More Like this

1. Efficient exploration of reaction pathways using reaction databases and active learning

   https://doi.org/10.1063/5.0235715  

   Kuryla, Domantas; Csányi, Gábor; van_Duin, Adri_C T; Michaelides, Angelos (March 2025, The Journal of Chemical Physics)

   Lian, T (Ed.)

   The fast and accurate simulation of chemical reactions is a major goal of computational chemistry. Recently, the pursuit of this goal has been aided by machine learning interatomic potentials (MLIPs), which provide energies and forces at quantum mechanical accuracy but at a fraction of the cost of the reference quantum mechanical calculations. Assembling the training set of relevant configurations is key to building the MLIP. Here, we demonstrate two approaches to training reactive MLIPs based on reaction pathway information. One approach exploits reaction datasets containing reactant, product, and transition state structures. Using an SN2 reaction dataset, we accurately locate reaction pathways and transition state geometries of up to 170 unseen reactions. In another approach, which does not depend on data availability, we present an efficient active learning procedure that yields an accurate MLIP and converged minimum energy path given only the reaction end point structures, avoiding quantum mechanics driven reaction pathway search at any stage of training set construction. We demonstrate this procedure on an SN2 reaction in the gas phase and with a small number of solvating water molecules, predicting reaction barriers within 20 meV of the reference quantum chemistry method. We then apply the active learning procedure on a more complex reaction involving a nucleophilic aromatic substitution and proton transfer, comparing the results against the reactive ReaxFF force field. Our active learning procedure, in addition to rapidly finding reaction paths for individual reactions, provides an approach to building large reaction path databases for training transferable reactive machine learning potentials. 

   more » « less
2. Closing the Sim-to-Real Gap in Guided Wave Damage Detection with Adversarial Training of Variational Auto-Encoders

   https://doi.org/10.1109/ICASSP43922.2022.9746196  

   Khurjekar, Ishan D.; Harley, Joel B. (May 2022, IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP))

   Guided wave testing is a popular approach for monitoring the structural integrity of infrastructures. We focus on the primary task of damage detection, where signal processing techniques are commonly employed. The detection performance is affected by a mismatch between the wave propagation model and experimental wave data. External variations, such as temperature, which are difficult to model, also affect the performance. While deep learning models can be an alternative detection method, there is often a lack of real-world training datasets. In this work, we counter this challenge by training an ensemble of variational autoencoders only on simulation data with a wave physics-guided adversarial component. We set up an experiment with non-uniform temperature variations to test the robustness of the methods. We compare our scheme with existing deep learning detection schemes and observe superior performance on experimental data. 

   more » « less
3. A Density Functional Tight Binding Layer for Deep Learning of Chemical Hamiltonians

   https://doi.org/10.1021/acs.jctc.8b00873  

   Li, Haichen; Collins, Christopher; Tanha, Matteus; Gordon, Geoffrey J; Yaron, David J (November 2018, Journal of Chemical Theory and Computation)

   Current neural networks for predictions of molecular properties use quantum chemistry only as a source of training data. This paper explores models that use quantum chemistry as an integral part of the prediction process. This is done by implementing self-consistent-charge Density-Functional-Tight-Binding (DFTB) theory as a layer for use in deep learning models. The DFTB layer takes, as input, Hamiltonian matrix elements generated from earlier layers and produces, as output, electronic properties from self-consistent field solutions of the corresponding DFTB Hamiltonian. Backpropagation enables efficient training of the model to target electronic properties. Two types of input to the DFTB layer are explored, splines and feed-forward neural networks. Because overfitting can cause models trained on smaller molecules to perform poorly on larger molecules, regularizations are applied that penalize nonmonotonic behavior and deviation of the Hamiltonian matrix elements from those of the published DFTB model used to initialize the model. The approach is evaluated on 15,700 hydrocarbons by comparing the root-mean-square error in energy and dipole moment, on test molecules with eight heavy atoms, to the error from the initial DFTB model. When trained on molecules with up to seven heavy atoms, the spline model reduces the test error in energy by 60% and in dipole moments by 42%. The neural network model performs somewhat better, with error reductions of 67% and 59%, respectively. Training on molecules with up to four heavy atoms reduces performance, with both the spline and neural net models reducing the test error in energy by about 53% and in dipole by about 25%. 

   more » « less
4. From ab initio to continuum: Linking multiple scales using deep-learned forces

   https://doi.org/10.1063/5.0166927  

   Wu, Haiyi; Liang, Chenxing; Jeong, Jinu; Aluru, N. R. (November 2023, The Journal of Chemical Physics)

   We develop a deep learning-based algorithm, called DeepForce, to link ab initio physics with the continuum theory to predict concentration profiles of confined water. We show that the deep-learned forces can be used to predict the structural properties of water confined in a nanochannel with quantum scale accuracy by solving the continuum theory given by Nernst–Planck equation. The DeepForce model has an excellent predictive performance with a relative error less than 7.6% not only for confined water in small channel systems (L < 6 nm) but also for confined water in large channel systems (L = 20 nm) which are computationally inaccessible through the high accuracy ab initio molecular dynamics simulations. Finally, we note that classical Molecular dynamics simulations can be inaccurate in capturing the interfacial physics of water in confinement (L < 4.0 nm) when quantum scale physics are neglected. 

   more » « less
5. Combining Empirical and Physics-Based Models for Solar Wind Prediction

   https://doi.org/10.3390/universe10050191  

   Johnson, Rob; Filali_Boubrahimi, Soukaina; Bahri, Omar; Hamdi, Shah Muhammad (May 2024, Universe)

   Solar wind modeling is classified into two main types: empirical models and physics-based models, each designed to forecast solar wind properties in various regions of the heliosphere. Empirical models, which are cost-effective, have demonstrated significant accuracy in predicting solar wind at the L1 Lagrange point. On the other hand, physics-based models rely on magnetohydrodynamics (MHD) principles and demand more computational resources. In this research paper, we build upon our recent novel approach that merges empirical and physics-based models. Our recent proposal involves the creation of a new physics-informed neural network that leverages time series data from solar wind predictors to enhance solar wind prediction. This innovative method aims to combine the strengths of both modeling approaches to achieve more accurate and efficient solar wind predictions. In this work, we show the variability of the proposed physics-informed loss across multiple deep learning models. We also study the effect of training the models on different solar cycles on the model’s performance. This work represents the first effort to predict solar wind by integrating deep learning approaches with physics constraints and analyzing the results across three solar cycles. Our findings demonstrate the superiority of our physics-constrained model over other unconstrained deep learning predictive models. 

   more » « less