Chemically Transferable Electronic Coarse Graining for Polythiophenes DOI
Zheng Yu, Nicholas E. Jackson

Journal of Chemical Theory and Computation, Год журнала: 2024, Номер 20(20), С. 9116 - 9127

Опубликована: Окт. 7, 2024

Recent advances in machine-learning-based electronic coarse graining (ECG) methods have demonstrated the potential to enable predictions soft materials at mesoscopic length scales. However, previous ECG models yet confront issue of chemical transferability. In this study, we develop chemically transferable for polythiophenes using graph neural networks. Our are trained on a data set that samples over conformational space random polythiophene sequences generated with 15 different monomer chemistries and three degrees polymerization. We systematically explore impact coarse-grained representation accuracy, highlighting significance preserving C-β coordinates thiophene. also find integrating unique polymer into training enhances model performance more efficiently than augmenting sampling already set. Moreover, our models, developed initially one property level quantum theory, can be transferred related properties higher levels theory minimal additional data. The introduced work will serve as foundation new classes across space.

Язык: Английский

Ten Problems in Polymer Reactivity Prediction DOI
Nicholas E. Jackson, Brett M. Savoie

Macromolecules, Год журнала: 2025, Номер unknown

Опубликована: Фев. 17, 2025

Язык: Английский

Процитировано

0
Synthesis, mesomorphic behaviour and luminescence of novel dithienopyrrole-based nematic liquid crystals DOI
Guang Hu, Wei Qian,

Xinyang Wu

и другие.

Liquid Crystals, Год журнала: 2025, Номер unknown, С. 1 - 12

Опубликована: Фев. 27, 2025

Язык: Английский

Процитировано

0
Chemically transferable electronic coarse graining for polythiophenes DOI Creative Commons
Zheng Yu, Nicholas E. Jackson

Опубликована: Июнь 21, 2024

Recent advances in machine-learning-based electronic coarse graining (ECG) methods have demonstrated the potential to enable predictions soft materials at mesoscopic length scales. However, previous ECG models yet confront issue of chemical transferability. In this study, we develop chemically transferable for polythiophenes using graph neural networks. Our are trained on a dataset that samples over conformational space random polythiophene sequences generated with 15 different monomer chemistries and three degrees polymerization. We systematically explore impact coarse-grained (CG) representation multiple resolutions accuracy, highlighting significance preserving C-beta coordinates thiophene. also find integrating unique polymer into training enhances model performance more efficiently than augmenting sampling already dataset. Moreover, our models, developed initially one property level quantum theory, can be transferred related properties higher levels theory minimal additional data. The introduced work will serve as foundation new classes across broader space.

Язык: Английский

Процитировано

1
Chemically Transferable Electronic Coarse Graining for Polythiophenes DOI
Zheng Yu, Nicholas E. Jackson

Journal of Chemical Theory and Computation, Год журнала: 2024, Номер 20(20), С. 9116 - 9127

Опубликована: Окт. 7, 2024

Recent advances in machine-learning-based electronic coarse graining (ECG) methods have demonstrated the potential to enable predictions soft materials at mesoscopic length scales. However, previous ECG models yet confront issue of chemical transferability. In this study, we develop chemically transferable for polythiophenes using graph neural networks. Our are trained on a data set that samples over conformational space random polythiophene sequences generated with 15 different monomer chemistries and three degrees polymerization. We systematically explore impact coarse-grained representation accuracy, highlighting significance preserving C-β coordinates thiophene. also find integrating unique polymer into training enhances model performance more efficiently than augmenting sampling already set. Moreover, our models, developed initially one property level quantum theory, can be transferred related properties higher levels theory minimal additional data. The introduced work will serve as foundation new classes across space.

Язык: Английский

Процитировано

0