlines the architecture of our multifunctional model, enhancing its efficiency and effectiveness. Utilizing Wannier functions as the basis lays the groundwork for predicting more physical quantities. This approach serves as a powerful tool to explore both the structural and electronic properties of large-scale systems characterized by low periodicities. By endowing an existing well-developed machine-learning force field with electronic structure simulation capabilities, the study marks a significant advancement in developing multimodal machine-learning-based computational methods that can achieve multiple functionalities traditionally exclusive to first-principles calculations.
[{'version': 'v1', 'created': 'Wed, 20 Mar 2024 15:33:46 GMT'}, {'version': 'v2', 'created': 'Mon, 1 Apr 2024 03:28:47 GMT'}]
2024-04-02
Orlando A. Mendible, Jonathan K. Whitmer, and Yamil J. Col\'on
Considerations in the use of ML interaction potentials for free energy calculations
null
10.1063/5.0252043
null
physics.chem-ph cond-mat.mtrl-sci cs.LG
Machine learning force fields (MLFFs) promise to accurately describe the potential energy surface of molecules at the ab initio level of theory with improved computational efficiency. Within MLFFs, equivariant graph neural networks (EQNNs) have shown great promise in accuracy and performance and are the focus of this work. The capability of EQNNs to recover free energy surfaces (FES) remains to be thoroughly investigated. In this work, we investigate the impact of collective variables (CVs) distribution within the training data on the accuracy of EQNNs predicting the FES of butane and alanine dipeptide (ADP). A generalizable workflow is presented in which training configurations are generated with classical molecular dynamics simulations, and energies and forces are obtained with ab initio calculations. We evaluate how bond and angle constraints in the training data influence the accuracy of EQNN force fields in reproducing the FES of the molecules at both classical and ab initio levels of theory. Results indicate that the model's accuracy is unaffected by the distribution of sampled CVs during training, given that the training data includes configurations from characteristic regions of the system's FES. However, when the training data is obtained from classical simulations, the EQNN struggles to extrapolate the free energy for configurations with high free energy. In contrast, models trained with the same configurations on ab initio data show improved extrapolation accuracy. The findings underscore the difficulties in creating a comprehensive training dataset for EQNNs to predict FESs and highlight the importance of prior knowledge of the system's FES.
[{'version': 'v1', 'created': 'Wed, 20 Mar 2024 19:49:21 GMT'}, {'version': 'v2', 'created': 'Tue, 13 May 2025 13:22:54 GMT'}, {'version': 'v3', 'created': 'Wed, 14 May 2025 14:50:01 GMT'}]
2025-05-15
Brian H. Lee, James P. Larentzos, John K. Brennan, and Alejandro Strachan
Graph neural network coarse-grain force field for the molecular crystal RDX
null
null
null
cond-mat.mes-hall cond-mat.mtrl-sci
Condense phase molecular systems organize in wide range of distinct molecular configurations, including amorphous melt and glass as well as crystals often exhibiting polymorphism, that originate from their intricate intra- and intermolecular forces. While accurate coarse-grain (CG) models for these materials are critical to understand phenomena beyond the reach of all-atom simulations, current models cannot capture the diversity of molecular structures. We introduce a generally applicable approach to develop CG force fields for molecular crystals combining graph neural networks (GNN) and data from an all-atom simulations and apply it to the high-energy density material RDX. We address the challenge of expanding the training data with relevant configurations via an iterative procedure that performs CG molecular dynamics of processes of interest and reconstructs the atomistic configurations using a pre-trained neural network decoder. The multi-site CG model uses a GNN architecture constructed to satisfy translational invariance and rotational covariance for forces. The resulting model captures both crystalline and amorphous states for a wide range of temperatures and densities.
[{'version': 'v1', 'created': 'Fri, 22 Mar 2024 15:06:06 GMT'}]
2024-03-25
Zhendong Cao, Xiaoshan Luo, Jian Lv and Lei Wang
Space Group Informed Transformer for Crystalline Materials Generation
null
null
null
cond-mat.mtrl-sci cs.LG physics.comp-ph
We introduce CrystalFormer, a transformer-based autoregressive model specifically designed for space group-controlled generation of crystalline materials. The incorporation of space group symmetry significantly simplifies the crystal space, which is crucial for data and compute efficient generative modeling of crystalline materials. Leveraging the prominent discrete and sequential nature of the Wyckoff positions, CrystalFormer learns to generate crystals by directly predicting the species and locations of symmetry-inequivalent atoms in the unit cell. We demonstrate the advantages of CrystalFormer in standard tasks such as symmetric structure initialization and element substitution compared to conventional methods implemented in popular crystal structure prediction software. Moreover, we showcase the application of CrystalFormer of property-guided materials design in a plug-and-play manner. Our analysis shows that CrystalFormer ingests sensible solid-state chemistry knowledge and heuristics by compressing the material dataset, thus enabling systematic exploration of crystalline materials. The simplicity, generality, and flexibility of CrystalFormer position it as a promising architecture to be the foundational model of the entire crystalline materials space, heralding a new era in materials modeling and discovery.
[{'version': 'v1', 'created': 'Sat, 23 Mar 2024 06:01:45 GMT'}, {'version': 'v2', 'created': 'Fri, 16 Aug 2024 02:57:35 GMT'}]
2024-08-19
Xiang Huang and Shenghong Ju
Tutorial: AI-assisted exploration and active design of polymers with high intrinsic thermal conductivity
Journal of Applied Physics 135, 171101, 2024
10.1063/5.0201522
null
cond-mat.soft cond-mat.mtrl-sci physics.app-ph physics.chem-ph physics.comp-ph
Designing polymers with high intrinsic thermal conductivity (TC) is critically important for the thermal management of organic electronics and photonics. However, this is a challenging task owing to the diversity of the chemical space and the barriers to advanced synthetic experiments/characterization techniques for polymers. In this Tutorial, the fundamentals and implementation of combining classical molecular dynamics simulation and machine learning (ML) for the development of polymers with high TC are comprehensively introduced. We begin by describing the core components of a universal ML framework, involving polymer datasets, property calculators, feature engineering and informatics algorithms. Then, the process of constructing interpretable regression algorithms for TC prediction is introduced, aiming to extract the underlying relationships between microstructures and TCs for polymers. We also explore the design of sequence-ordered polymers with high TC using lightweight and mainstream active learning algorithms. Lastly, we conclude by addressing the current limitations and suggesting potential avenues for future research on this topic.
[{'version': 'v1', 'created': 'Sat, 23 Mar 2024 16:52:56 GMT'}]
2024-05-09
Yuqi Song, Rongzhi Dong, Lai Wei, Qin Li, Jianjun Hu
AlphaCrystal-II: Distance matrix based crystal structure prediction using deep learning
null
null
null
cond-mat.mtrl-sci cs.LG
Computational prediction of stable crystal structures has a profound impact on the large-scale discovery of novel functional materials. However, predicting the crystal structure solely from a material's composition or formula is a promising yet challenging task, as traditional ab initio crystal structure prediction (CSP) methods rely on time-consuming global searches and first-principles free energy calculations. Inspired by the recent success of deep learning approaches in protein structure prediction, which utilize pairwise amino acid interactions to describe 3D structures, we present AlphaCrystal-II, a novel knowledge-based solution that exploits the abundant inter-atomic interaction patterns found in existing known crystal structures. AlphaCrystal-II predicts the atomic distance matrix of a target crystal material and employs this matrix to reconstruct its 3D crystal structure. By leveraging the wealth of inter-atomic relationships of known crystal structures, our approach demonstrates remarkable effectiveness and reliability in structure prediction through comprehensive experiments. This work highlights the potential of data-driven methods in accelerating the discovery and design of new materials with tailored properties.
[{'version': 'v1', 'created': 'Sun, 7 Apr 2024 05:17:43 GMT'}]
2024-04-09
Tomoya Shiota, Kenji Ishihara, Wataru Mizukami
Lowering the Exponential Wall: Accelerating High-Entropy Alloy Catalysts Screening using Local Surface Energy Descriptors from Neural Network Potentials
null
null
null
quant-ph cond-mat.mtrl-sci
Computational screening is indispensable for the efficient design of high-entropy alloys (HEAs), which hold considerable potential for catalytic applications. However, the chemical space of HEAs is exponentially vast with respect to the number of constituent elements, making even machine learning-based screening calculations time-intensive. To address this challenge, we propose a rapid method for predicting HEA properties using data from monometallic systems (or few-component alloys). Central to our approach is the newly introduced local surface energy (LSE) descriptor, which captures local surface reactivity at atomic resolution. We established a correlation between LSE and adsorption energies using monometallic systems. Using this correlation in a linear regression model, we successfully estimated molecular adsorption energies on HEAs with significantly higher accuracy than a conventional descriptor (i.e., generalized coordination numbers). Furthermore, we developed high-precision models by employing both classical and quantum machine learning. Our method enabled CO adsorption-energy calculations for 1000 quinary nanoparticles, comprising 201 atoms each, within a few days, considerably faster than density functional theory, which would require hundreds of years or neural network potentials, which would have taken hundreds of days. The proposed approach accelerates the exploration of the vast HEA chemical space, facilitating the design of novel catalysts.
[{'version': 'v1', 'created': 'Fri, 12 Apr 2024 11:54:06 GMT'}, {'version': 'v2', 'created': 'Sun, 6 Oct 2024 10:28:27 GMT'}, {'version': 'v3', 'created': 'Mon, 27 Jan 2025 08:54:38 GMT'}]
2025-01-28
Zhuo Diao, Keiichi Ueda, Linfeng Hou, Fengxuan Li, Hayato Yamashita, Masayuki Abe
AI-equipped scanning probe microscopy for autonomous site-specific atomic-level characterization at room temperature
null
null
null
physics.comp-ph cond-mat.mtrl-sci
We present an advanced scanning probe microscopy system enhanced with artificial intelligence (AI-SPM) designed for self-driving atomic-scale measurements. This system expertly identifies and manipulates atomic positions with high precision, autonomously performing tasks such as spectroscopic data acquisition and atomic adjustment. An outstanding feature of AI-SPM is its ability to detect and adapt to surface defects, targeting or avoiding them as necessary. It's also engineered to address typical challenges such as positional drift and tip apex atomic variations due to the thermal effect, ensuring accurate, site-specific surface analyses. Our tests under the demanding conditions of room temperature have demonstrated the robustness of the system, successfully navigating thermal drift and tip fluctuations. During these tests on the Si(111)-(7x7) surface, AI-SPM autonomously identified defect-free regions and performed a large number of current-voltage spectroscopy measurements at different adatom sites, while autonomously compensating for thermal drift and monitoring probe health. These experiments produce extensive data sets that are critical for reliable materials characterization and demonstrate the potential of AI-SPM to significantly improve data acquisition. The integration of AI into SPM technologies represents a step toward more effective, precise and reliable atomic-level surface analysis, revolutionizing materials characterization methods.
[{'version': 'v1', 'created': 'Wed, 17 Apr 2024 08:25:42 GMT'}]
2024-04-18
Shinnosuke Hattori and Qiang Zhu
Study of Entropy-Driven Polymorphic Stability for Aspirin Using Accurate Neural Network Interatomic Potential
null
null
null
cond-mat.mtrl-sci
In this study, we present a systematic computational investigation to analyze the long debated crystal stability of two well known aspirin polymorphs, labeled as Form I and Form II. Specifically, we developed a strategy to collect training configurations covering diverse interatomic interactions between representative functional groups in the aspirin crystals. Utilizing a state-of-the-art neural network interatomic potential (NNIP) model, we developed an accurate machine learning potential to simulate aspirin crystal dynamics under finite temperature conditions with $\sim$0.46 kJ/mol/molecule accuracy. Employing the trained NNIP model, we performed thermodynamic integration to assess the free energy difference between aspirin Forms I and II, accounting for the anharmonic effects in a large supercell consisting of 512 molecules. For the first time, our results convincingly demonstrated that Form I is more stable than Form II at 300 K, ranging from 0.74 to 1.83 kJ/mol/molecule, aligning with the experimental observations. Unlike the majority of previous simulations based on (quasi)harmonic approximations in a small super cell, which often found the degenerate energies between aspirin I and II, our findings underscore the importance of anharmonic effects in determining polymorphic stability ranking. Furthermore, we proposed the use of rotational degrees of freedom of methyl and ester/phenyl groups in the aspirin crystal, as characteristic motions to highlight rotational entropic contribution that favors the stability of Form I. Beyond the aspirin polymorphism, we anticipate that such entropy-driven stabilization can be broadly applicable to many other organic systems and thus our approach, suggesting our approach holds a great promise for stability studies in small molecule drug design.
[{'version': 'v1', 'created': 'Wed, 17 Apr 2024 17:34:52 GMT'}, {'version': 'v2', 'created': 'Fri, 19 Apr 2024 16:12:58 GMT'}]
2024-04-22
Adva Baratz, Galit Cohen, Sivan Refaely-Abramson
Unsupervised learning approach to quantum wavepacket dynamics from coupled temporal-spatial correlations
null
null
null
cond-mat.mtrl-sci
Understanding complex quantum dynamics in realistic materials requires insight into the underlying correlations dominating the interactions between the participating particles. Due to the wealth of information involved in these processes, applying artificial intelligence methods is compelling. Yet, unsupervised data-driven approaches typically focus on maximal variations of the individual components, rather than considering the correlations between them. Here we present an approach that recognizes correlation patterns to explore convoluted dynamical processes. Our scheme is using singular value decomposition (SVD) to extract dynamical features, unveiling the internal temporal-spatial interrelations that generate the dynamical mechanisms. We apply our approach to study light-induced wavepacket propagation in organic crystals, of interest for applications in material based quantum computing and quantum information science. We show how transformation from the input momentum and time coordinates onto a new correlation-induced coordinate space allows direct recognition of the relaxation and dephasing components dominating the dynamics and demonstrate their dependence on the initial pulse shape. Entanglement of the dynamical features is suggested as a pathway to reproduce the information required for further explainability of these mechanisms. Our method offers a route for elucidating complex dynamical processes using unsupervised AI-based analysis in multi-component systems.
