Fast Molecular Dynamics Research Articles

Enhanced deep potential model for fast and accurate molecular dynamics: application to the hydrated electron.

In molecular simulations, neural network force fields aim at achieving ab initio accuracy with reduced computational cost. This work introduces enhancements to the Deep Potential network architecture, integrating a message-passing framework and a new lightweight implementation with various improvements. Our model achieves accuracy on par with leading machine learning force fields and offers significant speed advantages, making it well-suited for large-scale, accuracy-sensitive systems. We also introduce a new iterative model for Wannier center prediction, allowing us to keep track of electron positions in simulations of general insulating systems. We apply our model to study the solvated electron in bulk water, an ostensibly simple system that is actually quite challenging to represent with neural networks. Our trained model is not only accurate, but can also transfer to larger systems. Our simulation confirms the cavity model, where the electron's localized state is observed to be stable. Through an extensive run, we accurately determine various structural and dynamical properties of the solvated electron.

Incorporating Noncovalent Interactions in Transfer Learning Gaussian Process Regression Models for Molecular Simulations.

FFLUX is a quantum chemical topology-based multipolar force field that uses Gaussian process regression machine learning models to predict atomic energies and multipole moments on the fly for fast and accurate molecular dynamics simulations. These models have previously been trained on monomers, meaning that many-body effects, for example, intermolecular charge transfer, are missed in simulations. Moreover, dispersion and repulsion have been modeled using Lennard-Jones potentials, necessitating careful parametrization. In this work, we take an important step toward addressing these shortcomings and show that models trained on clusters, in this case, a dimer, can be used in FFLUX simulations by preparing and benchmarking a formamide dimer model. To mitigate the computational costs associated with training higher-dimensional models, we rely on the transfer of hyperparameters from a smaller source model to a larger target model, enabling an order of magnitude faster training than with a direct learning approach. The dimer model allows for simulations that account for two-body effects, including intermolecular polarization and charge penetration, and that do not require nonbonded potentials. We show that addressing these limitations allows for simulations that are closer to quantum mechanics than previously possible with the monomeric models.

Fast and accurate nonadiabatic molecular dynamics enabled through variational interpolation of correlated electron wavefunctions.

We build on the concept of eigenvector continuation to develop an efficient multi-state method for the rigorous and smooth interpolation of a small training set of many-body wavefunctions through chemical space at mean-field cost. The inferred states are represented as variationally optimal linear combinations of the training states transferred between the many-body bases of different nuclear geometries. We show that analytic multi-state forces and nonadiabatic couplings from the model enable application to nonadiabatic molecular dynamics, developing an active learning scheme to ensure a compact and systematically improvable training set. This culminates in application to the nonadiabatic molecular dynamics of a photoexcited 28-atom hydrogen chain, with surprising complexity in the resulting nuclear motion. With just 22 DMRG calculations of training states from the low-energy correlated electronic structure at different geometries, we infer the multi-state energies, forces and nonadiabatic coupling vectors at 12 000 geometries with provable convergence to high accuracy along an ensemble of molecular trajectories, which would not be feasible with a brute force approach. This opens up a route to bridge the timescales between accurate single-point correlated electronic structure methods and timescales of relevance for photo-induced molecular dynamics.

Theory of Liquids for Studying the Conformational Flexibility of Biomolecules with Reference Interaction Site Model Approximation

The theory of fluids is used to modify the integral equations of the reference interaction site model (RISM) approximation. Its applicability to the study of biomolecules solvation is evaluated. Unlike traditional RISM applications, the new integral equation contains an intramolecular correlation matrix that only needs to be calculated once. This allows us to bypass the effort of repeatedly solving RISM equations and the time-consuming averaging of values obtained for each time point of a molecular trajectory. The new approach allows for the assessment of the conformational transience of dissolved molecules while taking into account the effects of solvation. The free energy of oxytocin, which is a peptide hormone, as well as self-assembled ionic peptide complexes calculated using both the traditional RISM and the new RISM with average matrix (RISM-AM) approach are estimated. The free energy of oxytocin calculated using RISM-AM shows that the statistical error does not exceed the error obtained by standard averaging of solutions in the RISM equation. Despite the somewhat ambiguous results obtained for ionic peptide self-assembly using RISM-AM with Lennard–Jones repulsion correction, this method can still be considered applicable for fast molecular dynamics analysis. Since the required computational power can be reduced by at least two orders of magnitude, the medium-matrix RISM is indeed a highly applicable tool for studying macromolecular conformations as well as corresponding solvation effects.

