Benchmark study on deep neural network potentials for small organic molecules

Abstract

There has been tremendous advancement in machine learning (ML) applications in computational chemistry, particularly in neural network potentials (NNP). NNPs can approximate potential energy surface (PES) as a high dimensional function by learning from existing reference data, thereby circumventing the need to solve the electronic Schrödinger equation explicitly. As a result, ML accelerates chemical space exploration and property prediction compared to quantum mechanical methods. Novel ML methods have the potential to provide efficient means for predicting the properties of molecules. However, this potential has been limited by the lack of standard comparative evaluations. In this work, we compare four selected models, that is, ANI, PhysNet, SchNet, and BAND‐NN, developed to represent the PES of small organic molecules. We evaluate these models for their accuracy and transferability on two different test sets (i) Small organic molecules of up to eight‐heavy atoms on which ANI and SchNet achieve root mean square error (RMSE) of 0.55 and 0.60 kcal/mol, respectively. (ii) On random selection of molecules from the GDB‐11 database with 10‐heavy atoms, ANI achieves RMSE of 1.17 kcal/mol and SchNet achieves RMSE of 1.89 kcal/mol. We examine their ability to produce smooth meaningful surface by performing PES scans for bond stretch, angle bend, and dihedral rotations on relatively large molecules to assess their possible application in molecular dynamics simulations. We also evaluate their performance for yielding minimum energy structures via geometry optimization using various minimization algorithms. All these models were also able to accurately differentiate different isomers of the same empirical formula C10H20. ANI and PhysNet achieve an RMSE of 0.29 and 0.52 kcal/mol, respectively, on C10H20 isomers.

Residual Life Prediction of Rotating Machinery Guided by Quantum Deep Neural Network

“…Implementation process of QGCCBNN based prediction method. The implementation process of the residual service life prediction method for rotating machinery based on QGCCBNN is shown in Figure 2.1 [8,9].…”

MolOpt: Autonomous Molecular Geometry Optimization Using Multiagent Reinforcement Learning

“…Neural network potentials (NNPs) learn to approximate the potential energy surface (PES) as a high dimensional function (HDF) f by learning from existing reference data. Once trained NNPs can successfully circumvent the need to solve the electronic Schrödinger equation explicitly as it has learned the mapping f ( Z i , r i ) → E , where Z i are the nuclear charges and r i are the atomic positions. − Machine learning (ML) methods in general have been successful in improving computational chemistry algorithms leading to accelerated property prediction and chemical space exploration . Recently, much emphasis has been on developing efficient ML-based search algorithms to explore chemical space, − but the same is not the case for conformational space.…”

MolOpt: Autonomous Molecular Geometry Optimization using Multi-Agent Reinforcement Learning

“…Machine learning (ML) methods have been successful in predicting the physicochemical properties of molecules. [1,2] Neural network potentials (NNPs) can successfully circumvent the need to solve the electronic Schrödinger equation explicitly by learning to approximate potential energy surface (PES) as a high dimensional function (HDF) by learning from existing reference data. [2,3,4,5,6,7,8] As a consequence, ML accelerates property prediction and chemical space exploration.…”

“…[1,2] Neural network potentials (NNPs) can successfully circumvent the need to solve the electronic Schrödinger equation explicitly by learning to approximate potential energy surface (PES) as a high dimensional function (HDF) by learning from existing reference data. [2,3,4,5,6,7,8] As a consequence, ML accelerates property prediction and chemical space exploration. Recently, much emphasis has been on developing efficient ML-based search algorithms to explore chemical space, but the same is not the case for conformational space.…”