[{'version': 'v1', 'created': 'Thu, 18 Apr 2024 08:20:30 GMT'}]
2024-04-19
Wonseok Lee, Yeonghun Kang, Taeun Bae, Jihan Kim
Harnessing Large Language Model to collect and analyze Metal-organic framework property dataset
null
null
null
cond-mat.mtrl-sci
This research was focused on the efficient collection of experimental Metal-Organic Framework (MOF) data from scientific literature to address the challenges of accessing hard-to-find data and improving the quality of information available for machine learning studies in materials science. Utilizing a chain of advanced Large Language Models (LLMs), we developed a systematic approach to extract and organize MOF data into a structured format. Our methodology successfully compiled information from more than 40,000 research articles, creating a comprehensive and ready-to-use dataset. The findings highlight the significant advantage of incorporating experimental data over relying solely on simulated data for enhancing the accuracy of machine learning predictions in the field of MOF research.
[{'version': 'v1', 'created': 'Sun, 31 Mar 2024 12:47:24 GMT'}]
2024-04-23
Bowen Hou, Jinyuan Wu, Diana Y. Qiu
Unsupervised Learning of Individual Kohn-Sham States: Interpretable Representations and Consequences for Downstream Predictions of Many-Body Effects
null
null
null
cond-mat.mtrl-sci physics.comp-ph
Representation learning for the electronic structure problem is a major challenge of machine learning in computational condensed matter and materials physics. Within quantum mechanical first principles approaches, Kohn-Sham density functional theory (DFT) is the preeminent tool for understanding electronic structure, and the high-dimensional wavefunctions calculated in this approach serve as the building block for downstream calculations of correlated many-body excitations and related physical observables. Here, we use variational autoencoders (VAE) for the unsupervised learning of high-dimensional DFT wavefunctions and show that these wavefunctions lie in a low-dimensional manifold within the latent space. Our model autonomously determines the optimal representation of the electronic structure, avoiding limitations due to manual feature engineering and selection in prior work. To demonstrate the utility of the latent space representation of the DFT wavefunction, we use it for the supervised training of neural networks (NN) for downstream prediction of the quasiparticle bandstructures within the GW formalism, which includes many-electron correlations beyond DFT. The GW prediction achieves a low error of 0.11 eV for a combined test set of metals and semiconductors drawn from the Computational 2D Materials Database (C2DB), suggesting that latent space representation captures key physical information from the original data. Finally, we explore the interpretability of the VAE representation and show that the successful representation learning and downstream prediction by our model is derived from the smoothness of the VAE latent space, which also enables the generation of wavefunctions on arbitrary points in latent space. Our work provides a novel and general machine-learning framework for investigating electronic structure and many-body physics.
[{'version': 'v1', 'created': 'Mon, 22 Apr 2024 21:50:50 GMT'}]
2024-04-24
Rajni Chahal, Michael D. Toomey, Logan T. Kearney, Ada Sedova, Joshua T. Damron, Amit K. Naskar, Santanu Roy
Deep Learning Interatomic Potential Connects Molecular Structural Ordering to Macroscale Properties of Polyacrylonitrile (PAN) Polymer
null
null
null
cond-mat.mtrl-sci
Polyacrylonitrile (PAN) is an important commercial polymer, bearing atactic stereochemistry resulting from nonselective radical polymerization. As such, an accurate, fundamental understanding of governing interactions among PAN molecular units are indispensable to advance the design principles of final products at reduced processability costs. While ab initio molecular dynamics (AIMD) simulations can provide the necessary accuracy for treating key interactions in polar polymers such as dipole-dipole interactions and hydrogen bonding, and analyzing their influence on molecular orientation, their implementation is limited to small molecules only. Herein, we show that the neural network interatomic potentials (NNIP) that are trained on the small-scale AIMD data (acquired for oligomers) can be efficiently employed to examine the structures/properties at large scales (polymers). NNIP provides critical insight into intra- and interchain hydrogen bonding and dipolar correlations, and accurately predicts the amorphous bulk PAN structure validated by modeling the experimental X-ray structure factor. Furthermore, the NNIP-predicted PAN properties such as density and elastic modulus are in good agreement with their experimental values. Overall, the trend in the elastic modulus is found to correlate strongly with the PAN structural orientations encoded in Hermans orientation factor. This study enables the ability to predict the structure-property relations for PAN and analogs with sustainable ab initio accuracy across scales.
[{'version': 'v1', 'created': 'Wed, 24 Apr 2024 20:21:54 GMT'}]
2024-04-26
Jiwei Yu, Zhangwei Wang, Aparna Saksena, Shaolou Wei, Ye Wei, Timoteo Colnaghi, Andreas Marek, Markus Rampp, Min Song, Baptiste Gault, Yue Li
3D deep learning for enhanced atom probe tomography analysis of nanoscale microstructures
null
null
null
cond-mat.mtrl-sci physics.data-an
Quantitative analysis of microstructural features on the nanoscale, including precipitates, local chemical orderings (LCOs) or structural defects (e.g. stacking faults) plays a pivotal role in understanding the mechanical and physical responses of engineering materials. Atom probe tomography (APT), known for its exceptional combination of chemical sensitivity and sub-nanometer resolution, primarily identifies microstructures through compositional segregations. However, this fails when there is no significant segregation, as can be the case for LCOs and stacking faults. Here, we introduce a 3D deep learning approach, AtomNet, designed to process APT point cloud data at the single-atom level for nanoscale microstructure extraction, simultaneously considering compositional and structural information. AtomNet is showcased in segmenting L12-type nanoprecipitates from the matrix in an AlLiMg alloy, irrespective of crystallographic orientations, which outperforms previous methods. AtomNet also allows for 3D imaging of L10-type LCOs in an AuCu alloy, a challenging task for conventional analysis due to their small size and subtle compositional differences. Finally, we demonstrate the use of AtomNet for revealing 2D stacking faults in a Co-based superalloy, without any defected training data, expanding the capabilities of APT for automated exploration of hidden microstructures. AtomNet pushes the boundaries of APT analysis, and holds promise in establishing precise quantitative microstructure-property relationships across a diverse range of metallic materials.
[{'version': 'v1', 'created': 'Thu, 25 Apr 2024 11:36:10 GMT'}]
2024-04-26
M. A. Maia, I. B. C. M. Rocha, D. Kova\v{c}evi\'c, F. P. van der Meer
Physically recurrent neural network for rate and path-dependent heterogeneous materials in a finite strain framework
null
null
null
cond-mat.mtrl-sci cs.LG cs.NA math.NA
In this work, a hybrid physics-based data-driven surrogate model for the microscale analysis of heterogeneous material is investigated. The proposed model benefits from the physics-based knowledge contained in the constitutive models used in the full-order micromodel by embedding them in a neural network. Following previous developments, this paper extends the applicability of the physically recurrent neural network (PRNN) by introducing an architecture suitable for rate-dependent materials in a finite strain framework. In this model, the homogenized deformation gradient of the micromodel is encoded into a set of deformation gradients serving as input to the embedded constitutive models. These constitutive models compute stresses, which are combined in a decoder to predict the homogenized stress, such that the internal variables of the history-dependent constitutive models naturally provide physics-based memory for the network. To demonstrate the capabilities of the surrogate model, we consider a unidirectional composite micromodel with transversely isotropic elastic fibers and elasto-viscoplastic matrix material. The extrapolation properties of the surrogate model trained to replace such micromodel are tested on loading scenarios unseen during training, ranging from different strain-rates to cyclic loading and relaxation. Speed-ups of three orders of magnitude with respect to the runtime of the original micromodel are obtained.
[{'version': 'v1', 'created': 'Fri, 5 Apr 2024 12:40:03 GMT'}]
2024-04-30
Adela Habib and Joshua Finkelstein and Anders M. N. Niklasson
Efficient Mixed-Precision Matrix Factorization of the Inverse Overlap Matrix in Electronic Structure Calculations with AI-Hardware and GPUs
null
null
null
physics.comp-ph cond-mat.mtrl-sci math-ph math.MP
In recent years, a new kind of accelerated hardware has gained popularity in the Artificial Intelligence (AI) and Machine Learning (ML) communities which enables extremely high-performance tensor contractions in reduced precision for deep neural network calculations. In this article, we exploit Nvidia Tensor cores, a prototypical example of such AI/ML hardware, to develop a mixed precision approach for computing a dense matrix factorization of the inverse overlap matrix in electronic structure theory, $S^{-1}$. This factorization of $S^{-1}$, written as $ZZ^T=S^{-1}$, is used to transform the general matrix eigenvalue problem into a standard matrix eigenvalue problem. Here we present a mixed precision iterative refinement algorithm where $Z$ is given recursively using matrix-matrix multiplications and can be computed with high performance on Tensor cores. To understand the performance and accuracy of Tensor cores, comparisons are made to GPU-only implementations in single and double precision. Additionally, we propose a non-parametric stopping criteria which is robust in the face of lower precision floating point operations. The algorithm is particularly useful when we have a good initial guess to $Z$, for example, from previous time steps in quantum-mechanical molecular dynamics simulations or from a previous iteration in a geometry optimization.
[{'version': 'v1', 'created': 'Mon, 29 Apr 2024 23:53:16 GMT'}]
2024-05-01
Sungwoo Kang
How Graph Neural Network Interatomic Potentials Extrapolate: Role of the Message-Passing Algorithm
J. Chem. Phys. 161, 244102 (2024)
10.1063/5.0234287
null
cond-mat.mtrl-sci
Graph neural network interatomic potentials (GNN-IPs) are gaining significant attention due to their capability of learning from large datasets. Specifically, universal interatomic potentials based on GNN, usually trained with crystalline geometries, often exhibit remarkable extrapolative behavior towards untrained domains, such as surfaces or amorphous configurations. However, the origin of this extrapolation capability is not well understood. This work provides a theoretical explanation of how GNN-IPs extrapolate to untrained geometries. First, we demonstrate that GNN-IPs can capture non-local electrostatic interactions through the message-passing algorithm, as evidenced by tests on toy models and DFT data. We find that GNN-IP models, SevenNet and MACE, accurately predict electrostatic forces in untrained domains, indicating that they have learned the exact functional form of the Coulomb interaction. Based on these results, we suggest that the ability to learn non-local electrostatic interactions, coupled with the embedding nature of GNN-IPs, explains their extrapolation ability. We find that the universal GNN-IP, SevenNet-0, effectively infers non-local Coulomb interactions in untrained domains but fails to extrapolate the non-local forces arising from the kinetic term, which supports the suggested theory. Finally, we address the impact of hyperparameters on the extrapolation performance of universal potentials, such as SevenNet-0 and MACE-MP-0, and discuss the limitations of the extrapolation capabilities.
[{'version': 'v1', 'created': 'Wed, 1 May 2024 02:55:15 GMT'}, {'version': 'v2', 'created': 'Tue, 13 Aug 2024 13:50:55 GMT'}, {'version': 'v3', 'created': 'Thu, 5 Dec 2024 06:49:06 GMT'}]
2025-01-08
Jihua Chen, Yue Yuan, Amir Koushyar Ziabari, Xuan Xu, Honghai Zhang, Panagiotis Christakopoulos, Peter V. Bonnesen, Ilia N. Ivanov, Panchapakesan Ganesh, Chen Wang, Karen Patino Jaimes, Guang Yang, Rajeev Kumar, Bobby G. Sumpter, Rigoberto Advincula
AI for Manufacturing and Healthcare: a chemistry and engineering perspective
null
null
null
cond-mat.mtrl-sci
Artificial Intelligence (AI) approaches are increasingly being applied to more and more domains of Science, Engineering, Chemistry, and Industries to not only improve efficiencies and enhance productivity, but also enable new capabilities. The new opportunities range from automated molecule design and screening, properties prediction, gaining insights of chemical reactions, to computer-aided design, predictive maintenance of systems, robotics, and autonomous vehicles. This review focuses on the new applications of AI in manufacturing and healthcare. For the Manufacturing Industries, we focus on AI and algorithms for (1) Battery, (2) Flow Chemistry, (3) Additive Manufacturing, (4) Sensors, and (5) Machine Vision. For Healthcare applications, we focus on: (1) Medical Vision (2) Diagnosis, (3) Protein Design, and (4) Drug Discovery. In the end, related topics are discussed, including physics integrated machine learning, model explainability, security, and governance during model deployment.
[{'version': 'v1', 'created': 'Thu, 2 May 2024 17:50:05 GMT'}]
2024-05-03
Nakul Rampal, Kaiyu Wang, Matthew Burigana, Lingxiang Hou, Juri Al-Johani, Anna Sackmann, Hanan S. Murayshid, Walaa Abdullah Al-Sumari, Arwa M. Al-Abdulkarim, Nahla Eid Al-Hazmi, Majed O. Al-Awad, Christian Borgs, Jennifer T. Chayes, Omar M. Yaghi
Single and Multi-Hop Question-Answering Datasets for Reticular Chemistry with GPT-4-Turbo
null
null
null
cs.CL cond-mat.mtrl-sci
The rapid advancement in artificial intelligence and natural language processing has led to the development of large-scale datasets aimed at benchmarking the performance of machine learning models. Herein, we introduce 'RetChemQA,' a comprehensive benchmark dataset designed to evaluate the capabilities of such models in the domain of reticular chemistry. This dataset includes both single-hop and multi-hop question-answer pairs, encompassing approximately 45,000 Q&As for each type. The questions have been extracted from an extensive corpus of literature containing about 2,530 research papers from publishers including NAS, ACS, RSC, Elsevier, and Nature Publishing Group, among others. The dataset has been generated using OpenAI's GPT-4 Turbo, a cutting-edge model known for its exceptional language understanding and generation capabilities. In addition to the Q&A dataset, we also release a dataset of synthesis conditions extracted from the corpus of literature used in this study. The aim of RetChemQA is to provide a robust platform for the development and evaluation of advanced machine learning algorithms, particularly for the reticular chemistry community. The dataset is structured to reflect the complexities and nuances of real-world scientific discourse, thereby enabling nuanced performance assessments across a variety of tasks. The dataset is available at the following link: https://github.com/nakulrampal/RetChemQA
[{'version': 'v1', 'created': 'Fri, 3 May 2024 14:29:54 GMT'}]
2024-05-06
Luis Mart\'in Encinar, Daniele Lanzoni, Andrea Fantasia, Fabrizio Rovaris, Roberto Bergamaschini, Francesco Montalenti
Quantitative analysis of the prediction performance of a Convolutional Neural Network evaluating the surface elastic energy of a strained film
null
10.1016/j.commatsci.2024.113657
null
physics.comp-ph cond-mat.mtrl-sci
A Deep Learning approach is devised to estimate the elastic energy density $\rho$ at the free surface of an undulated stressed film. About 190000 arbitrary surface profiles h(x) are randomly generated by Perlin noise and paired with the corresponding elastic energy density profiles $\rho(x)$, computed by a semi-analytical Green's function approximation, suitable for small-slope morphologies. The resulting dataset and smaller subsets of it are used for the training of a Fully Convolutional Neural Network. The trained models are shown to return quantitative predictions of $\rho$, not only in terms of convergence of the loss function during training, but also in validation and testing, with better results in the case of the larger dataset. Extensive tests are performed to assess the generalization capability of the Neural Network model when applied to profiles with localized features or assigned geometries not included in the original dataset. Moreover, its possible exploitation on domain sizes beyond the one used in the training is also analyzed in-depth. The conditions providing a one-to-one reproduction of the ground-truth $\rho(x)$ profiles computed by the Green's approximation are highlighted along with critical cases. The accuracy and robustness of the deep-learned $\rho(x)$ are further demonstrated in the time-integration of surface evolution problems described by simple partial differential equations of evaporation/condensation and surface diffusion.