Studying the mechanical behavior of a generic thermoplastic by means of a fast coarse-grained molecular dynamics model

Polymers play an emerging role in modern engineering applications due to their comparatively low cost, low density, and versatile manufacturing. The addition of nano-sized fillers further enhances the polymer’s properties but also induces a strong dependence on the resulting microstructure, particularly the matrix-filler interphase. Since an experimental characterization of this nano-sized interphase is extremely difficult, molecular dynamics (MD) simulations are used to study the effects at such small scales. However, MD’s high computational costs usually limit the scope of a mechanical characterization. Therefore, this study presents the methodology and tools to generate and analyze samples of an efficient generic thermoplastic model. In this first contribution, we focus on the neat polymer and introduce a versatile and numerically stable self-avoiding random walker with adjustable linearity of chain growth. Moreover, we verify our equilibration procedure by preparing samples in liquid and solid state which behave physically sound. Finally, we perform uniaxial tensile tests with a maximum strain of 10 % to evaluate the mechanical properties. In the liquid case, the polymer chains are sufficiently mobile, such that the tensile stresses fluctuate only around zero, while the solid exhibits an almost linear elastic regime followed by a nonlinear part. This contribution forms the basis for a thorough mechanical characterization of polymer nanocomposites which we will address in future studies. The methodology and tools introduced are not limited to our generic polymer, but applicable to many coarse-grained models.

Extending a generic and fast coarse-grained molecular dynamics model to examine the mechanical behavior of grafted polymer nanocomposites

Polymer nanocomposites are an important class of materials for engineering applications due to their high versatility and good mechanical properties combined with low density. By directly attaching the polymer chains to the nanofillers, the so-called grafting, a better load transfer between matrix and filler is achieved, and, in addition, a better dispersion of the fillers is obtained. Both result in enhanced mechanical properties. Since experimental investigations on the nanoscale are extremely challenging, complementary numerical studies are needed to unravel the mechanical behavior of polymer nanocomposites. To this end, molecular dynamics is ideally suited since it captures the microstructure, but is also numerically expensive. Therefore, this contribution presents a fast coarse-grained molecular dynamics model for the investigation of the mechanical behavior of grafted polymer nanocomposites. For this purpose, we extend an existing model by grafting bonds, which allows us to compare the effect of untreated and grafted fillers directly. In particular, we investigate the influence of filler content, grafting degree, and filler size on the stiffness and strength of the polymer (grafted) nanocomposites. We conclude that the grafting bonds have little effect on the stiffness, while the strength is significantly improved compared to the untreated fillers, which is in agreement with the literature. The presented molecular dynamics model for polymer grafted nanocomposites provides the basis for further investigations, particularly of the crucial matrix-filler interphase. In addition, this contribution translates molecular dynamics insights into mechanical properties, which bridges the gap to the engineering scale and thus represents a step towards exploiting the full potential of polymer (grafted) nanocomposites.

Achieving High Temporal Resolution in Single‐Molecule Fluorescence Techniques Using Plasmonic Nanoantennas

AbstractSingle‐molecule fluorescence techniques are essential for investigating the molecular mechanisms in biological processes. However, achieving sub‐millisecond temporal resolution to monitor fast molecular dynamics remains a significant challenge. The fluorescence brightness is the key parameter that generally defines the temporal resolution for these techniques. Conventional microscopes and standard fluorescent emitters fall short in achieving the high brightness required for sub‐millisecond monitoring. Plasmonic nanoantennas are proposed as a solution, but despite huge fluorescence enhancement having been obtained with these structures, the brightness generally remains below 1 million photons/s/molecule. Therefore, the improvement of temporal resolution is overlooked. This article presents a method for achieving high temporal resolution in single‐molecule fluorescence techniques using plasmonic nanoantennas, specifically optical horn antennas. This work demonstrates about 90% collection efficiency of the total emitted light, reaching a high fluorescence brightness of 2 million photons/s/molecule in the saturation regime. This enables observations of single molecules with microsecond binning time and fast fluorescence correlation spectroscopy measurements. This work expands the applications of plasmonic antennas and zero‐mode waveguides in the fluorescence saturation regime toward brighter single‐molecule signal, faster temporal resolutions, and improved detection rates to advance fluorescence sensing, DNA sequencing, and dynamic studies of molecular interactions.

