Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Method for calculating water molecule energy based on molecular degree-of-freedom depth learning

A technology of deep learning and water molecules, applied in the field of molecular energy calculation, can solve problems such as difficult to obtain satisfactory results, and achieve the effects of avoiding convergence problems, reducing influence, and accurately calculating results

Inactive Publication Date: 2019-03-01
DALIAN UNIV
View PDF3 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Restricted by this feature, it is difficult to obtain satisfactory results in the application of complex molecular systems or multi-molecular systems.

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Method for calculating water molecule energy based on molecular degree-of-freedom depth learning
  • Method for calculating water molecule energy based on molecular degree-of-freedom depth learning
  • Method for calculating water molecule energy based on molecular degree-of-freedom depth learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0037] A method to calculate the energy of water molecules through deep learning, using the spatial coordinates and corresponding energies of 1000 water molecules in different configurations. The change range of the two hydrogen-oxygen bonds of water molecules is The range of bond angles is 104.2±8.59°. Directly train the molecular energy with the molecular configuration, randomly select m configurations and corresponding energies as the training group, and the remaining 1000-m configurations and corresponding energies as the test group, calculate the energy of water molecules, and verify the correctness of the training results.

[0038] Transform the molecular space coordinates of the training group into two hydrogen-oxygen bond lengths (r O-H1 , r O-H2 ), the molecular bond angle (θ) and the reciprocal of the distance between the three atoms (1 / r O-H1 ,1 / r O-H2 ,1 / r H1-H2 ), each configuration parameter constitutes an independent six-column configuration parameter matr...

Embodiment 2

[0064] This embodiment provides a preferred solution of the hardware platform and software environment of the present invention.

[0065] Choose the low-end i5-6500 CPU@3.20GHz / NVIDIA Corporation GK208 [GeForce GT730] / 4G Mem hardware platform to obtain higher general performance; the software environment is Linux kernel 4.9 / TensorFlow-GPU 1.8.0 (via pip installed), the driver is CUDA 9.0 / cuDNN 7.1.

Embodiment 3

[0067] This embodiment provides a preferred scheme for selecting input data in the present invention.

[0068] Using the water data set provided by Brockherde et al., the data set contains the configuration of 1000 water molecules and the energy corresponding to one of them. The configuration is expressed by Bohr positions, and the energy unit is kcal / mol. This approach avoids the use of gradient descent to compute the minimized total energy while ensuring that the accuracy of molecular energies improves with larger training sets, and compares the results with those computed using the standard DFT approximation (PBE). The water molecule parameters are set to three: two bond lengths and one bond angle. According to the PBE results, the optimized water molecular configuration ( θ 0 =104.2°) is the starting point of training, at and ±8.59° to generate random combination configurations.

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses a method for calculating water molecule energy based on molecular degree-of-freedom depth learning, and belongs to the technical field of molecular energy calculation. The method comprises the following steps: S1, a water molecular database is established; S2, m configurations and corresponding energy are selected at random to serve as a training group, remaining 1000-m configurations and corresponding energy serve as a test group; S3, spatial coordinates of water molecules in the training group are transformed and serve as starting of calculation; S4, energy data of the training group are extracted to serve as training group output energy matrixes corresponding to column configuration parameter input matrixes in an one-to-one mode; S5, test group configuration parameter input matrixes and test group output energy matrixes are established; and S6, an energy matrix Ecalc is obtained through double neural layer calculation. According to the method, the influence of the ratio of the training group to the test group on accuracy of training results is reduced.

Description

technical field [0001] The invention relates to the technical field of molecular energy calculation, in particular to a method for calculating the energy of water molecules based on deep learning of molecular degrees of freedom. Background technique [0002] The machine learning of contemporary artificial intelligence, in the field of molecular structure optimization and minimum energy calculation, has a remarkable feature that as the degree of freedom of the initial matrix increases, that is, the more complex the molecular structure, the greater the degree of freedom, and the training group and the test group must be increased. ratio for accurate results. Restricted by this feature, it is difficult to obtain satisfactory results in the application of complex molecular systems or multi-molecular systems. Contents of the invention [0003] In order to solve the defects in the prior art, the present invention provides a method for calculating the energy of water molecules t...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
IPC IPC(8): G16C20/20
Inventor 崔洪光周立川商祎行周毅
Owner DALIAN UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com