[{'version': 'v1', 'created': 'Sun, 5 May 2024 20:34:16 GMT'}]
2025-03-04
Kamal Choudhary
AtomGPT: Atomistic Generative Pre-trained Transformer for Forward and Inverse Materials Design
null
null
null
cond-mat.mtrl-sci
Large language models (LLMs) such as generative pretrained transformers (GPTs) have shown potential for various commercial applications, but their applicability for materials design remains underexplored. In this article, we introduce AtomGPT, a model specifically developed for materials design based on transformer architectures, to demonstrate the capability for both atomistic property prediction and structure generation. We show that a combination of chemical and structural text descriptions can efficiently predict material properties with accuracy comparable to graph neural network models, including formation energies, electronic bandgaps from two different methods and superconducting transition temperatures. Furthermore, we demonstrate that AtomGPT can generate atomic structures for tasks such as designing new superconductors, with the predictions validated through density functional theory calculations. This work paves the way for leveraging LLMs in forward and inverse materials design, offering an efficient approach to the discovery and optimization of materials.
[{'version': 'v1', 'created': 'Mon, 6 May 2024 17:54:54 GMT'}, {'version': 'v2', 'created': 'Sat, 29 Jun 2024 06:24:30 GMT'}]
2024-07-02
Han Yang, Chenxi Hu, Yichi Zhou, Xixian Liu, Yu Shi, Jielan Li, Guanzhi Li, Zekun Chen, Shuizhou Chen, Claudio Zeni, Matthew Horton, Robert Pinsler, Andrew Fowler, Daniel Z\"ugner, Tian Xie, Jake Smith, Lixin Sun, Qian Wang, Lingyu Kong, Chang Liu, Hongxia Hao, Ziheng Lu
MatterSim: A Deep Learning Atomistic Model Across Elements, Temperatures and Pressures
null
null
null
cond-mat.mtrl-sci
Accurate and fast prediction of materials properties is central to the digital transformation of materials design. However, the vast design space and diverse operating conditions pose significant challenges for accurately modeling arbitrary material candidates and forecasting their properties. We present MatterSim, a deep learning model actively learned from large-scale first-principles computations, for efficient atomistic simulations at first-principles level and accurate prediction of broad material properties across the periodic table, spanning temperatures from 0 to 5000 K and pressures up to 1000 GPa. Out-of-the-box, the model serves as a machine learning force field, and shows remarkable capabilities not only in predicting ground-state material structures and energetics, but also in simulating their behavior under realistic temperatures and pressures, signifying an up to ten-fold enhancement in precision compared to the prior best-in-class. This enables MatterSim to compute materials' lattice dynamics, mechanical and thermodynamic properties, and beyond, to an accuracy comparable with first-principles methods. Specifically, MatterSim predicts Gibbs free energies for a wide range of inorganic solids with near-first-principles accuracy and achieves a 15 meV/atom resolution for temperatures up to 1000K compared with experiments. This opens an opportunity to predict experimental phase diagrams of materials at minimal computational cost. Moreover, MatterSim also serves as a platform for continuous learning and customization by integrating domain-specific data. The model can be fine-tuned for atomistic simulations at a desired level of theory or for direct structure-to-property predictions, achieving high data efficiency with a reduction in data requirements by up to 97%.
[{'version': 'v1', 'created': 'Wed, 8 May 2024 11:13:30 GMT'}, {'version': 'v2', 'created': 'Fri, 10 May 2024 16:49:52 GMT'}]
2024-05-13
Michael Vitz, Hamed Mohammadbagherpoor, Samarth Sandeep, Andrew Vlasic, Richard Padbury, and Anh Pham
Hybrid Quantum Graph Neural Network for Molecular Property Prediction
null
null
null
quant-ph cond-mat.mtrl-sci cs.LG
To accelerate the process of materials design, materials science has increasingly used data driven techniques to extract information from collected data. Specially, machine learning (ML) algorithms, which span the ML discipline, have demonstrated ability to predict various properties of materials with the level of accuracy similar to explicit calculation of quantum mechanical theories, but with significantly reduced run time and computational resources. Within ML, graph neural networks have emerged as an important algorithm within the field of machine learning, since they are capable of predicting accurately a wide range of important physical, chemical and electronic properties due to their higher learning ability based on the graph representation of material and molecular descriptors through the aggregation of information embedded within the graph. In parallel with the development of state of the art classical machine learning applications, the fusion of quantum computing and machine learning have created a new paradigm where classical machine learning model can be augmented with quantum layers which are able to encode high dimensional data more efficiently. Leveraging the structure of existing algorithms, we developed a unique and novel gradient free hybrid quantum classical convoluted graph neural network (HyQCGNN) to predict formation energies of perovskite materials. The performance of our hybrid statistical model is competitive with the results obtained purely from a classical convoluted graph neural network, and other classical machine learning algorithms, such as XGBoost. Consequently, our study suggests a new pathway to explore how quantum feature encoding and parametric quantum circuits can yield drastic improvements of complex ML algorithm like graph neural network.
[{'version': 'v1', 'created': 'Wed, 8 May 2024 16:43:25 GMT'}]
2024-05-09
Bowen Deng, Yunyeong Choi, Peichen Zhong, Janosh Riebesell, Shashwat Anand, Zhuohan Li, KyuJung Jun, Kristin A. Persson, Gerbrand Ceder
Overcoming systematic softening in universal machine learning interatomic potentials by fine-tuning
null
null
null
cond-mat.mtrl-sci cs.AI cs.LG
Machine learning interatomic potentials (MLIPs) have introduced a new paradigm for atomic simulations. Recent advancements have seen the emergence of universal MLIPs (uMLIPs) that are pre-trained on diverse materials datasets, providing opportunities for both ready-to-use universal force fields and robust foundations for downstream machine learning refinements. However, their performance in extrapolating to out-of-distribution complex atomic environments remains unclear. In this study, we highlight a consistent potential energy surface (PES) softening effect in three uMLIPs: M3GNet, CHGNet, and MACE-MP-0, which is characterized by energy and force under-prediction in a series of atomic-modeling benchmarks including surfaces, defects, solid-solution energetics, phonon vibration modes, ion migration barriers, and general high-energy states. We find that the PES softening behavior originates from a systematic underprediction error of the PES curvature, which derives from the biased sampling of near-equilibrium atomic arrangements in uMLIP pre-training datasets. We demonstrate that the PES softening issue can be effectively rectified by fine-tuning with a single additional data point. Our findings suggest that a considerable fraction of uMLIP errors are highly systematic, and can therefore be efficiently corrected. This result rationalizes the data-efficient fine-tuning performance boost commonly observed with foundational MLIPs. We argue for the importance of a comprehensive materials dataset with improved PES sampling for next-generation foundational MLIPs.
[{'version': 'v1', 'created': 'Sat, 11 May 2024 22:30:47 GMT'}]
2024-05-14
Ashley Lenau, Dennis M. Dimiduk, and Stephen R. Niezgoda
Importance of hyper-parameter optimization during training of physics-informed deep learning networks
null
null
null
cond-mat.mtrl-sci physics.data-an
Incorporating scientific knowledge into deep learning (DL) models for materials-based simulations can constrain the network's predictions to be within the boundaries of the material system. Altering loss functions or adding physics-based regularization (PBR) terms to reflect material properties informs a network about the physical constraints the simulation should obey. The training and tuning process of a DL network greatly affects the quality of the model, but how this process differs when using physics-based loss functions or regularization terms is not commonly discussed. In this manuscript, several PBR methods are implemented to enforce stress equilibrium on a network predicting the stress fields of a high elastic contrast composite. Models with PBR enforced the equilibrium constraint more accurately than a model without PBR, and the stress equilibrium converged more quickly. More importantly, it was observed that independently fine-tuning each implementation resulted in more accurate models. More specifically, each loss formulation and dataset required different learning rates and loss weights for the best performance. This result has important implications on assessing the relative effectiveness of different DL models and highlights important considerations when making a comparison between DL methods.
[{'version': 'v1', 'created': 'Tue, 14 May 2024 13:21:00 GMT'}, {'version': 'v2', 'created': 'Tue, 21 May 2024 21:31:46 GMT'}]
2024-05-24
Patxi Fernandez-Zelaia, Jason Mayeur, Jiahao Cheng, Yousub Lee, Kevin Knipe, Kai Kadau
Self-supervised feature distillation and design of experiments for efficient training of micromechanical deep learning surrogates
null
null
null
cs.CE cond-mat.mtrl-sci
Machine learning surrogate emulators are needed in engineering design and optimization tasks to rapidly emulate computationally expensive physics-based models. In micromechanics problems the local full-field response variables are desired at microstructural length scales. While there has been a great deal of work on establishing architectures for these tasks there has been relatively little work on establishing microstructural experimental design strategies. This work demonstrates that intelligent selection of microstructural volume elements for subsequent physics simulations enables the establishment of more accurate surrogate models. There exist two key challenges towards establishing a suitable framework: (1) microstructural feature quantification and (2) establishment of a criteria which encourages construction of a diverse training data set. Three feature extraction strategies are used as well as three design criteria. A novel contrastive feature extraction approach is established for automated self-supervised extraction of microstructural summary statistics. Results indicate that for the problem considered up to a 8\% improvement in surrogate performance may be achieved using the proposed design and training strategy. Trends indicate this approach may be even more beneficial when scaled towards larger problems. These results demonstrate that the selection of an efficient experimental design is an important consideration when establishing machine learning based surrogate models.
[{'version': 'v1', 'created': 'Thu, 16 May 2024 14:31:30 GMT'}]
2024-05-17
Stephen T. Lam, Shubhojit Banerjee, Rajni Chahal
Uncertainty and Exploration of Deep Learning-based Atomistic Models for Screening Molten Salt Properties and Compositions
null
null
null
cond-mat.mtrl-sci physics.chem-ph
Due to extreme chemical, thermal, and radiation environments, existing molten salt property databases lack the necessary experimental thermal properties of reactor-relevant salt compositions. Meanwhile, simulating these properties directly is typically either computationally expensive or inaccurate. In recent years, deep learning (DL)-based atomistic simulations have emerged as a method for achieving both efficiency and accuracy. However, there remain significant challenges in assessing model reliability in DL models when simulating properties and screening new systems. In this work, structurally complex LiF-NaF-ZrF$_4$ salt is studied. We show that neural network (NN) uncertainty can be quantified using ensemble learning to provide a 95% confidence interval (CI) for NN-based predictions. We show that DL models can successfully extrapolate to new compositions, temperatures, and timescales, but fail for significant changes in density, which is captured by ensemble-based uncertainty predictions. This enables improved confidence in utilizing simulated data for realistic reactor conditions, and guidelines for training deployable DL models.
[{'version': 'v1', 'created': 'Tue, 30 Apr 2024 21:20:55 GMT'}]
2024-05-20
Zijian Du, Luozhijie Jin, Le Shu, Yan Cen, Yuanfeng Xu, Yongfeng Mei and Hao Zhang
CTGNN: Crystal Transformer Graph Neural Network for Crystal Material Property Prediction
null
null
null
cond-mat.mtrl-sci physics.comp-ph
The combination of deep learning algorithm and materials science has made significant progress in predicting novel materials and understanding various behaviours of materials. Here, we introduced a new model called as the Crystal Transformer Graph Neural Network (CTGNN), which combines the advantages of Transformer model and graph neural networks to address the complexity of structure-properties relation of material data. Compared to the state-of-the-art models, CTGNN incorporates the graph network structure for capturing local atomic interactions and the dual-Transformer structures to model intra-crystal and inter-atomic relationships comprehensively. The benchmark carried on by the proposed CTGNN indicates that CTGNN significantly outperforms existing models like CGCNN and MEGNET in the prediction of formation energy and bandgap properties. Our work highlights the potential of CTGNN to enhance the performance of properties prediction and accelerates the discovery of new materials, particularly for perovskite materials.
[{'version': 'v1', 'created': 'Sun, 19 May 2024 10:00:06 GMT'}]
2024-05-21
Chinedu Ekuma
Computational toolkit for predicting thickness of 2D materials using machine learning and autogenerated dataset by large language model
null
null
null
cond-mat.mtrl-sci cond-mat.str-el
The thickness of 2D materials not only plays a crucial role in determining the performance of nanoelectronic and optoelectronic devices but also introduces complexities in predicting volume-dependent properties such as energy storage capacity, due to the intrinsic vacuum within these materials. Although a plethora of experimental techniques, including but not limited to optical contrast, Raman spectroscopy, nonlinear optical spectroscopy, near-field optical imaging, and hyperspectral imaging, facilitate the measurement of 2D material thickness, comprehensive data for many materials remains elusive. Over the last decade, the exponential proliferation of 2D materials and their heterostructures has outstripped the capabilities of conventional experimental and computational approaches. In this evolving landscape, machine learning (ML) has emerged as an indispensable tool, offering novel avenues to augment these traditional methodologies. Addressing the critical gap, we introduce THICK2D - Thickness Hierarchy Inference and Calculation Kit for 2D Materials. This Python-based computational framework harnesses an autogenerated thickness database, developed using large language models (LLMs), and advanced ML algorithms to facilitate the rapid and scalable estimation of material thickness, relying solely on crystallographic data. To demonstrate the utility and robustness of THICK2D, we successfully employed the toolkit to predict the thickness of more than 8000 2D-based materials, sourced from two extensive 2D material databases. THICK2D is disseminated as an open-source utility, accessible on GitHub https://github.com/gmp007/THICK2D, and archived on Zenodo at https://doi.org/10.5281/zenodo.11216648}{10.5281/zenodo.11216648.