Unusual water molecules at surfactant, lipid, and protein interfaces.

The structure and dynamics of water has captivated the interest of researchers for decades, motivated in large measure by the importance of these properties in biological systems at interfaces. However, techniques that can routinely deal with delicate, dynamic, and heterogenous natural systems, such as proteins embedded near or in lipid membranes, remains an area of active development. Magnetic Resonance (MR) experiments rely on resonance frequencies many orders of magnitude slower than those of, e.g., ultrafast laser experiments that have shown earlier promise for interrogating interfacial systems. Nonetheless, MR relaxometry techniques can still probe fast molecular dynamics in the frequency domain through a set of techniques known as relaxometry.

M-Chem: a modular software package for molecular simulation that spans scientific domains

ABSTRACT We present a new software package called M-Chem that is designed from scratch in C++ and parallelised on shared-memory multi-core architectures to facilitate efficient molecular simulations. Currently, M-Chem is a fast molecular dynamics (MD) engine that supports the evaluation of energies and forces from two-body to many-body all-atom potentials, reactive force fields, coarse-grained models, combined quantum mechanics molecular mechanics (QM/MM) models, and external force drivers from machine learning, augmented by algorithms that are focused on gains in computational simulation times. M-Chem also includes a range of standard simulation capabilities including thermostats, barostats, multi-timestepping, and periodic cells, as well as newer methods such as fast extended Lagrangians and high quality electrostatic potential generation. At present M-Chem is a developer friendly environment in which we encourage new software contributors from diverse fields to build their algorithms, models, and methods in our modular framework. The long-term objective of M-Chem is to create an interdisciplinary platform for computational methods with applications ranging from biomolecular simulations, reactive chemistry, to materials research.

Accelerating Hybrid Density Functional Theory Molecular Dynamics Simulations by Seminumerical Integration, Resolution-of-the-Identity Approximation, and Graphics Processing Units.

The computationally very demanding evaluation of the 4-center-2-electron (4c2e) integrals and their respective integral derivatives typically represents the major bottleneck within hybrid Kohn-Sham density functional theory molecular dynamics simulations. Building upon our previous works on seminumerical exact-exchange (sn-LinK) [Laqua, H., Thompsons, T. H., Kussmann, J., Ochsenfeld, C., J. Chem. Theory Comput. 2020, 16, 1465] and resolution-of-the-identity Coulomb (RI-J) [Kussmann, J., Laqua, H., Ochsenfeld, C., J. Chem. Theory Comput. 2021, 17, 1512], the expensive 4c2e integral evaluation can be avoided entirely, resulting in a highly efficient electronic structure theory method, allowing for fast ab initio molecular dynamics (AIMD) simulations even with large basis sets. Moreover, we propose to combine the final self-consistent field (SCF) step with the subsequent nuclear forces evaluation, providing the forces at virtually no additional cost after a converged SCF calculation, reducing the total runtime of an AIMD simulation by about another 25%. In addition, multiple independent MD trajectories can be computed concurrently on a single node, leading to a greatly increased utilization of the available hardware─especially when combined with graphics processing unit acceleration─improving the overall throughput by up to another 5 times in this way. With all of those optimizations combined, our proposed method provides nearly 3 orders of magnitude faster execution times than traditional 4c2e integral-based methods. To demonstrate the practical utility of the approach, quantum-mechanical/molecular-mechanical dynamics simulations on double-stranded DNA were performed, investigating the relative hydrogen bond strength between adenine-thymine and guanine-cytosine base pairs. In addition, this illustrative application also contains a general accuracy assessment of the introduced approximations (integration grids, resolution-of-the-identity) within AIMD simulations, serving as a protocol on how to apply these new methods to practical problems.

