Journal of Chemical Theory and Computation最新文献

The definition of heat current operator for systems for nonpairwise additive interactions and its impact on related lattice thermal conductivity (κ_L) via molecular dynamics (MD) simulation are ambiguous and controversial when migrating from empirical potential models to machine learning potential (MLP) models. Herein, we study and compare the significance of many-body interaction with heat current computation in one of the most popular MLP models, the moment tensor potential (MTP). Nonequilibrium MD simulations and equilibrium MD simulations among four different materials were performed, and inconsistencies in energy conservation between the simulation thermostat and the pairwise calculator were found. A new virial stress tensor expression with a many-body heat current description was integrated inside the MTP, and we uncovered the influence of the modification that could alter the κ_L results by 29-64% using the equilibrium MD computational approach. Our work demonstrates the importance of a many-body description during thermal analysis in MD simulations when MLPs are of concern.

A major challenge for many rare-event sampling strategies is the identification of progress coordinates that capture the slowest relevant motions. Machine-learning methods that can identify progress coordinates in an unsupervised manner have therefore been of great interest to the simulation community. Here, we developed a general method for identifying progress coordinates "on-the-fly" during weighted ensemble (WE) rare-event sampling via deep learning (DL) of outliers among sampled conformations. Our method identifies outliers in a latent space model of the system's sampled conformations that is periodically trained using a convolutional variational autoencoder. As a proof of principle, we applied our DL-enhanced WE method to simulate the NTL9 protein folding process. To enable rapid tests, our simulations propagated discrete-state synthetic molecular dynamics trajectories using a generative, fine-grained Markov state model. Results revealed that our on-the-fly DL of outliers enhanced the efficiency of WE by >3-fold in estimating the folding rate constant. Our efforts are a significant step forward in the unsupervised learning of slow coordinates during rare event sampling.

The identification of kinetically feasible reaction pathways that connect a reactant to its product, including numerous intermediates and transition states, is crucial for predicting chemical reactions and elucidating reaction mechanisms. However, as molecular systems become increasingly complex or larger, the number of local minimum structures and transition states grows, which makes this task challenging, even with advanced computational approaches. We introduced a reinforcement learning algorithm to efficiently identify a kinetically feasible reaction pathway between a given local minimum structure for the reactant and a given one for the product, starting from the reactant. The performance of the algorithm was validated using a benchmark data set of large-scale chemical reaction path networks. Several search policies were proposed, using metrics based on energetic or structural similarity to the product's goal structure, for each local minimum structure candidate found during the search. The performances of baseline greedy, random, and uniform search policies varied substantially depending on the system. In contrast, exploration-exploitation balanced policies such as Thompson sampling, probability of improvement, and expected improvement consistently demonstrated stable and high performance. Furthermore, we characterized the search mechanisms that depend on different policies in detail. This study also addressed potential avenues for further research, such as hierarchical reinforcement learning and multiobjective optimization, which could deepen the problem setting explored in this study.

We present a deep learning approach for analyzing two-dimensional scattering data of semiflexible polymers under external forces. In our framework, scattering functions are compressed into a three-dimensional latent space using a Variational Autoencoder (VAE), and two converter networks establish a bidirectional mapping between the polymer parameters (bending modulus, stretching force, and steady shear) and the scattering functions. The training data are generated using off-lattice Monte Carlo simulations to avoid the orientational bias inherent in lattice models, ensuring robust sampling of polymer conformations. The feasibility of this bidirectional mapping is demonstrated by the organized distribution of polymer parameters in the latent space. By integrating the converter networks with the VAE, we obtain a generator that produces scattering functions from given polymer parameters and an inferrer that directly extracts polymer parameters from scattering data. While the generator can be utilized in a traditional least-squares fitting procedure, the inferrer produces comparable results in a single pass and operates 3 orders of magnitude faster. This approach offers a scalable automated tool for polymer scattering analysis and provides a promising foundation for extending the method to other scattering models, experimental validation, and the study of time-dependent scattering data.

In this paper, we present the recent advances in the computation of the Dirac-Kohn-Sham (DKS) method of the BERTHA code. We show here that the simple underlined structure of the FORTRAN code also favors efficient porting of the code to GPUs, leading to a particularly efficient hybrid CPU/GPU implementation (OpenMP/OpenACC), where the most computationally intensive part for DKS matrix evaluation (three-center two-electron integrals evaluated via the McMurchie-Davidson scheme) is efficiently offloaded to the GPU via compiler directives based on the OpenACC programming model. This scheme in combination with the use of a linear algebra library optimized for GPUs (cuBLAS, cuSOLVER) significantly accelerates the DKS calculations. In addition, the low-level integral kernel developed here at FORTRAN level was used to port our real-time DKS (RT-TDDKS) implementation based on Python (PyBERTHART) for the utilization of the GPU. The results obtained on the new Tier-0 EuroHPC supercomputer (LEONARDO) of the CINECA Supercomputing Centre with a single NVIDIA A100 card are very satisfactory. We achieve a speedup up to 30 for Au₁₆ in a single-point DKS energy calculation and up to 10 for the Au₈ systems in an RT-TDDKS calculation, compared to our OpenMP (i.e., CPU only) parallel implementation (with 32 cores). The approach presented here is very general and, to our knowledge, represents the first port of a Python API to GPUs based on a FORTRAN kernel for the evaluation of two-electron integrals. The implementation is currently limited to the use of a single GPU accelerator, but future paths to an actual exascale implementation are discussed.