[{'version': 'v1', 'created': 'Fri, 24 May 2024 01:05:47 GMT'}]
2024-05-27
M. Sipil\"a, F. Mehryary, S. Pyysalo, F. Ginter and Milica Todorovi\'c
Question Answering models for information extraction from perovskite materials science literature
null
null
null
cond-mat.mtrl-sci
Scientific text is a promising source of data in materials science, with ongoing research into utilising textual data for materials discovery. In this study, we developed and tested a novel approach to extract material-property relationships from scientific publications using the Question Answering (QA) method. QA performance was evaluated for information extraction of perovskite bandgaps based on a human query. We observed considerable variation in results with five different large language models fine-tuned for the QA task. Best extraction accuracy was achieved with the QA MatBERT and F1-scores improved on the current state-of-the-art. This work demonstrates the QA workflow and paves the way towards further applications. The simplicity, versatility and accuracy of the QA approach all point to its considerable potential for text-driven discoveries in materials research.
[{'version': 'v1', 'created': 'Fri, 24 May 2024 07:24:21 GMT'}, {'version': 'v2', 'created': 'Fri, 13 Sep 2024 11:27:16 GMT'}]
2024-09-16
Avishek Singh and Nirmal Ganguli
Unsupervised Deep Neural Network Approach To Solve Bosonic Systems
null
null
null
cond-mat.mtrl-sci cond-mat.quant-gas
The simulation of quantum many-body systems poses a significant challenge in physics due to the exponential scaling of Hilbert space with the number of particles. Traditional methods often struggle with large system sizes and frustrated lattices. In this research article, we present a novel algorithm that leverages the power of deep neural networks combined with Markov Chain Monte Carlo simulation to address these limitations. Our method introduces a neural network architecture specifically designed to represent bosonic quantum states on a 1D lattice chain. We successfully achieve the ground state of the Bose-Hubbard model, demonstrating the superiority of the adaptive momentum optimizer for convergence speed and stability. Notably, our approach offers flexibility in simulating various lattice geometries and potentially larger system sizes, making it a valuable tool for exploring complex quantum phenomena. This work represents a substantial advancement in the field of quantum simulation, opening new possibilities for investigating previously challenging systems.
[{'version': 'v1', 'created': 'Fri, 24 May 2024 12:09:20 GMT'}]
2024-05-27
Avishek Singh and Nirmal Ganguli
Unsupervised Deep Neural Network Approach To Solve Fermionic Systems
null
null
null
cond-mat.mtrl-sci cond-mat.str-el
Solving the Schr\"{o}dinger equation for interacting many-body quantum systems faces computational challenges due to exponential scaling with system size. This complexity limits the study of important phenomena in materials science and physics. We develop an Artificial Neural Network (ANN)-driven algorithm to simulate fermionic systems on lattices. Our method uses Pauli matrices to represent quantum states, incorporates Markov Chain Monte Carlo sampling, and leverages an adaptive momentum optimizer. We demonstrate the algorithm's accuracy by simulating the Heisenberg Hamiltonian on a one-dimensional lattice, achieving results with an error in the order of $10^{-4}$ compared to exact diagonalization. Furthermore, we successfully model a magnetic phase transition in a two-dimensional lattice under an applied magnetic field. Importantly, our approach avoids the sign problem common to traditional Fermionic Monte Carlo methods, enabling the investigation of frustrated systems. This work demonstrates the potential of ANN-based algorithms for efficient simulation of complex quantum systems, opening avenues for discoveries in condensed matter physics and materials science.
[{'version': 'v1', 'created': 'Fri, 24 May 2024 12:41:02 GMT'}]
2024-05-27
Haosheng Xu, Dongheng Qian, and Jing Wang
Predicting Many Crystal Properties via an Adaptive Transformer-based Framework
null
null
null
cond-mat.mtrl-sci cond-mat.mes-hall cs.LG
Machine learning has revolutionized many fields, including materials science. However, predicting properties of crystalline materials using machine learning faces challenges in input encoding, output versatility, and interpretability. We introduce CrystalBERT, an adaptable transformer-based framework integrating space group, elemental, and unit cell information. This novel structure can seamlessly combine diverse features and accurately predict various physical properties, including topological properties, superconducting transition temperatures, dielectric constants, and more. CrystalBERT provides insightful interpretations of features influencing target properties. Our results indicate that space group and elemental information are crucial for predicting topological and superconducting properties, underscoring their intricate nature. By incorporating these features, we achieve 91\% accuracy in topological classification, surpassing prior studies and identifying previously misclassified materials. This research demonstrates that integrating diverse material information enhances the prediction of complex material properties, paving the way for more accurate and interpretable machine learning models in materials science.
[{'version': 'v1', 'created': 'Wed, 29 May 2024 09:56:00 GMT'}, {'version': 'v2', 'created': 'Fri, 13 Dec 2024 06:23:03 GMT'}]
2024-12-16
Harveen Kaur, Flaviano Della Pia, Ilyes Batatia, Xavier R. Advincula, Benjamin X. Shi, Jinggang Lan, G\'abor Cs\'anyi, Angelos Michaelides, and Venkat Kapil
Data-efficient fine-tuning of foundational models for first-principles quality sublimation enthalpies
null
null
null
cond-mat.mtrl-sci physics.chem-ph
Calculating sublimation enthalpies of molecular crystal polymorphs is relevant to a wide range of technological applications. However, predicting these quantities at first-principles accuracy -- even with the aid of machine learning potentials -- is a challenge that requires sub-kJ/mol accuracy in the potential energy surface and finite-temperature sampling. We present an accurate and data-efficient protocol based on fine-tuning of the foundational MACE-MP-0 model and showcase its capabilities on sublimation enthalpies and physical properties of ice polymorphs. Our approach requires only a few tens of training structures to achieve sub-kJ/mol accuracy in the sublimation enthalpies and sub 1 % error in densities for polymorphs at finite temperature and pressure. Exploiting this data efficiency, we explore simulations of hexagonal ice at the random phase approximation level of theory at experimental temperatures and pressures, calculating its physical properties, like pair correlation function and density, with good agreement with experiments. Our approach provides a way forward for predicting the stability of molecular crystals at finite thermodynamic conditions with the accuracy of correlated electronic structure theory.
[{'version': 'v1', 'created': 'Thu, 30 May 2024 16:18:29 GMT'}]
2024-05-31
Malte Grunert, Max Gro{\ss}mann, Erich Runge
Deep learning of spectra: Predicting the dielectric function of semiconductors
Phys. Rev. Materials 8, L122201 (2024)
10.1103/PhysRevMaterials.8.L122201
null
cond-mat.mtrl-sci
Predicting spectra and related properties such as the dielectric function of crystalline materials based on machine learning has a huge, hitherto unexplored, technological potential. For this reason, we create an ab initio database of 9915 dielectric tensors of semiconductors and insulators calculated in the independent-particle approximation (IPA). In addition, we present the OptiMate family of machine learning models, a series of graph attention neural networks (GAT) trained to predict the dielectric function and refractive index. OptiMate yields accurate prediction of spectra of semiconductors using only their crystal structure. Smooth, artifact-free curves are obtained without these properties being enforced by penalties.
[{'version': 'v1', 'created': 'Wed, 12 Jun 2024 13:21:29 GMT'}, {'version': 'v2', 'created': 'Fri, 20 Dec 2024 12:39:13 GMT'}]
2024-12-23
Huazhang Zhang, Hao-Cheng Thong, Louis Bastogne, Churen Gui, Xu He, Philippe Ghosez
Finite-temperature properties of antiferroelectric perovskite $\rm PbZrO_3$ from deep learning interatomic potential
null
null
null
cond-mat.mtrl-sci
The prototypical antiferroelectric perovskite $\rm PbZrO_3$ (PZO) has garnered considerable attentions in recent years due to its significance in technological applications and fundamental research. Many unresolved issues in PZO are associated with large length- and time-scales, as well as finite temperatures, presenting significant challenges for first-principles density functional theory studies. Here, we introduce a deep learning interatomic potential of PZO, enabling investigation of finite-temperature properties through large-scale atomistic simulations. Trained using an elaborately designed dataset, the model successfully reproduces a large number of phases, in particular, the recently discovered 80-atom antiferroelectric $Pnam$ phase and ferrielectric $Ima2$ phase, providing precise predictions for their structural and dynamical properties. Using this model, we investigated phase transitions of multiple phases, including $Pbam$/$Pnam$, $Ima2$ and $R3c$, which show high similarity to the experimental observation. Our simulation results also highlight the crucial role of free-energy in determining the low-temperature phase of PZO, reconciling the apparent contradiction: $Pbam$ is the most commonly observed phase in experiments, while theoretical calculations predict other phases exhibiting even lower energy. Furthermore, in the temperature range where the $Pbam$ phase is thermodynamically stable, typical double polarization hysteresis loops for antiferroelectrics were obtained, along with a detailed elucidation of the structural evolution during the electric-field induced transitions between the non-polar $Pbam$ and polar $R3c$ phases.
[{'version': 'v1', 'created': 'Thu, 13 Jun 2024 11:32:16 GMT'}, {'version': 'v2', 'created': 'Wed, 31 Jul 2024 10:22:29 GMT'}, {'version': 'v3', 'created': 'Wed, 21 Aug 2024 11:51:17 GMT'}]
2024-08-22
Davi M F\'ebba, Kingsley Egbo, William A. Callahan, Andriy Zakutayev
From Text to Test: AI-Generated Control Software for Materials Science Instruments
null
10.1039/D4DD00143E
null
cond-mat.mtrl-sci cs.AI
Large language models (LLMs) are transforming the landscape of chemistry and materials science. Recent examples of LLM-accelerated experimental research include virtual assistants for parsing synthesis recipes from the literature, or using the extracted knowledge to guide synthesis and characterization. Despite these advancements, their application is constrained to labs with automated instruments and control software, leaving much of materials science reliant on manual processes. Here, we demonstrate the rapid deployment of a Python-based control module for a Keithley 2400 electrical source measure unit using ChatGPT-4. Through iterative refinement, we achieved effective instrument management with minimal human intervention. Additionally, a user-friendly graphical user interface (GUI) was created, effectively linking all instrument controls to interactive screen elements. Finally, we integrated this AI-crafted instrument control software with a high-performance stochastic optimization algorithm to facilitate rapid and automated extraction of electronic device parameters related to semiconductor charge transport mechanisms from current-voltage (IV) measurement data. This integration resulted in a comprehensive open-source toolkit for semiconductor device characterization and analysis using IV curve measurements. We demonstrate the application of these tools by acquiring, analyzing, and parameterizing IV data from a Pt/Cr$_2$O$_3$:Mg/$\beta$-Ga$_2$O$_3$ heterojunction diode, a novel stack for high-power and high-temperature electronic devices. This approach underscores the powerful synergy between LLMs and the development of instruments for scientific inquiry, showcasing a path for further acceleration in materials science.
[{'version': 'v1', 'created': 'Sun, 23 Jun 2024 21:32:57 GMT'}, {'version': 'v2', 'created': 'Tue, 25 Jun 2024 11:34:15 GMT'}]
2024-11-12
Nguyen Tuan Hung, Ryotaro Okabe, Abhijatmedhi Chotrattanapituk, Mingda Li
Ensemble-Embedding Graph Neural Network for Direct Prediction of Optical Spectra from Crystal Structure
null
null
null
cond-mat.mtrl-sci physics.app-ph
Optical properties in solids, such as refractive index and absorption, hold vast applications ranging from solar panels to sensors, photodetectors, and transparent displays. However, first-principles computation of optical properties from crystal structures is a complex task due to the high convergence criteria and computational cost. Recent progress in machine learning shows promise in predicting material properties, yet predicting optical properties from crystal structures remains challenging due to the lack of efficient atomic embeddings. Here, we introduce GNNOpt, an equivariance graph-neural-network architecture featuring automatic embedding optimization. This enables high-quality optical predictions with a dataset of only 944 materials. GNNOpt predicts all optical properties based on the Kramers-Kr{\"o}nig relations, including absorption coefficient, complex dielectric function, complex refractive index, and reflectance. We apply the trained model to screen photovoltaic materials based on spectroscopic limited maximum efficiency and search for quantum materials based on quantum weight. First-principles calculations validate the efficacy of the GNNOpt model, demonstrating excellent agreement in predicting the optical spectra of unseen materials. The discovery of new quantum materials with high predicted quantum weight, such as SiOs which hosts exotic quasiparticles, demonstrates GNNOpt's potential in predicting optical properties across a broad range of materials and applications.
[{'version': 'v1', 'created': 'Mon, 24 Jun 2024 14:02:29 GMT'}]
2024-06-25
Zechen Tang, Nianlong Zou, He Li, Yuxiang Wang, Zilong Yuan, Honggeng Tao, Yang Li, Zezhou Chen, Boheng Zhao, Minghui Sun, Hong Jiang, Wenhui Duan, Yong Xu
Improving density matrix electronic structure method by deep learning
null
null
null
physics.comp-ph cond-mat.mtrl-sci
The combination of deep learning and ab initio materials calculations is emerging as a trending frontier of materials science research, with deep-learning density functional theory (DFT) electronic structure being particularly promising. In this work, we introduce a neural-network method for modeling the DFT density matrix, a fundamental yet previously unexplored quantity in deep-learning electronic structure. Utilizing an advanced neural network framework that leverages the nearsightedness and equivariance properties of the density matrix, the method demonstrates high accuracy and excellent generalizability in multiple example studies, as well as capability to precisely predict charge density and reproduce other electronic structure properties. Given the pivotal role of the density matrix in DFT as well as other computational methods, the current research introduces a novel approach to the deep-learning study of electronic structure properties, opening up new opportunities for deep-learning enhanced computational materials study.