Applying a generic and fast coarse-grained molecular dynamics model to extensively study the mechanical behavior of polymer nanocomposites

Applying a generic and fast coarse-grained molecular dynamics model to extensively study the mechanical behavior of polymer nanocomposites

Implementation of 3D Multi-Color Fluorescence Microscopy in a Quadruple Trap Optical Tweezers System.

Recent advances in the design and measurement capabilities of optical tweezers instruments, and especially the combination with multi-color fluorescence detection, have accommodated a dramatic increase in the versatility of optical trapping. Quadruple (Q)-trap optical tweezers are an excellent example of such an advance, by providing three-dimensional control over two constructs and thereby enabling for example DNA-DNA braiding. However, the implementation of fluorescence detection in such a Q-trapping system poses several challenges: (1) since typical samples span a distance in the order of tens of micrometers, it requires imaging of a large field of view, (2) in order to capture fast molecular dynamics, fast imaging with single-molecule sensitivity is desired, (3) in order to study three-dimensional objects, it could be needed to detect emission light at different axial heights while keeping the objective lens and thus the optically trapped microspheres in a fixed position. In this chapter, we describe design guidelines for a fluorescence imaging module on a Q-trap system that overcomes these challenges and provide a step-by-step description for construction and alignment of such a system. Finally, we present detailed instructions for proof-of-concept experiments that can be used to validate and highlight the capabilities of the instruments.

Thermodynamic and Transport Properties of LiF and FLiBe Molten Salts with Deep Learning Potentials.

Molten salts have attracted interest as potential heat carriers and/or fuel solvents in the development of new Gen IV nuclear reactor designs, high-temperature batteries, and thermal energy storage. In nuclear engineering, salts containing lithium fluoride-based compounds are of particular interest due to their ability to lower the melting points of mixtures and their compatibility with alloys. A machine learning potential (MLP) combined with a molecular dynamics study is performed on two popular molten salts, namely, LiF (50% Li) and FLiBe (66% LiF and 33% BeF2), to predict the thermodynamic and transport properties, such as density, diffusion coefficients, thermal conductivity, electrical conductivity, and shear viscosity. Due to the large possibilities of atomic environments, we employ training using Deep Potential Smooth Edition (DPSE) neural networks to learn from large datasets of 141,278 structures with 70 atoms for LiF and 238,610 structures with 91 atoms for FLiBe molten salts. These networks are then deployed in fast molecular dynamics to predict the thermodynamic and transport properties that are only accessible at longer time scales and are otherwise difficult to calculate with classical potentials, ab initio molecular dynamics, or experiments. The prospect of this work is to provide guidance for future works to develop general MLPs for high-throughput thermophysical database generation for a wide spectrum of molten salts.

Density Functional Theory Half-Electron Self-Energy Correction for Fast and Accurate Nonadiabatic Molecular Dynamics.

The nonadiabatic (NA) process is crucial to photochemistry and photophysics and requires an atomistic understanding. However, conventional NA molecular dynamics (MD) for condensed-phase materials on the nanoscale are generally limited to the semilocal exchange-correlation functional, which suffers from the bandgap and thus NA coupling (NAC) problems. We consider TiO2 and a black phosphorus monolayer as two prototypical systems, perform NA-MD simulations of nonradiative electron-hole recombination, and demonstrate for the first time that density functional theory (DFT) half-electron self-energy correction can reproduce the bandgap, effective masses of carriers, luminescence line widths, NAC, and excited-state lifetimes of the two systems at the hybrid functional level while the computational cost remains at that of the Predew-Burke-Ernzerhof functional. Our study indicates that the DFT-1/2 method can greatly accelerate NA-MD simulations while maintaining the accuracy of the hybrid functional, providing an advantage for studying photoexcitation dynamics for large-scale condensed-phase materials.

Conformational Shifts of Stacked Heteroaromatics: Vacuum vs. Water Studied by Machine Learning.

