scholarly journals Analytical gradients for molecular-orbital-based machine learning

2021 ◽  
Vol 154 (12) ◽  
pp. 124120
Author(s):  
Sebastian J. R. Lee ◽  
Tamara Husch ◽  
Feizhi Ding ◽  
Thomas F. Miller
Keyword(s):  
Machine Learning ◽  
Molecular Orbital ◽  
Analytical Gradients

scholarly journals High-Precision Atomic Charge Prediction for Protein Systems Using Fragment Molecular Orbital Calculation and Machine Learning

2020 ◽  
Vol 60 (7) ◽  
pp. 3361-3368
Author(s):  
Koichiro Kato ◽  
Tomohide Masuda ◽  
Chiduru Watanabe ◽  
Naoki Miyagawa ◽  
Hideo Mizouchi ◽  
...  
Keyword(s):  
Machine Learning ◽  
Molecular Orbital ◽  
High Precision ◽  
Atomic Charge ◽  
Molecular Orbital Calculation ◽  
Fragment Molecular Orbital

scholarly journals Machine Learning Predictor Models in the Electronic Properties of Alkanes based on Degree-Topology Indices

2021 ◽  
Vol 11 (11) ◽  
pp. 1-14
Author(s):  
Zubainun Mohamed Zabidi ◽  
◽  
Ahmad Nazib Alias ◽  
Nurul Aimi Zakaria ◽  
Zaidatul Salwa Mahmud ◽  
...  
Keyword(s):  
Molecular Structure ◽  
Machine Learning ◽  
Graph Theory ◽  
Linear Regression ◽  
Electronic Properties ◽  
Molecular Orbital ◽  
Machine Learning Algorithms ◽  
Chemical Graph ◽  
Chemical Graph Theory ◽  
Homo And Lumo

New topology indices that are degree-based have been introduced to represent molecular structure from chemical graph theory. The indices give a new sight into the physical properties of the chemical compounds. The correlation of physiochemical properties with chemical graph theory can be done using the Quantitative Structure Properties Relationship (QSPR). Highest Occupied Molecular Orbital (HOMO) and Lowest Unoccupied Molecular Orbital (LUMO) are two basic electronic properties that describe the physiochemical of molecular structure. In computational chemistry, HOMO and LUMO can be calculated by ab initio molecular orbital calculation such as semi-empirical and density functional theory (DFT) method. However, these methods are time-consuming computations. In this paper, predictor model of HOMO and LUMO were developed using Machine Learning algorithms namely Linear Regression, Ridge Regression, LASSO Regression and Elastic Net Regression. The results showed that the performance achievement of each of the machine learning algorithms varied in accordance to the topology indices descriptors and the most outperformed model was presented by Linear Regression with the Moment Balaban Indices (JJ). This paper provides the fundamental design and implementation framework of predicting the HOMO and LUMO electronic properties

scholarly journals Deep Learning Total Energies and Orbital Energies of Large Organic Molecules Using Hybridization of Molecular Fingerprints

2020 ◽  
Author(s):  
Obaidur Rahaman ◽  
Alessio Gagliardi
Keyword(s):  
Machine Learning ◽  
Deep Learning ◽  
Molecular Orbital ◽  
Large Scale ◽  
Organic Molecules ◽  
Chemical Properties ◽  
Many Body ◽  
Molecular Fingerprints ◽  
Total Energies ◽  
The Individual

<p>The ability to predict material properties without the need of resource consuming experimental efforts can immensely accelerate material and drug discovery. Although ab initio methods can be reliable and accurate in making such predictions, they are computationally too expensive at a large scale. The recent advancements in artificial intelligence and machine learning as well as availability of large quantum mechanics derived datasets enable us to train models on these datasets as benchmark and to make fast predictions on much larger datasets. The success of these machine learning models highly depends on the machine-readable fingerprints of the molecules that capture their chemical properties as well as topological information. In this work we propose a common deep learning based framework to combine different types of molecular fingerprints to enhance prediction accuracy. Graph Neural Network (GNN), Many Body Tensor Representation (MBTR) and a set of simple Molecular Descriptors (MD) were used to predict the total energies, Highest Occupied Molecular Orbital (HOMO) energies and Lowest Unoccupied Molecular Orbital (LUMO) energies of a dataset containing ~62k large organic molecules with complex aromatic rings and remarkably diverse functional groups. The results demonstrate that a combination of best performing molecular fingerprints can produce better results than the individual ones. The simple and flexible deep learning framework developed in this work can be easily adapted to incorporate other types of molecular fingerprints.<br></p>

scholarly journals Deep Learning Total Energies and Orbital Energies of Large Organic Molecules Using Hybridization of Molecular Fingerprints

