Neural networks as a tool for compact representation of ab initio molecular potential energy surfaces |
| |
| Institution: | 1. Physics Department, Belarusian State University, 4 Nezaležnaści Ave., 220030 Minsk, Belarus;2. A.N. Sevchenko Institute of Applied Physical Problems at Belarusian State University, 7 Kurčataǔ Str., 220108 Minsk, Belarus;1. Key Laboratory of Materials Modification by Laser, Electron, and Ion Beams, Ministry of Education, School of Physics and Optoelectronic Technology, Dalian University of Technology, Dalian 116024, PR China;2. Department of College Foundation Education, BoHai University, Jinzhou 121000, Liaoning, PR China;3. State Key Laboratory of Molecular Reaction Dynamics, Dalian Institute of Chemical Physics, Chinese Academy of Science, Zhongshan Road 457, Dalian 116023, PR China;1. Institute of Modern Physics, Northwest University, Xi’an, Shaanxi 710127, China;2. School of Physics, Northwest University, Xi’an, Shaanxi 710127, China;3. Beijing Computational Science Research Center, Beijing 100193, China;4. Shaanxi Key Laboratory for Theoretical Physics Frontiers, Xi’an, Shaanxi 710127, China;1. Faculty of Physics, Belarusian State University, 4 Nezaležnaści Ave., 220030 Minsk, Belarus;2. A.N. Sevchenko Institute of Applied Physical Problems at Belarusian State University, 7 Kurčataǔ Str., 220108 Minsk, Belarus |
| |
| Abstract: | Ab initio quantum chemical calculations of molecular properties such as, e.g., torsional potential energies, require massive computational effort even for moderately sized molecules, if basis sets with a reasonable quality are employed. Using ab initio data on conformational properties of the cofactor (6R,1′R,2′S)-5,6,7,8-tetrahydrobiopterin, we demonstrate that error backpropagation networks can be established that efficiently approximate complicated functional relationships such as torsional potential energy surfaces of a flexible molecule. Our pilot simulations suggest that properly trained neural networks might provide an extremely compact storage medium for quantum chemically obtained information. Moreover, they are outstandingly comfortable tools when it comes to making use of the stored information. One possible application is demonstrated, namely, computation of relaxed torsional energy surfaces. |
| |
| Keywords: | |
| 本文献已被 ScienceDirect 等数据库收录！ |
|