Stacking interactions play a crucial role in drug design, as we can find aromatic cores or scaffolds in almost any available small molecule drug. To predict optimal binding geometries and enhance stacking interactions, usually high-level quantum mechanical calculations are performed. These calculations have two major drawbacks: they are very time consuming, and solvation can only be considered using implicit solvation. Therefore, most calculations are performed in vacuum. However, recent studies have revealed a direct correlation between the desolvation penalty, vacuum stacking interactions and binding affinity, making predictions even more difficult. To overcome the drawbacks of quantum mechanical calculations, in this study we use neural networks to perform fast geometry optimizations and molecular dynamics simulations of heteroaromatics stacked with toluene in vacuum and in explicit solvation. We show that the resulting energies in vacuum are in good agreement with high-level quantum mechanical calculations. Furthermore, we show that using explicit solvation substantially influences the favored orientations of heteroaromatic rings thereby emphasizing the necessity to include solvation properties starting from the earliest phases of drug design.

Molecular Insight into Cu2+-Induced Conformational Transitions of Amyloid β-Protein from Fast Kinetic Analysis and Molecular Dynamics Simulations.

Cu2+-mediated amyloid β-protein (Aβ) aggregation is implicated in the pathogenesis of Alzheimer's disease, so it is of significance to understand Cu2+-mediated conformational transitions of Aβ. Herein, four Aβ mutants were created by using the environment-sensitive cyanophenylalanine to respectively substitute F4, Y10, F19, and F20 residues of Aβ40. By using stopped-flow fluorescence spectroscopy and molecular dynamics (MD) simulations, the early stage conformational transitions of the mutants mediated by Cu2+ binding were investigated. The fast kinetics unveils that Cu2+ has more significant influence on the conformational changes of N-terminal (F4 and Y10) than on the central hydrophobic core (CHC, F19, and F20) under different pH conditions (pH 6.6-8.0), especially Y10. Interestingly, lag periods of the conformational transitions are observed for the F19 and F20 mutants at pH 8.0, indicating the slow response of the two mutation sites on the conformational transitions. More importantly, significantly longer lag periods for F20 than for F19 indicate the conduction of the transition from F19 to F20. The conduction time (difference in lag period) decreases from 4.5 s at Cu2+ = 0 to undetectable (<1 ms) at Cu2+ = 10 μM. The significant difference in the response time of F19 and F20 and the fast local conformational changes of Y10 imply that the conformational transitions of Aβ start around Y10. MD simulations support the observation of hydrophobicity increase at N-terminal during the conformational transitions of Aβ-Cu2+. It also reveals that Y10 is immediately approached by Cu2+, supporting the speculation that the starting point of conformational transitions of Aβ is near Y10. The work has provided molecular insight into the early stage conformational transitions of Aβ40 mediated by Cu2+.

Fast Nonadiabatic Molecular Dynamics via Spin-Flip Time-Dependent Density-Functional Tight-Binding Approach: Application to Nonradiative Relaxation of Tetraphenylethylene with Locked Aromatic Rings.

Nonadiabatic dynamics around conical intersections between ground and excited states are crucial to understand excited-state phenomena in complex chemical systems. With this background in mind, we present an approach combining fewest-switches trajectory surface hopping and spin-flip (SF) time-dependent (TD) density-functional tight binding (DFTB), which is a simplified version of SF-TD density functional theory (DFT) with semiempirical parametrizations, for computationally efficient nonadiabatic molecular dynamics simulations. The estimated computational time of the SF-TD-DFTB approach is several orders of magnitude lower than that of SF-TD-DFT. In addition, the proposed method reproduces the time scales and quantum yields in photoisomerization reactions of azobenzene at a level comparable with conventional ab initio approaches, demonstrating reasonable accuracy. Finally, we report a practical application of the developed technique to explore the nonradiative relaxation processes of tetraphenylethylene and its derivative with torsionally locked aromatic rings and discuss the effect of locking the rings on the excited-state lifetime.

