Crystal structure prediction using age-fitness multi-objective genetic algorithm and coordination number constraints
Wenhui Yang, Edirisuriya M. Dilanga Siriwardane, Jianjun Hu

et al.

arXiv (Cornell University), Journal Year: 2021, Volume and Issue: unknown

Published: July 3, 2021

Crystal structure prediction (CSP) has emerged as one of the most important approaches for discovering new materials. CSP algorithms based on evolutionary and particle swarm optimization have discovered a great number However, these ab initio calculation free energy are inefficient. Moreover, they severe limitations in terms scalability. We recently proposed promising crystal method atomic contact maps, using global to search Wyckoff positions by maximizing match between map predicted true structure. our previous two major limitations: (1) loss capability due getting trapped local optima; (2) it only uses connection atoms unit cell predict structure, ignoring chemical environment outside cell, which may lead unreasonable coordination environments. Herein we propose novel multi-objective genetic map-based optimizing three objectives, including accuracy, individual age, match. Furthermore, assign age values all individuals GA try minimize aiming avoid premature convergence problem. Our experimental results show that compared CMCrystal algorithm, algorithm (CMCrystalMOO) can reconstruct with higher quality alleviate problem convergence.

Language: Английский

End‐to‐End Crystal Structure Prediction from Powder X‐Ray Diffraction DOI Creative Commons

Qingsi Lai,

Fanjie Xu,

Lin Yao

et al.

Advanced Science, Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 4, 2025

Abstract Powder X‐ray diffraction (PXRD) is a prevalent technique in materials characterization. While the analysis of PXRD often requires extensive human manual intervention, and most automated method only achieved at coarse‐grained level. The more difficult important task fine‐grained crystal structure prediction from remains unaddressed. This study introduces XtalNet, first equivariant deep generative model for end‐to‐end PXRD. Unlike previous methods that rely solely on composition, XtalNet leverages as an additional condition, eliminating ambiguity enabling generation complex organic structures with up to 400 atoms unit cell. comprises two modules: Contrastive PXRD‐Crystal Pretraining (CPCP) module aligns space space, Conditional Crystal Structure Generation (CCSG) generates candidate conditioned patterns. Evaluation MOF datasets (hMOF‐100 hMOF‐400) demonstrates XtalNet's effectiveness. achieves top‐10 Match Rate 90.2% 79% hMOF‐100 hMOF‐400 conditional task, respectively. enables direct experimental measurements, need intervention external databases. opens new possibilities determination accelerated discovery novel materials.

Language: Английский

Citations

2
MatGPT: A Vane of Materials Informatics from Past, Present, to Future DOI
Zhilong Wang, An Chen, Kehao Tao

et al.

Advanced Materials, Journal Year: 2023, Volume and Issue: 36(6)

Published: Oct. 10, 2023

Abstract Combining materials science, artificial intelligence (AI), physical chemistry, and other disciplines, informatics is continuously accelerating the vigorous development of new materials. The emergence “GPT (Generative Pre‐trained Transformer) AI” shows that scientific research field has entered era intelligent civilization with “data” as basic factor “algorithm + computing power” core productivity. continuous innovation AI will impact cognitive laws methods, reconstruct knowledge wisdom system. This leads to think more about informatics. Here, a comprehensive discussion models infrastructures provided, advances in discovery design are reviewed. With rise paradigms triggered by “AI for Science”, vane informatics: “MatGPT”, proposed technical path planning from aspects data, descriptors, generative models, pretraining directed collaborative training, experimental robots, well efforts preparations needed develop generation informatics, carried out. Finally, challenges constraints faced discussed, order achieve digital, intelligent, automated construction joint interdisciplinary scientists.

Language: Английский

Citations

30
Mlatticeabc: Generic Lattice Constant Prediction of Crystal Materials Using Machine Learning DOI Creative Commons
Yuxin Li, Wenhui Yang, Rongzhi Dong

et al.

ACS Omega, Journal Year: 2021, Volume and Issue: 6(17), P. 11585 - 11594

Published: April 20, 2021

Lattice constants such as unit cell edge lengths and plane angles are important parameters of the periodic structures crystal materials. Predicting lattice has wide applications in structure prediction materials property prediction. Previous work used machine learning models neural networks support vector machines combined with composition features for constant achieved a maximum performance cubic an average coefficient determination (R2) 0.82. Other tailored special family fixed form ABX3 perovskites can achieve much higher due to homogeneity structures. However, these trained small data sets usually not applicable generic parameter diverse compositions. Herein, we report MLatticeABC, random forest model new descriptor set length (a, b, c) which achieves R2 score 0.973 crystals 0.80 all systems. The scores between 0.498 0.757 over b c model, could be by just inputting molecular formula material get constants. Our results also show significant improvement angle predictions. Source code freely accessed at https://github.com/usccolumbia/MLatticeABC.

Language: Английский

Citations

49
Reflections on one million compounds in the open quantum materials database (OQMD) DOI Creative Commons
Jiahong Shen, Sean D. Griesemer, Abhijith Gopakumar

et al.