[{'version': 'v1', 'created': 'Tue, 25 Jun 2024 13:55:40 GMT'}]
2024-06-26
Michael Moran, Vladimir V. Gusev, Michael W. Gaultois, Dmytro Antypov, Matthew J. Rosseinsky
Establishing Deep InfoMax as an effective self-supervised learning methodology in materials informatics
null
null
null
cs.LG cond-mat.mtrl-sci
The scarcity of property labels remains a key challenge in materials informatics, whereas materials data without property labels are abundant in comparison. By pretraining supervised property prediction models on self-supervised tasks that depend only on the "intrinsic information" available in any Crystallographic Information File (CIF), there is potential to leverage the large amount of crystal data without property labels to improve property prediction results on small datasets. We apply Deep InfoMax as a self-supervised machine learning framework for materials informatics that explicitly maximises the mutual information between a point set (or graph) representation of a crystal and a vector representation suitable for downstream learning. This allows the pretraining of supervised models on large materials datasets without the need for property labels and without requiring the model to reconstruct the crystal from a representation vector. We investigate the benefits of Deep InfoMax pretraining implemented on the Site-Net architecture to improve the performance of downstream property prediction models with small amounts (<10^3) of data, a situation relevant to experimentally measured materials property databases. Using a property label masking methodology, where we perform self-supervised learning on larger supervised datasets and then train supervised models on a small subset of the labels, we isolate Deep InfoMax pretraining from the effects of distributional shift. We demonstrate performance improvements in the contexts of representation learning and transfer learning on the tasks of band gap and formation energy prediction. Having established the effectiveness of Deep InfoMax pretraining in a controlled environment, our findings provide a foundation for extending the approach to address practical challenges in materials informatics.
[{'version': 'v1', 'created': 'Sun, 30 Jun 2024 11:33:49 GMT'}]
2024-07-02
Somnath Bharech, Yangyiwei Yang, Michael Selzer, Britta Nestler, Bai-Xiang Xu
ML-extendable framework for multiphysics-multiscale simulation workflow and data management using Kadi4Mat
null
null
null
cond-mat.mtrl-sci
As material modeling and simulation has become vital for modern materials science, research data with distinctive physical principles and extensive volume are generally required for full elucidation of the material behavior across all relevant scales. Effective workflow and data management, with corresponding metadata descriptions, helps leverage the full potential of data-driven analyses for computer-aided material design. In this work, we propose a research workflow and data management (RWDM) framework to manage complex workflows and resulting research (meta)data, while following FAIR principles. Multiphysics multiscale simulations for additive manufacturing investigations are treated as showcase and implemented on Kadi4Mat: an open source research data infrastructure. The input and output data of the simulations, together with the associated setups and scripts realizing the simulation workflow, are curated in corresponding standardized Kadi4Mat records with extendibility for further research and data-driven analyses. These records are interlinked to indicate information flow and form an ontology based knowledge graph. Automation scheme for performing high-throughput simulation and post-processing integrated with the proposed RWDM framework is also presented.
[{'version': 'v1', 'created': 'Tue, 2 Jul 2024 11:13:41 GMT'}]
2024-07-03
Seifallah Elfetni and Reza Darvishi Kamachali
PINNs-MPF: A Physics-Informed Neural Network Framework for Multi-Phase-Field Simulation of Interface Dynamics
null
null
null
cond-mat.mtrl-sci physics.comp-ph
We present an application of Physics-Informed Neural Networks to handle MultiPhase-Field simulations of microstructure evolution. It has been showcased that a combination of optimization techniques extended and adapted from the PINNs literature, and the introduction of specific techniques inspired by the MPF Method background, is required. The numerical resolution is realized through a multi-variable time-series problem by using fully discrete resolution. Within each interval, space, time, and phases are treated separately, constituting discrete subdomains. An extended multi-networking concept is implemented to subdivide the simulation domain into multiple batches, with each batch associated with an independent Neural Network trained to predict the solution. To ensure efficient interaction across different phasesand in the spatio-temporal-phasic subdomain, a Master NN handles efficient interaction among the multiple networks, as well as the transfer of learning in different directions. A set of systematic simulations with increasing complexity was performed, that benchmarks various critical aspects of MPF simulations, including different geometries, types of interface dynamics and the evolution of an interfacial triple junction. A comprehensive approach is adopted to specifically focus the attention on the interfacial regions through an automatic and dynamic meshing process, significantly simplifying the tuning of hyper-parameters and serving as a fundamental key for addressing MPF problems using Machine Learning. The pyramidal training approach is proposed to the PINN community as a dual-impact method: it facilitates the initialization of training and allows an extended transfer of learning. The proposed PINNs-MPF framework successfully reproduces benchmark tests with high fidelity and Mean Squared Error loss values ranging from 10$^{-4}$ to 10$^{-6}$ compared to ground truth solutions.
[{'version': 'v1', 'created': 'Tue, 2 Jul 2024 12:55:01 GMT'}, {'version': 'v2', 'created': 'Fri, 30 Aug 2024 18:07:34 GMT'}]
2024-09-04
Ji Wei Yoon, Bangjian Zhou, J Senthilnath
SG-NNP: Species-separated Gaussian Neural Network Potential with Linear Elemental Scaling and Optimized Dimensions for Multi-component Materials
null
null
null
cond-mat.mtrl-sci
Accurate simulations of materials at long-time and large-length scales have increasingly been enabled by Machine-learned Interatomic Potentials (MLIPs). There have been increasing interest on improving the robustness of such models. To this end, we engineer a novel set of Gaussian-type descriptors that scale linearly with the number of atoms, reduce informational degeneracy for multi-component atomic environments and apply them in Species-separated Gaussian Neural Network Potentials (SG-NNPs). The robustness of our method was tested by analyzing the impact of various design choices and hyperparameters on Molybdenum (Mo) SG-NNP performance during training and inference/simulation. With less dimensions, SG-NNPs are shown to have superior atomic forces and total energy predictions than other traditional and ML descriptor-based interatomic potentials on diverse set of materials - Ni, Cu, Li, Mo, Si, Ge, NiMo, Li3N and NbMoTaW. From the obtained results we can observe that the proposed method improves the performance of atomic descriptors of complex environments with multiple species.
[{'version': 'v1', 'created': 'Tue, 9 Jul 2024 07:46:34 GMT'}]
2024-07-10
Zhilong Song, Shuaihua Lu, Minggang Ju, Qionghua Zhou and Jinlan Wang
Is Large Language Model All You Need to Predict the Synthesizability and Precursors of Crystal Structures?
null
null
null
cond-mat.mtrl-sci
Accessing the synthesizability of crystal structures is pivotal for advancing the practical application of theoretical material structures designed by machine learning or high-throughput screening. However, a significant gap exists between the actual synthesizability and thermodynamic or kinetic stability, which is commonly used for screening theoretical structures for experiments. To address this, we develop the Crystal Synthesis Large Language Models (CSLLM) framework, which includes three LLMs for predicting the synthesizability, synthesis methods, and precursors. We create a comprehensive synthesizability dataset including 140,120 crystal structures and develop an efficient text representation method for crystal structures to fine-tune the LLMs. The Synthesizability LLM achieves a remarkable 98.6% accuracy, significantly outperforming traditional synthesizability screening based on thermodynamic and kinetic stability by 106.1% and 44.5%, respectively. The Methods LLM achieves a classification accuracy of 91.02%, and the Precursors LLM has an 80.2% success rate in predicting synthesis precursors. Furthermore, we develop a user-friendly graphical interface that enables automatic predictions of synthesizability and precursors from uploaded crystal structure files. Through these contributions, CSLLM bridges the gap between theoretical material design and experimental synthesis, paving the way for the rapid discovery of novel and synthesizable functional materials.
[{'version': 'v1', 'created': 'Tue, 9 Jul 2024 16:35:12 GMT'}]
2024-07-10
Joseph Musielewicz, Janice Lan, Matt Uyttendaele, and John R. Kitchin
Improved Uncertainty Estimation of Graph Neural Network Potentials Using Engineered Latent Space Distances
null
null
null
cs.LG cond-mat.mtrl-sci
Graph neural networks (GNNs) have been shown to be astonishingly capable models for molecular property prediction, particularly as surrogates for expensive density functional theory calculations of relaxed energy for novel material discovery. However, one limitation of GNNs in this context is the lack of useful uncertainty prediction methods, as this is critical to the material discovery pipeline. In this work, we show that uncertainty quantification for relaxed energy calculations is more complex than uncertainty quantification for other kinds of molecular property prediction, due to the effect that structure optimizations have on the error distribution. We propose that distribution-free techniques are more useful tools for assessing calibration, recalibrating, and developing uncertainty prediction methods for GNNs performing relaxed energy calculations. We also develop a relaxed energy task for evaluating uncertainty methods for equivariant GNNs, based on distribution-free recalibration and using the Open Catalyst Project dataset. We benchmark a set of popular uncertainty prediction methods on this task, and show that latent distance methods, with our novel improvements, are the most well-calibrated and economical approach for relaxed energy calculations. Finally, we demonstrate that our latent space distance method produces results which align with our expectations on a clustering example, and on specific equation of state and adsorbate coverage examples from outside the training dataset.
[{'version': 'v1', 'created': 'Mon, 15 Jul 2024 15:59:39 GMT'}, {'version': 'v2', 'created': 'Mon, 26 Aug 2024 17:31:16 GMT'}]
2024-08-27
Erwin Cazares and Brian E. Schuster
Deep Learning for Quantitative Dynamic Fragmentation Analysis
null
null
null
cond-mat.mtrl-sci
We have developed an image-based convolutional neural network (CNN) that is applicable for quantitative time-resolved measurements of the fragmentation behavior of opaque brittle materials using ultra-high speed optical imaging. This model extends previous work on the U-net model, where we trained binary, 3 and 5 class models using supervised learning on experimentally measured dynamic fracture experiments on various opaque structural ceramic materials that were adhered on transparent polymer (polycarbonate or acrylic) backing materials. Full details of the experimental investigations are outside the scope of this manuscript but briefly, several different ceramics were loaded using spatially and time-varying mechanical loads to induce inelastic deformation and fracture processes that were recorded at frequencies as high as 5 MHz using high speed optical imaging. These experiments provided a rich and diverse dataset that includes many of the common fracture modes found in static and dynamic fracture including cone cracking, median cracking, comminution, and combined complex failure modes that involve effectively simultaneous activation and propagation of multiple fragmentation modes. While the training data presented here was obtained from dynamic fragmentation experiments, this study is applicable to static loading of these materials as the crack speeds typically higher a kilometer per second in these materials are on the order of 1-10 km/s regardless of the loading rate. We believe the methodologies presented here will be useful in quantifying the failure processes in structural materials for protection applications and can be used for direct validation of engineering models used in design.
[{'version': 'v1', 'created': 'Wed, 17 Jul 2024 19:35:57 GMT'}]
2024-07-19
Zilong Yuan, Zechen Tang, Honggeng Tao, Xiaoxun Gong, Zezhou Chen, Yuxiang Wang, He Li, Yang Li, Zhiming Xu, Minghui Sun, Boheng Zhao, Chong Wang, Wenhui Duan, Yong Xu
Deep learning density functional theory Hamiltonian in real space
null
null
null
physics.comp-ph cond-mat.mtrl-sci
Deep learning electronic structures from ab initio calculations holds great potential to revolutionize computational materials studies. While existing methods proved success in deep-learning density functional theory (DFT) Hamiltonian matrices, they are limited to DFT programs using localized atomic-like bases and heavily depend on the form of the bases. Here, we propose the DeepH-r method for deep-learning DFT Hamiltonians in real space, facilitating the prediction of DFT Hamiltonian in a basis-independent manner. An equivariant neural network architecture for modeling the real-space DFT potential is developed, targeting a more fundamental quantity in DFT. The real-space potential exhibits simplified principles of equivariance and enhanced nearsightedness, further boosting the performance of deep learning. When applied to evaluate the Hamiltonian matrix, this method significantly improved in accuracy, as exemplified in multiple case studies. Given the abundance of data in the real-space potential, this work may pave a novel pathway for establishing a ``large materials model" with increased accuracy.
[{'version': 'v1', 'created': 'Fri, 19 Jul 2024 15:07:22 GMT'}]
2024-07-22
Nihang Fu, Sadman Sadeed Omee, Jianjun Hu
Physical Encoding Improves OOD Performance in Deep Learning Materials Property Prediction
null
null
null
cond-mat.mtrl-sci
Deep learning (DL) models have been widely used in materials property prediction with great success, especially for properties with large datasets. However, the out-of-distribution (OOD) performances of such models are questionable, especially when the training set is not large enough. Here we showed that using physical encoding rather than the widely used one-hot encoding can significantly improve the OOD performance by increasing the models' generalization performance, which is especially true for models trained with small datasets. Our benchmark results of both composition- and structure-based deep learning models over six datasets including formation energy, band gap, refractive index, and elastic properties predictions demonstrated the importance of physical encoding to OOD generalization for models trained on small datasets.
[{'version': 'v1', 'created': 'Sun, 21 Jul 2024 16:40:28 GMT'}]
2024-07-23
Alexander Gorfer and David Heuser and Rainer Abart and Christoph Dellago
Thermodynamics of alkali feldspar solid solutions with varying Al-Si order: atomistic simulations using a neural network potential
null
null
null
cond-mat.mtrl-sci physics.comp-ph physics.geo-ph
The thermodynamic mixing properties of alkali feldspar solid solutions between the Na and K end members were computed through atomistic simulations using a neural network potential. We performed combined molecular dynamics and Monte Carlo simulations in the semi-grand canonical ensemble at 800 {\deg}C and considered three quenched disorder states in the Al-Si-O framework ranging from fully ordered to fully disordered. The excess Gibbs energy of mixing, excess enthalpy of mixing and excess entropy of mixing are in good agreement with literature data. In particular, the notion that increasing disorder in the Al-Si-O framework correlates with increasing ideality of Na-K mixing is successfully predicted. Finally, a recently proposed short range ordering of Na and K in the alkali sublattice is observed, which may be considered as a precursor to exsolution lamellae, a characteristic phenomenon in alkali feldspar of intermediate composition leading to perthite formation during cooling.
