针对目标性能的多元合金成分设计因具有巨大的成分参数空间而极具挑战,而且传统的试错实验由于效率低能探索的合金成分有限。提出利用高通量实验结合机器学习方法加速非等摩尔比的硬质高熵合金Co x Cr y Ti z Mo u W v的成分设计。首先...针对目标性能的多元合金成分设计因具有巨大的成分参数空间而极具挑战,而且传统的试错实验由于效率低能探索的合金成分有限。提出利用高通量实验结合机器学习方法加速非等摩尔比的硬质高熵合金Co x Cr y Ti z Mo u W v的成分设计。首先通过自主研发的全流程高通量合金制备系统制备了138个不同成分的高熵合金铸态样品。然后根据测量的维氏硬度(HV)数据,使用随机森林法和支持向量机法进行机器学习建模,并预测了五元合金体系内潜在的3876个不同成分合金的硬度。随机森林机器学习模型的预测结果在高(HV>800 MPa)、中(600<HV<800 MPa)、低(HV<600 MPa)硬度区域的平均误差分别为2.87%,3.30%和6.70%,实验硬度值在对应区域的测量误差分别为1.69%,1.88%和1.87%。根据机器学习模型预测结果建立的“成分-硬度”与“描述因子-硬度”关系图谱展示了全成分空间内高熵合金的硬度变化规律及影响硬度的重要描述因子——原子半径差。研究结果表明,高通量实验与机器学习相结合可使多元合金成分优化效率提高百倍以上。此外,建议未来研究应在“机器学习”基础上加强“向机器学习”,在更高层次上获得新的专业知识认知。展开更多
The ring-polymer molecular dynamics(RPMD)was used to calculate the thermal rate coefficients of the multi-channel roaming reaction H+MgH→Mg+H_(2).Two reaction channels,tight and roaming,are explicitly considered.This...The ring-polymer molecular dynamics(RPMD)was used to calculate the thermal rate coefficients of the multi-channel roaming reaction H+MgH→Mg+H_(2).Two reaction channels,tight and roaming,are explicitly considered.This is a pioneering attempt of exerting RPMD method to multichannel reactions.With the help of a newly developed optimization-interpolation protocol for preparing the initial structures and adaptive protocol for choosing the force constants,we have successfully obtained the thermal rate coefficients.The results are consistent with those from other theoretical methods,such as variational transition state theory and quantum dynamics.Especially,RPMD results exhibit negative temperature dependence,which is similar to the results from variational transition state theory but different from the ones from ground state quantum dynamics calculations.展开更多
In this work,the solidification of liquid iron with or without external magnetic field was investigated by using two molecular dynamics methods,namely direct cooling and two-phase simulation.The influence of external ...In this work,the solidification of liquid iron with or without external magnetic field was investigated by using two molecular dynamics methods,namely direct cooling and two-phase simulation.The influence of external magnetic field on the solidification is characterized by the critical temperature and radial distribution functions.Our computational results show that under external magnetic field,the solidification point tends to decrease significantly.By further analyzing the diffusion coefficients and viscosity,we attribute the effect to the stronger fluctuation of liquid iron atoms driven by the external magnetic field.展开更多
基金Supported by the National Natural Science Foundation of China(61674161,61504083)Open Project of State Key Laboratory of Functional Materials for Informatics,Public welfare capacity building in Guangdong Province(2015A010103016)the Science and Technology Foundation of Shenzhen(JCYJ20160226192033020)
文摘针对目标性能的多元合金成分设计因具有巨大的成分参数空间而极具挑战,而且传统的试错实验由于效率低能探索的合金成分有限。提出利用高通量实验结合机器学习方法加速非等摩尔比的硬质高熵合金Co x Cr y Ti z Mo u W v的成分设计。首先通过自主研发的全流程高通量合金制备系统制备了138个不同成分的高熵合金铸态样品。然后根据测量的维氏硬度(HV)数据,使用随机森林法和支持向量机法进行机器学习建模,并预测了五元合金体系内潜在的3876个不同成分合金的硬度。随机森林机器学习模型的预测结果在高(HV>800 MPa)、中(600<HV<800 MPa)、低(HV<600 MPa)硬度区域的平均误差分别为2.87%,3.30%和6.70%,实验硬度值在对应区域的测量误差分别为1.69%,1.88%和1.87%。根据机器学习模型预测结果建立的“成分-硬度”与“描述因子-硬度”关系图谱展示了全成分空间内高熵合金的硬度变化规律及影响硬度的重要描述因子——原子半径差。研究结果表明,高通量实验与机器学习相结合可使多元合金成分优化效率提高百倍以上。此外,建议未来研究应在“机器学习”基础上加强“向机器学习”,在更高层次上获得新的专业知识认知。
基金supported by the National Natural Science Foundation of China(No.21503130 and No.11674212,and No.21603144)supported by the Young Eastern Scholar Program of the Shanghai Municipal Education Commission(QD2016021)+1 种基金the Shanghai Key Laboratory of High Temperature Superconductors(No.14DZ2260700)supported by Shanghai Sailing Program(No.2016YF1408400).
文摘The ring-polymer molecular dynamics(RPMD)was used to calculate the thermal rate coefficients of the multi-channel roaming reaction H+MgH→Mg+H_(2).Two reaction channels,tight and roaming,are explicitly considered.This is a pioneering attempt of exerting RPMD method to multichannel reactions.With the help of a newly developed optimization-interpolation protocol for preparing the initial structures and adaptive protocol for choosing the force constants,we have successfully obtained the thermal rate coefficients.The results are consistent with those from other theoretical methods,such as variational transition state theory and quantum dynamics.Especially,RPMD results exhibit negative temperature dependence,which is similar to the results from variational transition state theory but different from the ones from ground state quantum dynamics calculations.
基金funded by the National Natural Science Foundation of China(No.22173057 for Yongle Li and No.51690164 for Xi Li)the Foundation of Shanghai Science and Technology Commission(No.21JC1402700 and No.21DZ2304900 for Yongle Li)supported by Open Project of State Key Laboratory of Advanced Special Steel,Shanghai Key Laboratory of Advanced Ferrometallurgy,Shanghai University。
文摘In this work,the solidification of liquid iron with or without external magnetic field was investigated by using two molecular dynamics methods,namely direct cooling and two-phase simulation.The influence of external magnetic field on the solidification is characterized by the critical temperature and radial distribution functions.Our computational results show that under external magnetic field,the solidification point tends to decrease significantly.By further analyzing the diffusion coefficients and viscosity,we attribute the effect to the stronger fluctuation of liquid iron atoms driven by the external magnetic field.
