Estimating correlation energy of diatomic molecules and atoms with neural networks

Abstract

The electronic correlation energy of diatomic molecules and heavy atoms is estimated using a back propagation neural network approach. The supervised learning is accomplished using known exact results of the electronic correlation energy. The recall rate, that is, the performance of the net in recognizing the training set, is about 96%. The correctness of values given to the test set and prediction rate is at the 90% level. We generate tables for the electronic correlation energy of several diatomic molecules and all the neutral atoms up to radon (Rn). © 1997 by John Wiley & Sons, Inc. J Comput Chem 18: 1407–1414, 1997

Machine learning for quantum mechanics in a nutshell

“…Predating the introduction and formalization of the kernel learning framework, concepts like regularization were used early on, e.g., by Rabitz . Interpolation between QM results for different systems, e.g., molecular property estimates, started roughly a decade later with usage of ANN to predict correlation energies and bond dissociation enthalpies . Later, other methods such as support vector machines were used as well .…”

Neural network potential-energy surfaces in chemistry: a tool for large-scale simulations

“…25,41 They have been applied to the classification of NMR spectra, 42 the investigation of mass spectra of organic compounds, 43 the modeling of kinetic rates, 44 the prediction of protein secondary and ternary structures based on the sequence of amino acids, 45,46 quantitative structure-activity relationship (QSAR) studies, [47][48][49] the prediction of binding sites of biomolecules, 50 the classification of medical data in clinical chemistry, 51 the analysis of operation conditions in polymerization reactions, 52 the prediction of nuclear ground state spins, 53 the classification of atomic energy levels, 54,55 the analysis of nucleic acid sequences, 56 to solve the Schro¨dinger equation for simple model systems, [57][58][59][60][61] to predict the outcome of molecular trajectories, 62,63 to estimate the binding free energies of enzymatic inhibitors, 64 and to extract information on pair potentials from structure factors. 65 Further, a number of applications have been reported, which are related to the PES, like the estimation of DFT energies with converged basis sets using lower level electronic structure calculations, 66 the estimation of the correlation energy of diatomic molecules and heavy atoms, 67 and the derivation of improved enthalpies, 68,69 bond energies 70 and heats of formation. 71,72 All these applications are based on the ability of NNs to detect hidden patterns in complex data sets.…”

Neural networks study on the correlation between impact sensitivity and molecular structures for nitramine explosives

“…In addition, the use of neural network methods has grown constantly in a variety of applications in chemistry and physics since their first utilization for the prediction of protein secondary structure. For example, neural networks have been used to calculate the ground-state eigenenergy of two-dimensional harmonic oscillators [34], to solve nonhomogenous ordinary and partial differential equations [35], and to obtain the electronic correlation energy for atoms and diatomic molecules [36]. But for the explosive engineering, there exist few reports about the impact sensitivity based on neural networks method, although Nefati [3] and Cho et al [37] predicted the impact sensitivity of various types of explosive molecules via neural networks.…”