Active learning to overcome exponential-wall problem for effective structure prediction of chemical-disordered materials
Abstract Chemical-disordered materials have a wide range of applications whereas the determination of their structures or configurations is one of the most important and challenging problems. Traditional methods are extremely inefficient or intractable for large systems due to the notorious exponent...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Nature Portfolio
2023-01-01
|
| Series: | npj Computational Materials |
| Online Access: | https://doi.org/10.1038/s41524-023-00967-z |
| _version_ | 1828056821404794880 |
|---|---|
| author | Xiaoze Yuan Yuwei Zhou Qing Peng Yong Yang Yongwang Li Xiaodong Wen |
| author_facet | Xiaoze Yuan Yuwei Zhou Qing Peng Yong Yang Yongwang Li Xiaodong Wen |
| author_sort | Xiaoze Yuan |
| collection | DOAJ |
| description | Abstract Chemical-disordered materials have a wide range of applications whereas the determination of their structures or configurations is one of the most important and challenging problems. Traditional methods are extremely inefficient or intractable for large systems due to the notorious exponential-wall issue that the number of possible structures increase exponentially for N-body systems. Herein, we introduce an efficient approach to predict the thermodynamically stable structures of chemical-disordered materials via active-learning accompanied by first-principles calculations. Our method, named LAsou, can efficiently compress the sampling space and dramatically reduce the computational cost. Three distinct and typical finite-size systems are investigated, including the anion-disordered BaSc(OxF1−x)3 (x = 0.667), the cation-disordered Ca1−xMnxCO3 (x = 0.25) with larger size and the defect-disordered ε-FeCx (x = 0.5) with larger space. The commonly used enumeration method requires to explicitly calculate 2664, 1033, and 10496 configurations, respectively, while the LAsou method just needs to explicitly calculate about 15, 20, and 10 configurations, respectively. Besides the finite-size system, our LAsou method is ready for quasi-infinite size systems empowering materials design. |
| first_indexed | 2024-04-10T21:01:30Z |
| format | Article |
| id | doaj.art-e2950d2ad85045acb387a6af8451cde7 |
| institution | Directory Open Access Journal |
| issn | 2057-3960 |
| language | English |
| last_indexed | 2024-04-10T21:01:30Z |
| publishDate | 2023-01-01 |
| publisher | Nature Portfolio |
| record_format | Article |
| series | npj Computational Materials |
| spelling | doaj.art-e2950d2ad85045acb387a6af8451cde72023-01-22T12:19:47ZengNature Portfolionpj Computational Materials2057-39602023-01-01911910.1038/s41524-023-00967-zActive learning to overcome exponential-wall problem for effective structure prediction of chemical-disordered materialsXiaoze Yuan0Yuwei Zhou1Qing Peng2Yong Yang3Yongwang Li4Xiaodong Wen5State Key Laboratory of Coal Conversion, Institute of Coal Chemistry, Chinese Academy of SciencesState Key Laboratory of Coal Conversion, Institute of Coal Chemistry, Chinese Academy of SciencesThe State Key Laboratory of Nolinear Mechanics, Institute of Mechanics, Chinese Academy of SciencesState Key Laboratory of Coal Conversion, Institute of Coal Chemistry, Chinese Academy of SciencesState Key Laboratory of Coal Conversion, Institute of Coal Chemistry, Chinese Academy of SciencesState Key Laboratory of Coal Conversion, Institute of Coal Chemistry, Chinese Academy of SciencesAbstract Chemical-disordered materials have a wide range of applications whereas the determination of their structures or configurations is one of the most important and challenging problems. Traditional methods are extremely inefficient or intractable for large systems due to the notorious exponential-wall issue that the number of possible structures increase exponentially for N-body systems. Herein, we introduce an efficient approach to predict the thermodynamically stable structures of chemical-disordered materials via active-learning accompanied by first-principles calculations. Our method, named LAsou, can efficiently compress the sampling space and dramatically reduce the computational cost. Three distinct and typical finite-size systems are investigated, including the anion-disordered BaSc(OxF1−x)3 (x = 0.667), the cation-disordered Ca1−xMnxCO3 (x = 0.25) with larger size and the defect-disordered ε-FeCx (x = 0.5) with larger space. The commonly used enumeration method requires to explicitly calculate 2664, 1033, and 10496 configurations, respectively, while the LAsou method just needs to explicitly calculate about 15, 20, and 10 configurations, respectively. Besides the finite-size system, our LAsou method is ready for quasi-infinite size systems empowering materials design.https://doi.org/10.1038/s41524-023-00967-z |
| spellingShingle | Xiaoze Yuan Yuwei Zhou Qing Peng Yong Yang Yongwang Li Xiaodong Wen Active learning to overcome exponential-wall problem for effective structure prediction of chemical-disordered materials npj Computational Materials |
| title | Active learning to overcome exponential-wall problem for effective structure prediction of chemical-disordered materials |
| title_full | Active learning to overcome exponential-wall problem for effective structure prediction of chemical-disordered materials |
| title_fullStr | Active learning to overcome exponential-wall problem for effective structure prediction of chemical-disordered materials |
| title_full_unstemmed | Active learning to overcome exponential-wall problem for effective structure prediction of chemical-disordered materials |
| title_short | Active learning to overcome exponential-wall problem for effective structure prediction of chemical-disordered materials |
| title_sort | active learning to overcome exponential wall problem for effective structure prediction of chemical disordered materials |
| url | https://doi.org/10.1038/s41524-023-00967-z |
| work_keys_str_mv | AT xiaozeyuan activelearningtoovercomeexponentialwallproblemforeffectivestructurepredictionofchemicaldisorderedmaterials AT yuweizhou activelearningtoovercomeexponentialwallproblemforeffectivestructurepredictionofchemicaldisorderedmaterials AT qingpeng activelearningtoovercomeexponentialwallproblemforeffectivestructurepredictionofchemicaldisorderedmaterials AT yongyang activelearningtoovercomeexponentialwallproblemforeffectivestructurepredictionofchemicaldisorderedmaterials AT yongwangli activelearningtoovercomeexponentialwallproblemforeffectivestructurepredictionofchemicaldisorderedmaterials AT xiaodongwen activelearningtoovercomeexponentialwallproblemforeffectivestructurepredictionofchemicaldisorderedmaterials |