Journal of Physics Materials, Journal Year: 2022, Volume and Issue: 5(3), P. 031001 - 031001

Published: June 23, 2022

Abstract Density functional theory (DFT) has been widely applied in modern materials discovery and many databases, including the open quantum database (OQMD), contain large collections of calculated DFT properties experimentally known crystal structures hypothetical predicted compounds. Since beginning OQMD late 2010, over one million compounds have now stored database, which is constantly used by worldwide researchers advancing studies. The growth depends on project-based high-throughput calculations, structure-based projects, property-based most recently, machine-learning-based projects. Another major goal to ensure openness its data public developers are working with other databases reach a universal querying protocol support FAIR principles.

Language: Английский

Citations

30
Composition based crystal materials symmetry prediction using machine learning with enhanced descriptors DOI
Yuxin Li, Rongzhi Dong, Wenhui Yang

et al.

Computational Materials Science, Journal Year: 2021, Volume and Issue: 198, P. 110686 - 110686

Published: July 6, 2021

Language: Английский

Citations

29
TCSP: a Template-Based Crystal Structure Prediction Algorithm for Materials Discovery DOI
Lai Wei, Nihang Fu, Edirisuriya M. Dilanga Siriwardane

et al.

Inorganic Chemistry, Journal Year: 2022, Volume and Issue: 61(22), P. 8431 - 8439

Published: April 14, 2022

Fast and accurate crystal structure prediction (CSP) algorithms web servers are highly desirable for the exploration discovery of new materials out infinite chemical design space. However, currently, computationally expensive first-principles calculation-based CSP applicable to relatively small systems reach most researchers. Several teams have used an element substitution approach generating or predicting structures, but usually in ad hoc way. Here we develop a template-based (TCSP) algorithm its companion server, which makes this tool accessible all Our uses elemental/chemical similarity oxidation states guide selection template structures then rank them based on compatibility can return multiple predictions with ranking scores few minutes. A benchmark study 98290 formulas Materials Project database using leave-one-out evaluation shows that our achieve high accuracy (for 13145 target TCSP predicted their root-mean-square deviation < 0.1) large portion formulas. We also discover Ga-B-N system, showing potential high-throughput discovery. user-friendly app be accessed freely at www.materialsatlas.org/crystalstructure MaterialsAtlas.org platform.

Language: Английский

Citations

21
Deep learning-assisted methods for accelerating the intelligent screening of novel 2D materials: New perspectives focusing on data collection and description DOI

Yuandong Lin,

Ji Ma, Yong‐Guang Jia

et al.

Coordination Chemistry Reviews, Journal Year: 2025, Volume and Issue: 529, P. 216436 - 216436

Published: Jan. 16, 2025

Language: Английский

Citations

0
Search methods for inorganic materials crystal structure prediction DOI Creative Commons
Xiangyu Yin, Chrysanthos E. Gounaris

Current Opinion in Chemical Engineering, Journal Year: 2021, Volume and Issue: 35, P. 100726 - 100726

Published: Oct. 8, 2021

Crystal structure prediction (CSP) is the problem of determining most stable crystalline arrangements materials given their chemical compositions. In general, CSP methodologies include two algorithmic steps, namely a method for assessing material stability any design, and search algorithm exploring design space. For inorganic crystals, in particular, critical aspect to develop an effective algorithm. This paper summarizes previous research discusses recent progress methods developed CSP. Empirical methods, guided-sampling algorithms, more data-driven approaches are discussed. Additionally, we describe mathematical optimization-based paradigm that has been recently introduced as alternative approach. A semiconductor nanowire approach then presented illustrate this paradigm.

Language: Английский

Citations

21
Comparative study of crystal structure prediction approaches based on a graph network and an optimization algorithm DOI Open Access

Fan Yang,

Guanjian Cheng, Wan‐Jian Yin

et al.

Science China Materials, Journal Year: 2024, Volume and Issue: 67(4), P. 1273 - 1281

Published: March 25, 2024

Language: Английский

Citations

3
Crystal structure prediction of materials with high symmetry using differential evolution DOI
Wenhui Yang, Edirisuriya M. Dilanga Siriwardane, Rongzhi Dong

et al.

Journal of Physics Condensed Matter, Journal Year: 2021, Volume and Issue: 33(45), P. 455902 - 455902

Published: Aug. 13, 2021

Crystal structure determines properties of materials. With the crystal a chemical substance, many physical and can be predicted by first-principles calculations or machine learning models. Since it is relatively easy to generate hypothetical chemically valid formula, prediction becomes an important method for discovering new In our previous work, we proposed contact map-based method, which uses global optimization algorithms such as genetic maximize match between map real search coordinates at Wyckoff Positions(WP). However, when predicting with high symmetry, found that algorithm has difficulty find effective combination WPs satisfies mainly caused inconsistency dimensionality target structure. This makes challenging predict structures high-symmetry crystals. order solve this problem, here propose use PyXtal filter random given symmetry constraints based on information formulas space groups. goal, differential evolution non-special positions realize Our experimental results show CMCrystalHS effectively problem inconsistent dimensions symmetry.

Language: Английский

Citations

18