This work introduces the Strategic Escape (SE) algorithm, an approach that systematically ensures effective exploration of the potential energy surface during global minimum searches for atomic clusters. The SE algorithm prioritizes the escape from local minima prior to geometry optimization, leveraging a combination of randomized direction vectors, distance-based uniqueness criteria, and covalent bonding heuristics. These principles enhance structural diversity and computational efficiency by reducing redundant geometry optimizations. Additionally, a symmetry-guided seed generation method based on an adaptive polygon is proposed to provide diverse and physically realistic initial configurations. Together, these methods achieve a 2.3-fold improvement in computational efficiency compared to conventional Basin-Hopping approaches. The effectiveness of the SE algorithm is demonstrated through its application to boron, metal clusters, and binary-composition clusters, achieving rapid convergence to global minimum structures with high reliability. These advancements establish the SE algorithm as a robust and scalable tool for exploring complex chemical systems.

Linear regression equations were developed for different density functionals using data from the CCCBDB, along with a test set of 89 ionization energies (IE) and 76 electron affinities (EA) so that experimental IE and EA can be predicted from orbital energies. Separate equations were determined for different classes of atoms and molecules. These relationships were also applied to all occupied orbitals to simulate the photoemission spectra of organic molecules with accuracy similar to that of other computational methods at a fraction of the cost. The error for large molecules (up to 200 atoms) can be below 0.2 eV with many functionals for the prediction of the IE and EA.

Given the great importance of linear alkanes in fundamental and applied research, an accurate machine-learned potential (MLP) would be a major advance in computational modeling of these hydrocarbons. Recently, we reported a novel, many-body permutationally invariant model that was trained specifically for the 44-atom hydrocarbon C₁₄H₃₀ on roughly 250,000 B3LYP energies (Qu, C.; Houston, P. L.; Allison, T.; Schneider, B. I.; Bowman, J. M. J. Chem. Theory Comput. 2024, 20, 9339-9353). Here, we demonstrate the accuracy of the transferability of this potential for linear alkanes ranging from butane C₄H₁₀ up to C₃₀H₆₂. Unlike other approaches for transferability that aim for universal applicability, the present approach is targeted for linear alkanes. The mean absolute error (MAE) for energy ranges from 0.26 kcal/mol for butane and rises to 0.73 kcal/mol for C₃₀H₆₂ over the energy range up to 80 kcal/mol for butane and 600 kcal/mol for C₃₀H₆₂. These values are unprecedented for transferable potentials and indicate the high performance of a targeted transferable potential. The conformational barriers are shown to be in excellent agreement with high-level ab initio calculations for pentane, the largest alkane for which such calculations have been reported. Vibrational power spectra of C₃₀H₆₂ from molecular dynamics calculations are presented and briefly discussed. Finally, the evaluation time for the potential is shown to vary linearly with the number of atoms.

Computer simulations play a pivotal role in interpreting experimental two-photon absorption (2PA) spectra. One of the key aspects of the simulation of these spectra is to take into account the vibrational fine structure of the bands in electronic spectra. This is typically done by employing Franck-Condon (FC) term and low-order terms in the Herzberg-Teller (HT) expansion. In this work, we present a systematic study of first-order HT effects on the vibronic structure of π → π* electronic bands in 2PA spectra of 13 common fluorophores. We begin by evaluating the performance of several density functional approximations (DFAs) against the second-order coupled cluster singles and doubles model (CC2) for reproducing two-photon transition moments and their first- and second-order derivatives with respect to normal modes of vibration on a set of six donor-acceptor molecules. Our findings reveal that most DFAs produce inaccurate values for these derivatives, with the exception of the LC-BLYP functionals with range-separation parameters of 0.33 and 0.47. Although these functionals underestimate the HT contribution to the 2PA total intensities of the π → π* electronic bands, they offer a reasonable qualitative reproduction of the HT vibrational fine structure of the reference spectra. We further explore HT effects on fluorescent chromophores, finding that HT contributions are secondary to FC effects, leading to small shifts of the wavelengths peaks, and minimal changes in the intensities. Additionally, the adiabatic Hessian, vertical Hessian, and vertical gradient vibronic models are assessed. The general agreement among these models confirms that the harmonic approximation is suitable for studying the selected fluorophores.