[{'version': 'v1', 'created': 'Wed, 24 Jul 2024 17:34:03 GMT'}]
2024-07-25
Suchona Akter, Yong Li, Minbum Kim, Md Omar Faruque, Zhonghua Peng, Praveen K. Thallapally, and Mohammad R. Momeni
Fine-tuning Microporosity of Crystalline Vanadomolybdate Frameworks for Selective Adsorptive Separation of Kr from Xe
Langmuir 2024 40 (47), 24934-24944
10.1021/acs.langmuir.4c02910
null
cond-mat.mtrl-sci
Selective adsorptive capture and separation of chemically inert Kr and Xe noble gases with very low ppmv concentrations in air and industrial off-gases constitute an important technological challenge. Here, using a synergistic combination of experiment and theory, the microporous crystalline vanadomolybdates (MoVOx) as highly selective Kr sorbents are studied in detail. By varying the Mo/V ratios, we show for the first time that their one-dimensional pores can be fine-tuned for the size-selective adsorption of Kr over the larger Xe with selectivities reaching >100. Using extensive electronic structure calculations and grand canonical Monte-Carlo simulations, the competition between Kr uptake with CO2 and N2 was also investigated. As most materials reported so far are selective toward the larger, more polarizable Xe than Kr, this work constitutes an important step toward robust Kr-selective sorbent materials. This work highlights the potential use of porous crystalline transition metal oxides as energy-efficient and selective noble gas capture sorbents for industrial applications.
[{'version': 'v1', 'created': 'Sat, 27 Jul 2024 12:54:17 GMT'}]
2025-05-08
Zihan Wang, Anindya Bhaduri, Hongyi Xu, Liping Wang
An Uncertainty-aware Deep Learning Framework-based Robust Design Optimization of Metamaterial Units
null
null
null
eess.SP cond-mat.mtrl-sci cs.LG
Mechanical metamaterials represent an innovative class of artificial structures, distinguished by their extraordinary mechanical characteristics, which are beyond the scope of traditional natural materials. The use of deep generative models has become increasingly popular in the design of metamaterial units. The effectiveness of using deep generative models lies in their capacity to compress complex input data into a simplified, lower-dimensional latent space, while also enabling the creation of novel optimal designs through sampling within this space. However, the design process does not take into account the effect of model uncertainty due to data sparsity or the effect of input data uncertainty due to inherent randomness in the data. This might lead to the generation of undesirable structures with high sensitivity to the uncertainties in the system. To address this issue, a novel uncertainty-aware deep learning framework-based robust design approach is proposed for the design of metamaterial units with optimal target properties. The proposed approach utilizes the probabilistic nature of the deep learning framework and quantifies both aleatoric and epistemic uncertainties associated with surrogate-based design optimization. We demonstrate that the proposed design approach is capable of designing high-performance metamaterial units with high reliability. To showcase the effectiveness of the proposed design approach, a single-objective design optimization problem and a multi-objective design optimization problem are presented. The optimal robust designs obtained are validated by comparing them to the designs obtained from the topology optimization method as well as the designs obtained from a deterministic deep learning framework-based design optimization where none of the uncertainties in the system are explicitly considered.
[{'version': 'v1', 'created': 'Fri, 19 Jul 2024 22:21:27 GMT'}]
2024-07-31
Christian Venturella, Jiachen Li, Christopher Hillenbrand, Ximena Leyva Peralta, Jessica Liu, Tianyu Zhu
Unified Deep Learning Framework for Many-Body Quantum Chemistry via Green's Functions
null
null
null
physics.chem-ph cond-mat.mtrl-sci physics.comp-ph
Quantum many-body methods provide a systematic route to computing electronic properties of molecules and materials, but high computational costs restrict their use in large-scale applications. Due to the complexity in many-electron wavefunctions, machine learning models capable of capturing fundamental many-body physics remain limited. Here, we present a deep learning framework targeting the many-body Green's function, which unifies predictions of electronic properties in ground and excited states, while offering deep physical insights into electron correlation effects. By learning the $GW$ or coupled-cluster self-energy from mean-field features, our graph neural network achieves competitive performance in predicting one- and two-particle excitations and quantities derivable from one-particle density matrix. We demonstrate its high data efficiency and good transferability across chemical species, system sizes, molecular conformations, and correlation strengths in bond breaking, through multiple molecular and nanomaterial benchmarks. This work opens up new opportunities for utilizing machine learning to solve many-electron problems.
[{'version': 'v1', 'created': 'Mon, 29 Jul 2024 19:20:52 GMT'}]
2024-07-31
Isaiah A. Moses, Wesley F. Reinhart
Transfer Learning for Multi-material Classification of Transition Metal Dichalcogenides with Atomic Force Microscopy
null
null
null
cond-mat.mtrl-sci physics.comp-ph
Deep learning models are widely used for the data-driven design of materials based on atomic force microscopy (AFM) and other scanning probe microscopy. These tools enhance efficiency in inverse design and characterization of materials. However, limited and imbalanced experimental materials data typically available is a major challenge. Also important is the need to interpret trained models, which have typically been complex enough to be uninterpretable by humans. Here, we present a systemic evaluation of transfer learning strategies to accommodate low-data scenarios in materials synthesis and a model latent feature analysis to draw connections to the human-interpretable characteristics of the samples. Our models show accurate predictions in five classes of transition metal dichalcogenides (TMDs) (MoS$_2$, WS$_2$, WSe$_2$, MoSe$_2$, and Mo-WSe$_2$) with up to 89$\%$ accuracy on held-out test samples. Analysis of the latent features reveals a correlation with physical characteristics such as grain density, DoG blob, and local variation. The transfer learning optimization modality and the exploration of the correlation between the latent and physical features provide important frameworks that can be applied to other classes of materials beyond TMDs to enhance the models' performance and explainability which can accelerate the inverse design of materials for technological applications.
[{'version': 'v1', 'created': 'Tue, 30 Jul 2024 17:06:42 GMT'}, {'version': 'v2', 'created': 'Tue, 10 Dec 2024 22:27:58 GMT'}]
2024-12-12
Shunya Minami, Yoshihiro Hayashi, Stephen Wu, Kenji Fukumizu, Hiroki Sugisawa, Masashi Ishii, Isao Kuwajima, Kazuya Shiratori, Ryo Yoshida
Scaling Law of Sim2Real Transfer Learning in Expanding Computational Materials Databases for Real-World Predictions
null
null
null
cond-mat.mtrl-sci cs.LG
To address the challenge of limited experimental materials data, extensive physical property databases are being developed based on high-throughput computational experiments, such as molecular dynamics simulations. Previous studies have shown that fine-tuning a predictor pretrained on a computational database to a real system can result in models with outstanding generalization capabilities compared to learning from scratch. This study demonstrates the scaling law of simulation-to-real (Sim2Real) transfer learning for several machine learning tasks in materials science. Case studies of three prediction tasks for polymers and inorganic materials reveal that the prediction error on real systems decreases according to a power-law as the size of the computational data increases. Observing the scaling behavior offers various insights for database development, such as determining the sample size necessary to achieve a desired performance, identifying equivalent sample sizes for physical and computational experiments, and guiding the design of data production protocols for downstream real-world tasks.
[{'version': 'v1', 'created': 'Wed, 7 Aug 2024 18:47:58 GMT'}]
2024-08-09
Ali Riza Durmaz, Akhil Thomas, Lokesh Mishra, Rachana Niranjan Murthy, Thomas Straub
MaterioMiner -- An ontology-based text mining dataset for extraction of process-structure-property entities
null
null
null
cs.CL cond-mat.mtrl-sci
While large language models learn sound statistical representations of the language and information therein, ontologies are symbolic knowledge representations that can complement the former ideally. Research at this critical intersection relies on datasets that intertwine ontologies and text corpora to enable training and comprehensive benchmarking of neurosymbolic models. We present the MaterioMiner dataset and the linked materials mechanics ontology where ontological concepts from the mechanics of materials domain are associated with textual entities within the literature corpus. Another distinctive feature of the dataset is its eminently fine-granular annotation. Specifically, 179 distinct classes are manually annotated by three raters within four publications, amounting to a total of 2191 entities that were annotated and curated. Conceptual work is presented for the symbolic representation of causal composition-process-microstructure-property relationships. We explore the annotation consistency between the three raters and perform fine-tuning of pre-trained models to showcase the feasibility of named-entity recognition model training. Reusing the dataset can foster training and benchmarking of materials language models, automated ontology construction, and knowledge graph generation from textual data.
[{'version': 'v1', 'created': 'Mon, 5 Aug 2024 21:42:59 GMT'}]
2024-08-12
A. K. Shargh, C. D. Stiles, J. A. El-Awady
Deep Learning Accelerated Phase Prediction of Refractory Multi-Principal Element Alloys
null
null
null
cond-mat.mtrl-sci
The tunability of the mechanical properties of refractory multi-principal-element alloys (RMPEAs) make them attractive for numerous high-temperature applications. It is well-established that the phase stability of RMPEAs control their mechanical properties. In this study, we develop a deep learning framework that is trained on a CALPHAD-derived database that is predictive of RMPEAs phases with high accuracy up to eight phases within the elemental space of Ti, Fe, Al, V, Ni, Nb, and Zr with an accuracy of approximately 90%. We further investigate the causes for the low out of domain performance of the deep learning models in predicting phases of RMPEA with new elemental sets and propose a strategy to mitigate this performance shortfall.
[{'version': 'v1', 'created': 'Mon, 12 Aug 2024 15:42:52 GMT'}]
2024-08-13
Yan Chen, Xueru Wang, Xiaobin Deng, Yilun Liu, Xi Chen, Yunwei Zhang, Lei Wang, Hang Xiao
MatterGPT: A Generative Transformer for Multi-Property Inverse Design of Solid-State Materials
null
null
null
cond-mat.mtrl-sci physics.comp-ph
Inverse design of solid-state materials with desired properties represents a formidable challenge in materials science. Although recent generative models have demonstrated potential, their adoption has been hindered by limitations such as inefficiency, architectural constraints and restricted open-source availability. The representation of crystal structures using the SLICES (Simplified Line-Input Crystal-Encoding System) notation as a string of characters enables the use of state-of-the-art natural language processing models, such as Transformers, for crystal design. Drawing inspiration from the success of GPT models in generating coherent text, we trained a generative Transformer on the next-token prediction task to generate solid-state materials with targeted properties. We demonstrate MatterGPT's capability to generate de novo crystal structures with targeted single properties, including both lattice-insensitive (formation energy) and lattice-sensitive (band gap) properties. Furthermore, we extend MatterGPT to simultaneously target multiple properties, addressing the complex challenge of multi-objective inverse design of crystals. Our approach showcases high validity, uniqueness, and novelty in generated structures, as well as the ability to generate materials with properties beyond the training data distribution. This work represents a significant step forward in computational materials discovery, offering a powerful and open tool for designing materials with tailored properties for various applications in energy, electronics, and beyond.
[{'version': 'v1', 'created': 'Wed, 14 Aug 2024 15:12:05 GMT'}]
2024-08-15
Qinyang Li, Nicholas Miklaucic, Jianjun Hu
Out-of-distribution materials property prediction using adversarial learning based fine-tuning
null
null
null
cond-mat.mtrl-sci cs.LG
The accurate prediction of material properties is crucial in a wide range of scientific and engineering disciplines. Machine learning (ML) has advanced the state of the art in this field, enabling scientists to discover novel materials and design materials with specific desired properties. However, one major challenge that persists in material property prediction is the generalization of models to out-of-distribution (OOD) samples,i.e., samples that differ significantly from those encountered during training. In this paper, we explore the application of advancements in OOD learning approaches to enhance the robustness and reliability of material property prediction models. We propose and apply the Crystal Adversarial Learning (CAL) algorithm for OOD materials property prediction,which generates synthetic data during training to bias the training towards those samples with high prediction uncertainty. We further propose an adversarial learning based targeting finetuning approach to make the model adapted to a particular OOD dataset, as an alternative to traditional fine-tuning. Our experiments demonstrate the success of our CAL algorithm with its high effectiveness in ML with limited samples which commonly occurs in materials science. Our work represents a promising direction toward better OOD learning and materials property prediction.
[{'version': 'v1', 'created': 'Sat, 17 Aug 2024 21:22:21 GMT'}]
2024-08-20
Salvatore Romano, Pablo Montero de Hijes, Matthias Meier, Georg Kresse, Cesare Franchini, Christoph Dellago
Structure and dynamics of the magnetite(001)/water interface from molecular dynamics simulations based on a neural network potential
null
null
null
physics.comp-ph cond-mat.mtrl-sci physics.chem-ph
The magnetite/water interface is commonly found in nature and plays a crucial role in various technological applications. However, our understanding of its structural and dynamical properties at the molecular scale remains still limited. In this study, we develop an efficient Behler-Parrinello neural network potential (NNP) for the magnetite/water system, paying particular attention to the accurate generation of reference data with density functional theory. Using this NNP, we performed extensive molecular dynamics simulations of the magnetite (001) surface across a wide range of water coverages, from the single molecule to bulk water. Our simulations revealed several new ground states of low coverage water on the Subsurface Cation Vacancy (SCV) model and yielded a density profile of water at the surface that exhibits marked layering. By calculating mean square displacements, we obtained quantitative information on the diffusion of water molecules on the SCV for different coverages, revealing significant anisotropy. Additionally, our simulations provided qualitative insights into the dissociation mechanisms of water molecules at the surface.
[{'version': 'v1', 'created': 'Wed, 21 Aug 2024 11:33:24 GMT'}, {'version': 'v2', 'created': 'Fri, 6 Sep 2024 09:22:33 GMT'}]
2024-09-09
Xiangxiang Shen, Zheng Wan, Lingfeng Wen, Licheng Sun, Ou Yang Ming Jie, JiJUn Cheng, Xuan Tang, Xian Wei
PDDFormer: Pairwise Distance Distribution Graph Transformer for Crystal Material Property Prediction
null
null
null
cond-mat.mtrl-sci cs.AI
The crystal structure can be simplified as a periodic point set repeating across the entire three-dimensional space along an underlying lattice. Traditionally, methods for representing crystals rely on descriptors like lattice parameters, symmetry, and space groups to characterize the structure. However, in reality, atoms in material always vibrate above absolute zero, causing continuous fluctuations in their positions. This dynamic behavior disrupts the underlying periodicity of the lattice, making crystal graphs based on static lattice parameters and conventional descriptors discontinuous under even slight perturbations. To this end, chemists proposed the Pairwise Distance Distribution (PDD) method, which has been used to distinguish all periodic structures in the world's largest real materials collection, the Cambridge Structural Database. However, achieving the completeness of PDD requires defining a large number of neighboring atoms, resulting in high computational costs. Moreover, it does not account for atomic information, making it challenging to directly apply PDD to crystal material property prediction tasks. To address these challenges, we propose the atom-Weighted Pairwise Distance Distribution (WPDD) and Unit cell Pairwise Distance Distribution (UPDD) for the first time, incorporating them into the construction of multi-edge crystal graphs. Based on this, we further developed WPDDFormer and UPDDFormer, graph transformer architecture constructed using WPDD and UPDD crystal graphs. We demonstrate that this method maintains the continuity and completeness of crystal graphs even under slight perturbations in atomic positions.
