Building foundation of polymer informatics: ensemble of generators and virtual libraries of diverse functional polymers

构建聚合物信息学基础：各种功能聚合物的生成器和虚拟库的集合

• 批准号: 21K14675
• 负责人: Wu Stephen
• 金额: $ 2.91万
• 依托单位: The Institute of Statistical Mathematics
• 依托单位国家: 日本
• 项目类别: Grant-in-Aid for Early-Career Scientists
• 财政年份: 2021
• 资助国家: 日本
• 起止时间: 2021-04-01 至 2024-03-31
• 项目状态: 已结题
• 来源: https://kaken.nii.ac.jp/en/grant/KAKENHI-PROJECT-21K14675/
• 关键词: polymer informatics; generative models; open source software; ensemble learning; virtual library

In 2022, multiple types of polymer generators have been prepared under the all-in-one materials informatics platform, XenonPy. (1) language-model-based generators, NGrams, for 21 classes of polymers have been successfully trained; (2) reaction-based polymer candidate generator has been developed; (3) polymer generator based on rule-based virtual synthesis is going to be implemented. On top of the on-going expansion of the polymer generator library, application of our polymer informatics has been test on the design of new liquid crystalline polymers. Five new liquid crystalline polymers have been discovered. One of them further demonstrated a relatively high thermal conductivity, showing potential to be tested for industrial use. Our results are currently under preparation for publications. Seven presentations have been given around the world for this year's research achievement, and more will be given in the following year. Also, the development of our XenonPy platform has motivated new collaboration opportunities with the industrial companies. Last but not least, our work has been served as a foundation for another polymer informatics project, called RadonPy, which aims at producing the largest open calculation database for polymers.

2022年，在一体化材料信息学平台XenonPy下制备了多种类型的聚合物发生器。(1)在此基础上，本文提出了一种基于语言模型的聚合物生成器NGrams，并成功地训练了21类聚合物的生成器NGrams; 2）开发了基于反应的聚合物候选生成器; 3）实现了基于规则虚拟合成的聚合物生成器。在聚合物生成器库的持续扩展之上，我们的聚合物信息学的应用已经在新的液晶聚合物的设计上进行了测试。发现了五种新的液晶聚合物。其中一种进一步显示出相对较高的导热性，显示出可用于工业用途的潜力。我们的研究结果目前正在准备出版。今年的研究成果已在世界各地进行了七次报告，明年还将进行更多的报告。此外，我们XenonPy平台的开发激发了与工业公司的新合作机会。最后但并非最不重要的是，我们的工作已经成为另一个聚合物信息学项目的基础，称为RadonPy，其目的是产生最大的聚合物开放计算数据库。

项目成果

Design of liquid-crystalline polyimides by integrating expert knowledge to a data-driven machine learning framework

将专业知识集成到数据驱动的机器学习框架中设计液晶聚酰亚胺

• 发表时间: 2022
• 作者: Ju Shenghong;Yoshida Ryo;Liu Chang;Wu Stephen;Hongo Kenta;Tadano Terumasa;Shiomi Junichiro;Stephen Wu;Stephen Wu;Stephen Wu;Stephen Wu;Stephen Wu;Stephen Wu;Stephen Wu;Stephen Wu;Stephen Wu
• 通讯作者: Stephen Wu

Exploring diamondlike lattice thermal conductivity crystals via feature-based transfer learning

• DOI: 10.1103/physrevmaterials.5.053801
• 发表时间: 2019-09
• 期刊: Physical Review Materials
• 影响因子: 3.4
• 作者: S. Ju;Ryo Yoshida;Chang Liu;Stephen Wu;K. Hongo;T. Tadano;J. Shiomi

データ駆動型の液晶ポリイミド設計

数据驱动的液晶聚酰亚胺设计

• 发表时间: 2022
• 作者: Ju Shenghong;Yoshida Ryo;Liu Chang;Wu Stephen;Hongo Kenta;Tadano Terumasa;Shiomi Junichiro;Stephen Wu;Stephen Wu;Stephen Wu;Stephen Wu;Stephen Wu;Stephen Wu
• 通讯作者: Stephen Wu

Design of high thermal conductivity liquid-crystalline polyimides based on interplay of expert knowledge and machine learning

基于专家知识和机器学习相互作用的高导热液晶聚酰亚胺设计

• 发表时间: 2022
• 作者: Ju Shenghong;Yoshida Ryo;Liu Chang;Wu Stephen;Hongo Kenta;Tadano Terumasa;Shiomi Junichiro;Stephen Wu;Stephen Wu;Stephen Wu;Stephen Wu;Stephen Wu;Stephen Wu;Stephen Wu
• 通讯作者: Stephen Wu

Bayesian Sequential Stacking Algorithm for Concurrently Designing Molecules and Synthetic Reaction Networks

• DOI: 10.48550/arxiv.2204.01847
• 发表时间: 2022-03
• 期刊: ArXiv
• 作者: Qi Zhang-;Chang Liu;Stephen Wu;Ryo Yoshida