Machine learning product state distributions from initial reactant states for a reactive atom–diatom collision system

Abstract

A machine-learned model for predicting product state distributions from specific initial states (state-to-distribution or STD) for reactive atom–diatom collisions is presented and quantitatively tested for the N(4S) + O2(X3Σg−) → NO(X2Π) + O(3P) reaction. The reference dataset for training the neural network consists of final state distributions determined from quasi-classical trajectory (QCT) simulations for ∼2000 initial conditions. Overall, the prediction accuracy as quantified by the root-mean-squared difference (∼0.003) and the R2 (∼0.99) between the reference QCT and predictions of the STD model is high for the test set, for off-grid state-specific initial conditions, and for initial conditions drawn from reactant state distributions characterized by translational, rotational, and vibrational temperatures. Compared with a more coarse grained distribution-to-distribution (DTD) model evaluated on the same initial state distributions, the STD model shows comparable performance with the additional benefit of the state resolution in the reactant preparation. Starting from specific initial states also leads to a more diverse range of final state distributions, which requires a more expressive neural network compared with DTD. A direct comparison between QCT simulations, the STD model, and the widely used Larsen–Borgnakke (LB) model shows that the STD model is quantitative, whereas the LB model is qualitative at best for rotational distributions P(j′) and fails for vibrational distributions P(v′). As such, the STD model can be well-suited for simulating nonequilibrium high-speed flows, e.g., using the direct simulation Monte Carlo method.

Rovibrational-Specific QCT and Master Equation Study on N2(X1Σg+) + O(3P) and NO(X2Π) + N(4S) Systems in High-Energy Collisions

“…In addition to the Zel’dovich mechanism (), collisional energy-transfer and dissociation processes of individual collision pairs in the N 2 O system (i.e., N 2 + O and NO + N) are also relevant in developing a reliable thermochemical nonequilibrium model for high-temperature air. Regarding the N 2 O system, previous theoretical studies − have been carried out by means of ab initio potential energy surface (PES) constructions followed by quasiclassical trajectory (QCT) calculations for particular chemical reaction channels. Each of the PESs developed by Bose and Candler and Gamallo et al was employed to compute thermal rate coefficients for the Zel’dovich reaction ().…”

Atomistic Simulations for Reactions and Vibrational Spectroscopy in the Era of Machine Learning─Quo Vadis?

“…Comparison with rates directly determined from QCT simulations shows that the trained NN reaches accuracies better than 99% over a wide temperature range (1000 ≤ T ≤ 20000 K). A very recent generalization was concerned with learning entire product state distributions for specific reactant states (state-to-distribution model; STD) . The performance of this model is equally good as the STS model but at considerably reduced computational cost.…”

Quantitative molecular simulations

“…Computed thermal and vibrational relaxation rates agree to within a few percent with those measured experimentally which is a good basis for developing more coarse grained models. 191–193…”

“…Computed thermal and vibrational relaxation rates agree to within a few percent with those measured experimentally which is a good basis for developing more coarse grained models. [191][192][193] Malonaldehyde (MA), acetylacetone, and formic acid dimer (FAD) are topical systems for quantitative simulations of gasphase spectroscopy and reactions. In particular MA and FAD have attracted considerable interest and quantitatively accurate results are available for tunneling splittings.…”