[{'version': 'v1', 'created': 'Fri, 23 Aug 2024 11:05:48 GMT'}, {'version': 'v2', 'created': 'Mon, 26 Aug 2024 02:42:23 GMT'}, {'version': 'v3', 'created': 'Sun, 22 Sep 2024 13:35:30 GMT'}, {'version': 'v4', 'created': 'Sun, 24 Nov 2024 08:10:52 GMT'}]
2024-11-26
Saurabh Tiwari, Prathamesh Satpute, Supriyo Ghosh
Time series forecasting of multiphase microstructure evolution using deep learning
Computational Materials Science 247, 113518, 2025
10.1016/j.commatsci.2024.113518
null
cond-mat.mtrl-sci
Microstructure evolution, which plays a critical role in determining materials properties, is commonly simulated by the high-fidelity but computationally expensive phase-field method. To address this, we approximate microstructure evolution as a time series forecasting problem within the domain of deep learning. Our approach involves implementing a cost-effective surrogate model that accurately predicts the spatiotemporal evolution of microstructures, taking an example of spinodal decomposition in binary and ternary mixtures. Our surrogate model combines a convolutional autoencoder to reduce the dimensional representation of these microstructures with convolutional recurrent neural networks to forecast their temporal evolution. We use different variants of recurrent neural networks to compare their efficacy in developing surrogate models for phase-field predictions. On average, our deep learning framework demonstrates excellent accuracy and speedup relative to the "ground truth" phase-field simulations. We use quantitative measures to demonstrate how surrogate model predictions can effectively replace the phase-field timesteps without compromising accuracy in predicting the long-term evolution trajectory. Additionally, by emulating a transfer learning approach, our framework performs satisfactorily in predicting new microstructures resulting from alloy composition and physics unknown to the model. Therefore, our approach offers a useful data-driven alternative and accelerator to the materials microstructure simulation workflow.
[{'version': 'v1', 'created': 'Thu, 22 Aug 2024 06:14:06 GMT'}, {'version': 'v2', 'created': 'Thu, 21 Nov 2024 11:32:58 GMT'}]
2024-11-22
Harikrishnan Vijayakumaran, Jonathan B. Russ, Glaucio H. Paulino, Miguel A. Bessa
Consistent machine learning for topology optimization with microstructure-dependent neural network material models
null
null
null
cond-mat.mtrl-sci cs.LG cs.NA math.NA
Additive manufacturing methods together with topology optimization have enabled the creation of multiscale structures with controlled spatially-varying material microstructure. However, topology optimization or inverse design of such structures in the presence of nonlinearities remains a challenge due to the expense of computational homogenization methods and the complexity of differentiably parameterizing the microstructural response. A solution to this challenge lies in machine learning techniques that offer efficient, differentiable mappings between the material response and its microstructural descriptors. This work presents a framework for designing multiscale heterogeneous structures with spatially varying microstructures by merging a homogenization-based topology optimization strategy with a consistent machine learning approach grounded in hyperelasticity theory. We leverage neural architectures that adhere to critical physical principles such as polyconvexity, objectivity, material symmetry, and thermodynamic consistency to supply the framework with a reliable constitutive model that is dependent on material microstructural descriptors. Our findings highlight the potential of integrating consistent machine learning models with density-based topology optimization for enhancing design optimization of heterogeneous hyperelastic structures under finite deformations.
[{'version': 'v1', 'created': 'Sun, 25 Aug 2024 14:17:43 GMT'}, {'version': 'v2', 'created': 'Tue, 27 Aug 2024 14:24:52 GMT'}]
2024-08-28
Fanjie Xu, Wentao Guo, Feng Wang, Lin Yao, Hongshuai Wang, Fujie Tang, Zhifeng Gao, Linfeng Zhang, Weinan E, Zhong-Qun Tian, Jun Cheng
Towards a Unified Benchmark and Framework for Deep Learning-Based Prediction of Nuclear Magnetic Resonance Chemical Shifts
null
null
null
physics.comp-ph cond-mat.dis-nn cond-mat.mtrl-sci physics.chem-ph
The study of structure-spectrum relationships is essential for spectral interpretation, impacting structural elucidation and material design. Predicting spectra from molecular structures is challenging due to their complex relationships. Herein, we introduce NMRNet, a deep learning framework using the SE(3) Transformer for atomic environment modeling, following a pre-training and fine-tuning paradigm. To support the evaluation of NMR chemical shift prediction models, we have established a comprehensive benchmark based on previous research and databases, covering diverse chemical systems. Applying NMRNet to these benchmark datasets, we achieve state-of-the-art performance in both liquid-state and solid-state NMR datasets, demonstrating its robustness and practical utility in real-world scenarios. This marks the first integration of solid and liquid state NMR within a unified model architecture, highlighting the need for domainspecific handling of different atomic environments. Our work sets a new standard for NMR prediction, advancing deep learning applications in analytical and structural chemistry.
[{'version': 'v1', 'created': 'Wed, 28 Aug 2024 10:11:00 GMT'}]
2024-08-29
Xiuying Zhang, Linqiang Xu, Jing Lu, Zhaofu Zhang, and Lei Shen
Physics-integrated Neural Network for Quantum Transport Prediction of Field-effect Transistor
null
null
null
cond-mat.dis-nn cond-mat.mtrl-sci physics.comp-ph
Quantum-mechanics-based transport simulation is of importance for the design of ultra-short channel field-effect transistors (FETs) with its capability of understanding the physical mechanism, while facing the primary challenge of the high computational intensity. Traditional machine learning is expected to accelerate the optimization of FET design, yet its application in this field is limited by the lack of both high-fidelity datasets and the integration of physical knowledge. Here, we introduced a physics-integrated neural network framework to predict the transport curves of sub-5-nm gate-all-around (GAA) FETs using an in-house developed high-fidelity database. The transport curves in the database are collected from literature and our first-principles calculations. Beyond silicon, we included indium arsenide, indium phosphide, and selenium nanowires with different structural phases as the FET channel materials. Then, we built a physical-knowledge-integrated hyper vector neural network (PHVNN), in which five new physical features were added into the inputs for prediction transport characteristics, achieving a sufficiently low mean absolute error of 0.39. In particular, ~98% of the current prediction residuals are within one order of magnitude. Using PHVNN, we efficiently screened out the symmetric p-type GAA FETs that possess the same figures of merit with the n-type ones, which are crucial for the fabrication of homogeneous CMOS circuits. Finally, our automatic differentiation analysis provides interpretable insights into the PHVNN, which highlights the important contributions of our new input parameters and improves the reliability of PHVNN. Our approach provides an effective method for rapidly screening appropriate GAA FETs with the prospect of accelerating the design process of next-generation electronic devices.
[{'version': 'v1', 'created': 'Fri, 30 Aug 2024 05:38:12 GMT'}]
2024-09-02
Alexander New, Nam Q. Le, Michael J. Pekala, Christopher D. Stiles
Self-supervised learning for crystal property prediction via denoising
null
null
null
cs.LG cond-mat.mtrl-sci
Accurate prediction of the properties of crystalline materials is crucial for targeted discovery, and this prediction is increasingly done with data-driven models. However, for many properties of interest, the number of materials for which a specific property has been determined is much smaller than the number of known materials. To overcome this disparity, we propose a novel self-supervised learning (SSL) strategy for material property prediction. Our approach, crystal denoising self-supervised learning (CDSSL), pretrains predictive models (e.g., graph networks) with a pretext task based on recovering valid material structures when given perturbed versions of these structures. We demonstrate that CDSSL models out-perform models trained without SSL, across material types, properties, and dataset sizes.
[{'version': 'v1', 'created': 'Fri, 30 Aug 2024 12:53:40 GMT'}]
2024-09-02
Tsz Wai Ko and Shyue Ping Ong
Data-Efficient Construction of High-Fidelity Graph Deep Learning Interatomic Potentials
null
null
null
physics.comp-ph cond-mat.mtrl-sci physics.chem-ph
Machine learning potentials (MLPs) have become an indispensable tool in large-scale atomistic simulations because of their ability to reproduce ab initio potential energy surfaces (PESs) very accurately at a fraction of computational cost. For computational efficiency, the training data for most MLPs today are computed using relatively cheap density functional theory (DFT) methods such as the Perdew-Burke-Ernzerhof (PBE) generalized gradient approximation (GGA) functional. Meta-GGAs such as the recently developed strongly constrained and appropriately normed (SCAN) functional have been shown to yield significantly improved descriptions of atomic interactions for diversely bonded systems, but their higher computational cost remains an impediment to their use in MLP development. In this work, we outline a data-efficient multi-fidelity approach to constructing Materials 3-body Graph Network (M3GNet) interatomic potentials that integrate different levels of theory within a single model. Using silicon and water as examples, we show that a multi-fidelity M3GNet model trained on a combined dataset of low-fidelity GGA calculations with 10% of high-fidelity SCAN calculations can achieve accuracies comparable to a single-fidelity M3GNet model trained on a dataset comprising 8x the number of SCAN calculations. This work paves the way for the development of high-fidelity MLPs in a cost-effective manner by leveraging existing low-fidelity datasets.
[{'version': 'v1', 'created': 'Mon, 2 Sep 2024 05:57:32 GMT'}]
2024-09-04
Zirui Zhao, Xiaoke Wang, Si Wu, Pengfei Zhou, Qian Zhao, Guanping Xu, Kaitong Sun, Hai-Feng Li
Deep learning-driven evaluation and prediction of ion-doped NASICON materials for enhanced solid-state battery performance
AAPPS Bulletin, 2024, 34(1): 26
10.1007/s43673-024-00131-9
null
cond-mat.mtrl-sci
We developed a convolutional neural network (CNN) model capable of predicting the performance of various ion-doped NASICON compounds by leveraging extensive datasets from prior experimental investigation.The model demonstrated high accuracy and efficiency in predicting ionic conductivity and electrochemical properties. Key findings include the successful synthesis and validation of three NASICON materials predicted by the model, with experimental results closely matching the model predictions. This research not only enhances the understanding of ion-doping effects in NASICON materials but also establishes a robust framework for material design and practical applications. It bridges the gap between theoretical predictions and experimental validations.
[{'version': 'v1', 'created': 'Mon, 2 Sep 2024 02:20:44 GMT'}, {'version': 'v2', 'created': 'Mon, 9 Sep 2024 02:46:15 GMT'}]
2025-01-13
Koki Ueno, Satoru Ohuchi, Kazuhide Ichikawa, Kei Amii, Kensuke Wakasugi
SpinMultiNet: Neural Network Potential Incorporating Spin Degrees of Freedom with Multi-Task Learning
null
null
null
cond-mat.mtrl-sci cs.LG
Neural Network Potentials (NNPs) have attracted significant attention as a method for accelerating density functional theory (DFT) calculations. However, conventional NNP models typically do not incorporate spin degrees of freedom, limiting their applicability to systems where spin states critically influence material properties, such as transition metal oxides. This study introduces SpinMultiNet, a novel NNP model that integrates spin degrees of freedom through multi-task learning. SpinMultiNet achieves accurate predictions without relying on correct spin values obtained from DFT calculations. Instead, it utilizes initial spin estimates as input and leverages multi-task learning to optimize the spin latent representation while maintaining both $E(3)$ and time-reversal equivariance. Validation on a dataset of transition metal oxides demonstrates the high predictive accuracy of SpinMultiNet. The model successfully reproduces the energy ordering of stable spin configurations originating from superexchange interactions and accurately captures the rhombohedral distortion of the rocksalt structure. These results pave the way for new possibilities in materials simulations that consider spin degrees of freedom, promising future applications in large-scale simulations of various material systems, including magnetic materials.
[{'version': 'v1', 'created': 'Thu, 5 Sep 2024 05:13:28 GMT'}, {'version': 'v2', 'created': 'Sun, 8 Sep 2024 23:58:44 GMT'}]
2024-09-10
Wei Lu and Rachel K. Luu and Markus J. Buehler
Fine-tuning large language models for domain adaptation: Exploration of training strategies, scaling, model merging and synergistic capabilities
null
null
null
cs.CL cond-mat.mtrl-sci cs.AI
The advancement of Large Language Models (LLMs) for domain applications in fields such as materials science and engineering depends on the development of fine-tuning strategies that adapt models for specialized, technical capabilities. In this work, we explore the effects of Continued Pretraining (CPT), Supervised Fine-Tuning (SFT), and various preference-based optimization approaches, including Direct Preference Optimization (DPO) and Odds Ratio Preference Optimization (ORPO), on fine-tuned LLM performance. Our analysis shows how these strategies influence model outcomes and reveals that the merging of multiple fine-tuned models can lead to the emergence of capabilities that surpass the individual contributions of the parent models. We find that model merging leads to new functionalities that neither parent model could achieve alone, leading to improved performance in domain-specific assessments. Experiments with different model architectures are presented, including Llama 3.1 8B and Mistral 7B models, where similar behaviors are observed. Exploring whether the results hold also for much smaller models, we use a tiny LLM with 1.7 billion parameters and show that very small LLMs do not necessarily feature emergent capabilities under model merging, suggesting that model scaling may be a key component. In open-ended yet consistent chat conversations between a human and AI models, our assessment reveals detailed insights into how different model variants perform and show that the smallest model achieves a high intelligence score across key criteria including reasoning depth, creativity, clarity, and quantitative precision. Other experiments include the development of image generation prompts based on disparate biological material design concepts, to create new microstructures, architectural concepts, and urban design based on biological materials-inspired construction principles.