2020 ◽  
Author(s):  
Obaidur Rahaman ◽  
Alessio Gagliardi
Keyword(s):  
Machine Learning ◽  
Deep Learning ◽  
Molecular Orbital ◽  
Large Scale ◽  
Organic Molecules ◽  
Chemical Properties ◽  
Many Body ◽  
Molecular Fingerprints ◽  
Total Energies ◽  
The Individual

<p>The ability to predict material properties without the need of resource consuming experimental efforts can immensely accelerate material and drug discovery. Although ab initio methods can be reliable and accurate in making such predictions, they are computationally too expensive at a large scale. The recent advancements in artificial intelligence and machine learning as well as availability of large quantum mechanics derived datasets enable us to train models on these datasets as benchmark and to make fast predictions on much larger datasets. The success of these machine learning models highly depends on the machine-readable fingerprints of the molecules that capture their chemical properties as well as topological information. In this work we propose a common deep learning based framework to combine different types of molecular fingerprints to enhance prediction accuracy. Graph Neural Network (GNN), Many Body Tensor Representation (MBTR) and a set of simple Molecular Descriptors (MD) were used to predict the total energies, Highest Occupied Molecular Orbital (HOMO) energies and Lowest Unoccupied Molecular Orbital (LUMO) energies of a dataset containing ~62k large organic molecules with complex aromatic rings and remarkably diverse functional groups. The results demonstrate that a combination of best performing molecular fingerprints can produce better results than the individual ones. The simple and flexible deep learning framework developed in this work can be easily adapted to incorporate other types of molecular fingerprints.<br></p>

scholarly journals Regression Clustering for Improved Accuracy and Training Costs with Molecular-Orbital-Based Machine Learning

2019 ◽  
Vol 15 (12) ◽  
pp. 6668-6677 ◽  
Author(s):  
Lixue Cheng ◽  
Nikola B. Kovachki ◽  
Matthew Welborn ◽  
Thomas F. Miller
Keyword(s):  
Machine Learning ◽  
Molecular Orbital ◽  
Training Costs ◽  
Improved Accuracy ◽  
And Training

scholarly journals Prediction of Molecular Properties by Molecular Orbital Calculations and Machine Learning

2020 ◽  
Vol 19 (2) ◽  
pp. 43-45
Author(s):  
Hiroyuki TERAMAE ◽  
Tetsuhide MATSUO ◽  
Kazuma NIWATSUKINO ◽  
Ryota INOUE ◽  
Shinji NOGUCHI ◽  
...  
Keyword(s):  
Machine Learning ◽  
Molecular Orbital ◽  
Molecular Properties ◽  
Molecular Orbital Calculations

scholarly journals A universal density matrix functional from molecular orbital-based machine learning: Transferability across organic molecules

2019 ◽  
Vol 150 (13) ◽  
pp. 131103 ◽  
Author(s):  
Lixue Cheng ◽  
Matthew Welborn ◽  
Anders S. Christensen ◽  
Thomas F. Miller
Keyword(s):  
Machine Learning ◽  
Density Matrix ◽  
Molecular Orbital ◽  
Organic Molecules

scholarly journals Machine learning property prediction for organic photovoltaic devices

2020 ◽  
Vol 6 (1) ◽  
Author(s):  
Nastaran Meftahi ◽  
Mykhailo Klymenko ◽  
Andrew J. Christofferson ◽  
Udo Bach ◽  
David A. Winkler ◽  
...  
Keyword(s):  
Machine Learning ◽  
Molecular Orbital ◽  
Short Circuit ◽  
Open Circuit ◽  
Green Energy ◽  
Organic Photovoltaic ◽  
Learning Approaches ◽  
Potential Donor ◽  
Energy Applications ◽  
Donor And Acceptor

Abstract Organic photovoltaic (OPV) materials are promising candidates for cheap, printable solar cells. However, there are a very large number of potential donors and acceptors, making selection of the best materials difficult. Here, we show that machine-learning approaches can leverage computationally expensive DFT calculations to estimate important OPV materials properties quickly and accurately. We generate quantitative relationships between simple and interpretable chemical signature and one-hot descriptors and OPV power conversion efficiency (PCE), open circuit potential (Voc), short circuit density (Jsc), highest occupied molecular orbital (HOMO) energy, lowest unoccupied molecular orbital (LUMO) energy, and the HOMO–LUMO gap. The most robust and predictive models could predict PCE (computed by DFT) with a standard error of ±0.5 for percentage PCE for both the training and test set. This model is useful for pre-screening potential donor and acceptor materials for OPV applications, accelerating design of these devices for green energy applications.

Sign in / Sign up

Export Citation Format

Share Document