[{'version': 'v1', 'created': 'Thu, 5 Sep 2024 11:49:53 GMT'}]
2024-09-06
Abdelwahab Kawafi, Lars K\"urten, Levke Ortlieb, Yushi Yang, Abraham Mauleon Amieva, James E. Hallett and C.Patrick Royall
Colloidoscope: Detecting Dense Colloids in 3d with Deep Learning
null
null
null
cond-mat.soft cond-mat.mtrl-sci cond-mat.stat-mech
Colloidoscope is a deep learning pipeline employing a 3D residual Unet architecture, designed to enhance the tracking of dense colloidal suspensions through confocal microscopy. This methodology uses a simulated training dataset that reflects a wide array of real-world imaging conditions, specifically targeting high colloid volume fraction and low-contrast scenarios where traditional detection methods struggle. Central to our approach is the use of experimental signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR), and point-spread-functions (PSFs) to accurately quantify and simulate the experimental data. Our findings reveal that Colloidoscope achieves superior recall in particle detection (finds more particles) compared to conventional heuristic methods. Simultaneously, high precision is maintained (high fraction of true positives.) The model demonstrates a notable robustness to photobleached samples, thereby prolonging the imaging time and number of frames than may be acquired. Furthermore, Colloidoscope maintains small scale resolution sufficient to classify local structural motifs. Evaluated across both simulated and experimental datasets, Colloidoscope brings the advancements in computer vision offered by deep learning to particle tracking at high volume fractions. We offer a promising tool for researchers in the soft matter community, this model is deployed and available to use pretrained: https://github.com/wahabk/colloidoscope.
[{'version': 'v1', 'created': 'Fri, 6 Sep 2024 20:21:33 GMT'}]
2024-09-10
Ayush Jain, Rishi Gurnani, Arunkumar Rajan, H. Jerry Qi, Rampi Ramprasad
A Physics-Enforced Neural Network to Predict Polymer Melt Viscosity
null
10.1038/s41524-025-01532-6
null
cs.CE cond-mat.mtrl-sci
Achieving superior polymeric components through additive manufacturing (AM) relies on precise control of rheology. One key rheological property particularly relevant to AM is melt viscosity ($\eta$). Melt viscosity is influenced by polymer chemistry, molecular weight ($M_w$), polydispersity, induced shear rate ($\dot\gamma$), and processing temperature ($T$). The relationship of $\eta$ with $M_w$, $\dot\gamma$, and $T$ may be captured by parameterized equations. Several physical experiments are required to fit the parameters, so predicting $\eta$ of a new polymer material in unexplored physical domains is a laborious process. Here, we develop a Physics-Enforced Neural Network (PENN) model that predicts the empirical parameters and encodes the parametrized equations to calculate $\eta$ as a function of polymer chemistry, $M_w$, polydispersity, $\dot\gamma$, and $T$. We benchmark our PENN against physics-unaware Artificial Neural Network (ANN) and Gaussian Process Regression (GPR) models. Finally, we demonstrate that the PENN offers superior values of $\eta$ when extrapolating to unseen values of $M_w$, $\dot\gamma$, and $T$ for sparsely seen polymers.
[{'version': 'v1', 'created': 'Sun, 8 Sep 2024 22:52:24 GMT'}]
2025-04-25
Nicholas Beaver, Aniruddha Dive, Marina Wong, Keita Shimanuki, Ananya Patil, Anthony Ferrell, Mohsen B. Kivy
Rapid Assessment of Stable Crystal Structures in Single Phase High Entropy Alloys Via Graph Neural Network Based Surrogate Modelling
null
null
null
cond-mat.mtrl-sci cond-mat.dis-nn
In an effort to develop a rapid, reliable, and cost-effective method for predicting the structure of single-phase high entropy alloys, a Graph Neural Network (ALIGNN-FF) based approach was introduced. This method was successfully tested on 132 different high entropy alloys, and the results were analyzed and compared with density functional theory and valence electron concentration calculations. Additionally, the effects of various factors, including lattice parameters and the number of supercells with unique atomic configurations, on the prediction accuracy were investigated. The ALIGNN-FF based approach was subsequently used to predict the structure of a novel cobalt-free 3d high entropy alloy, and the result was experimentally verified.
[{'version': 'v1', 'created': 'Wed, 11 Sep 2024 23:34:48 GMT'}]
2024-09-13
Jun Li, Wenqi Fang, Shangjian Jin, Tengdong Zhang, Yanling Wu, Xiaodan Xu, Yong Liu and Dao-Xin Yao
A deep learning approach to search for superconductors from electronic bands
null
null
null
cond-mat.supr-con cond-mat.mtrl-sci
Energy band theory is a foundational framework in condensed matter physics. In this work, we employ a deep learning method, BNAS, to find a direct correlation between electronic band structure and superconducting transition temperature. Our findings suggest that electronic band structures can act as primary indicators of superconductivity. To avoid overfitting, we utilize a relatively simple deep learning neural network model, which, despite its simplicity, demonstrates predictive capabilities for superconducting properties. By leveraging the attention mechanism within deep learning, we are able to identify specific regions of the electronic band structure most correlated with superconductivity. This novel approach provides new insights into the mechanisms driving superconductivity from an alternative perspective. Moreover, we predict several potential superconductors that may serve as candidates for future experimental synthesis.
[{'version': 'v1', 'created': 'Thu, 12 Sep 2024 03:02:59 GMT'}]
2024-09-13
Xiao-Qi Han, Zhenfeng Ouyang, Peng-Jie Guo, Hao Sun, Ze-Feng Gao and Zhong-Yi Lu
InvDesFlow: An AI-driven materials inverse design workflow to explore possible high-temperature superconductors
Chin. Phys. Lett. 2025,42(4): 047301
10.1088/0256-307X/42/4/047301
null
cond-mat.supr-con cond-mat.mtrl-sci cs.AI physics.comp-ph
The discovery of new superconducting materials, particularly those exhibiting high critical temperature ($T_c$), has been a vibrant area of study within the field of condensed matter physics. Conventional approaches primarily rely on physical intuition to search for potential superconductors within the existing databases. However, the known materials only scratch the surface of the extensive array of possibilities within the realm of materials. Here, we develop InvDesFlow, an AI search engine that integrates deep model pre-training and fine-tuning techniques, diffusion models, and physics-based approaches (e.g., first-principles electronic structure calculation) for the discovery of high-$T_c$ superconductors. Utilizing InvDesFlow, we have obtained 74 dynamically stable materials with critical temperatures predicted by the AI model to be $T_c \geq$ 15 K based on a very small set of samples. Notably, these materials are not contained in any existing dataset. Furthermore, we analyze trends in our dataset and individual materials including B$_4$CN$_3$ (at 5 GPa) and B$_5$CN$_2$ (at ambient pressure) whose $T_c$s are 24.08 K and 15.93 K, respectively. We demonstrate that AI technique can discover a set of new high-$T_c$ superconductors, outline its potential for accelerating discovery of the materials with targeted properties.
[{'version': 'v1', 'created': 'Thu, 12 Sep 2024 14:16:56 GMT'}, {'version': 'v2', 'created': 'Mon, 2 Dec 2024 14:29:14 GMT'}, {'version': 'v3', 'created': 'Tue, 13 May 2025 08:22:00 GMT'}]
2025-05-14
Israrul H. Hashmi, Himanshu, Rahul Karmakar and Tarak K Patra
Extrapolative ML Models for Copolymers
null
null
null
cond-mat.soft cond-mat.mtrl-sci cs.LG
Machine learning models have been progressively used for predicting materials properties. These models can be built using pre-existing data and are useful for rapidly screening the physicochemical space of a material, which is astronomically large. However, ML models are inherently interpolative, and their efficacy for searching candidates outside a material's known range of property is unresolved. Moreover, the performance of an ML model is intricately connected to its learning strategy and the volume of training data. Here, we determine the relationship between the extrapolation ability of an ML model, the size and range of its training dataset, and its learning approach. We focus on a canonical problem of predicting the properties of a copolymer as a function of the sequence of its monomers. Tree search algorithms, which learn the similarity between polymer structures, are found to be inefficient for extrapolation. Conversely, the extrapolation capability of neural networks and XGBoost models, which attempt to learn the underlying functional correlation between the structure and property of polymers, show strong correlations with the volume and range of training data. These findings have important implications on ML-based new material development.
[{'version': 'v1', 'created': 'Sun, 15 Sep 2024 11:02:01 GMT'}]
2024-09-17
Shaswat Mohanty, Yifan Wang, Wei Cai
Generalizability of Graph Neural Network Force Fields for Predicting Solid-State Properties
null
null
null
cs.LG cond-mat.mtrl-sci cs.NA math.NA
Machine-learned force fields (MLFFs) promise to offer a computationally efficient alternative to ab initio simulations for complex molecular systems. However, ensuring their generalizability beyond training data is crucial for their wide application in studying solid materials. This work investigates the ability of a graph neural network (GNN)-based MLFF, trained on Lennard-Jones Argon, to describe solid-state phenomena not explicitly included during training. We assess the MLFF's performance in predicting phonon density of states (PDOS) for a perfect face-centered cubic (FCC) crystal structure at both zero and finite temperatures. Additionally, we evaluate vacancy migration rates and energy barriers in an imperfect crystal using direct molecular dynamics (MD) simulations and the string method. Notably, vacancy configurations were absent from the training data. Our results demonstrate the MLFF's capability to capture essential solid-state properties with good agreement to reference data, even for unseen configurations. We further discuss data engineering strategies to enhance the generalizability of MLFFs. The proposed set of benchmark tests and workflow for evaluating MLFF performance in describing perfect and imperfect crystals pave the way for reliable application of MLFFs in studying complex solid-state materials.
[{'version': 'v1', 'created': 'Mon, 16 Sep 2024 02:14:26 GMT'}, {'version': 'v2', 'created': 'Sat, 21 Dec 2024 16:21:51 GMT'}]
2024-12-24
Amir Omranpour and J\"org Behler
A High-Dimensional Neural Network Potential for Co$_3$O$_4$
null
null
null
cond-mat.mtrl-sci
The Co$_3$O$_4$ spinel is an important material in oxidation catalysis. Its properties under catalytic conditions, i.e., at finite temperatures, can be studied by molecular dynamics simulations, which critically depend on an accurate description of the atomic interactions. Due to the high complexity of Co$_3$O$_4$, which is related to the presence of multiple oxidation states of the cobalt ions, to date \textit{ab initio} methods have been essentially the only way to reliably capture the underlying potential energy surface, while more efficient atomistic potentials are very challenging to construct. Consequently, the accessible length and time scales of computer simulations of systems containing Co$_3$O$_4$ are still severely limited. Rapid advances in the development of modern machine learning potentials (MLPs) trained on electronic structure data now make it possible to bridge this gap. In this work, we employ a high-dimensional neural network potential (HDNNP) to construct a MLP for bulk Co$_3$O$_4$ spinel based on density functional theory calculations. After a careful validation of the potential, we compute various structural, vibrational, and dynamical properties of the Co$_3$O$_4$ spinel with a particular focus on its temperature-dependent behavior, including the thermal expansion coefficient.
[{'version': 'v1', 'created': 'Tue, 17 Sep 2024 10:02:27 GMT'}]
2024-09-18
Luke P. J. Gilligan, Matteo Cobelli, Hasan M. Sayeed, Taylor D. Sparks and Stefano Sanvito
Sampling Latent Material-Property Information From LLM-Derived Embedding Representations
null
null
null
cs.CL cond-mat.mtrl-sci
Vector embeddings derived from large language models (LLMs) show promise in capturing latent information from the literature. Interestingly, these can be integrated into material embeddings, potentially useful for data-driven predictions of materials properties. We investigate the extent to which LLM-derived vectors capture the desired information and their potential to provide insights into material properties without additional training. Our findings indicate that, although LLMs can be used to generate representations reflecting certain property information, extracting the embeddings requires identifying the optimal contextual clues and appropriate comparators. Despite this restriction, it appears that LLMs still have the potential to be useful in generating meaningful materials-science representations.
[{'version': 'v1', 'created': 'Wed, 18 Sep 2024 13:22:04 GMT'}]
2024-09-19
Jaime A. Berkovich and Markus J. Buehler
LifeGPT: Topology-Agnostic Generative Pretrained Transformer Model for Cellular Automata
null
null
null
cs.AI cond-mat.mtrl-sci cond-mat.stat-mech math.DS
Conway's Game of Life (Life), a well known algorithm within the broader class of cellular automata (CA), exhibits complex emergent dynamics, with extreme sensitivity to initial conditions. Modeling and predicting such intricate behavior without explicit knowledge of the system's underlying topology presents a significant challenge, motivating the development of algorithms that can generalize across various grid configurations and boundary conditions. We develop a decoder-only generative pretrained transformer (GPT) model to solve this problem, showing that our model can simulate Life on a toroidal grid with no prior knowledge on the size of the grid, or its periodic boundary conditions (LifeGPT). LifeGPT is topology-agnostic with respect to its training data and our results show that a GPT model is capable of capturing the deterministic rules of a Turing-complete system with near-perfect accuracy, given sufficiently diverse training data. We also introduce the idea of an `autoregressive autoregressor' to recursively implement Life using LifeGPT. Our results pave the path towards true universal computation within a large language model framework, synthesizing of mathematical analysis with natural language processing, and probing AI systems for situational awareness about the evolution of such algorithms without ever having to compute them. Similar GPTs could potentially solve inverse problems in multicellular self-assembly by extracting CA-compatible rulesets from real-world biological systems to create new predictive models, which would have significant consequences for the fields of bioinspired materials, tissue engineering, and architected materials design.
[{'version': 'v1', 'created': 'Tue, 3 Sep 2024 11:43:16 GMT'}, {'version': 'v2', 'created': 'Thu, 17 Oct 2024 16:55:02 GMT'}]
2024-10-18
Teng Long, Yixuan Zhang, Hongbin Zhang
Generative deep learning for the inverse design of materials
null
null
null
cond-mat.mtrl-sci physics.comp-ph
In addition to the forward inference of materials properties using machine learning, generative deep learning techniques applied on materials science allow the inverse design of materials, i.e., assessing the composition-processing-(micro-)structure-property relationships in a reversed way. In this review, we focus on the (micro-)structure-property mapping, i.e., crystal structure-intrinsic property and microstructure-extrinsic property, and summarize comprehensively how generative deep learning can be performed. Three key elements, i.e., the construction of latent spaces for both the crystal structures and microstructures, generative learning approaches, and property constraints, are discussed in detail. A perspective is given outlining the challenges of the existing methods in terms of computational resource consumption, data compatibility, and yield of generation.
[{'version': 'v1', 'created': 'Fri, 27 Sep 2024 20:10:19 GMT'}]
2024-10-01