Molecular dynamics simulation as a tool for prediction of the properties of TiO2 and TiO2:MoO3 based chemical gas sensors

Fast and simple molecular dynamics simulation method was used to investigate the interaction of water, hydrogen, methane, and ethanol molecules with {100} set atomic planes of anatase TiO2 and α-MoO3 at 300 K and 573 K. Preliminary molecular simulation of gas molecules interaction with oxide surface provides an opportunity to eliminate empirical search of the oxide material with preassigned gas sensing properties. Facilities of molecular dynamics simulation approach without involving time-consuming quantum chemical calculations turn to be sufficient for this purpose. From the molecular simulation results it followed, that the most energy-efficient interaction of hydrogen molecules turned to be with (001) anatase plane. The less energy-efficient and localized among all the investigated processes was the interaction of methane molecules both with anatase and MoO3 surfaces. Interaction of water molecules was more energy-efficient with anatase surface than with MoO3 surface. Molecular dynamics simulation results were confronted with experimental results on gas sensing properties of one-electrode thermocatalytic chemical gas sensors on the basis of anatase TiO2 and TiO2:MoO3 composite materials. Predicted low output signal value of the sensors towards CH4, high output value towards hydrogen with its decrease at high MoO3 content in TiO2:MoO3 composite and sufficiently high sensitivity towards ethanol were proved in practice.

Probing relaxation models by means of Fast Field-Cycling relaxometry, NMR spectroscopy and molecular dynamics simulations: Detailed insight into the translational and rotational dynamics of a protic ionic liquid

A combination of temperature dependent Fast Field-Cycling (FFC) relaxometry and high-resolution NMR spectroscopy provides more than five orders of magnitude in frequency range for studying translational and rotational dynamics of ionic liquids. However, to make use of this broad frequency range, certain requirements have to be met: The viscosity of the liquid has to be of the right order of magnitude, and the material has to exhibit a sufficiently broad liquid range. In addition, NMR sensitive nuclei have to be present (ideally) on both, cation and anion, and appropriate relaxation models properly describing the molecular dynamics have to be available. For the latter, Molecular Dynamics (MD) simulations are ideally suited to suggest meaningful relaxation models. In this study we employ the protic ionic liquid triethylammonium bis(trifluoromethylsulfonyl)-imide [TEA][NTf2] as model compound to gain insight into details of the molecular dynamical processes. By addressing different NMR sensitive nuclei on both ions, 1H nuclei on the triethylammonium cation, and 19F on the NTf2 anion, we are able to obtain translational dynamics as well as rotational dynamics for both species at the same time. The obtained temperature dependent translatoric diffusion coefficients are consistent with our MD simulations, and are found to be in agreement with data reported in the literature. In addition, two types of NMR relaxation processes are employed to investigate the intramolecular relaxation: (1) dipolar relaxation at low frequencies (employing FFC) addressing 1H and 19F nuclei, and (2) quadrupolar relaxation at high frequencies addressing 2H nuclei. In near quantitative agreement with MD simulations, we show that both, dipolar relaxation obtained via FFC and quadrupolar relaxation obtained via high-field NMR are leading to consistent rotational correlation times, suggesting that the rotational motion of the cation is dominantly isotropic. Hence, the intramolecular 1H relaxation rate of the cation determined via FFC can be appropriately expressed by a single rotational correlation time, compatible with the Bloembergen Purcell Pound (BPP) approach. MD simulations of the NTf2 anion, however, suggest that its reorientational dynamics is strongly anisotropic. Consequently, the proper description of the intramolecular 19F relaxation rate of the anion requires a more complex model beyond the BPP approach employing at least two correlation times: By taking internal rotation of the CF3 groups into account, we are able to determine correlation times compatible with our MD simulations.

Efficient Atomic-Resolution Uncertainty Estimation for Neural Network Potentials Using a Replica Ensemble.

Neural network potentials (NNPs) are gaining much attention as they enable fast molecular dynamics (MD) simulations for a wide range of systems while maintaining the accuracy of density functional theory calculations. Since NNP is constructed by machine learning on training data, its prediction uncertainty increases drastically as atomic environments deviate from training points. Therefore, it is essential to monitor the uncertainty level during MD simulations to judge the soundness of the results. In this work, we propose an uncertainty estimator based on the replica ensemble in which NNPs are trained over atomic energies of a reference NNP that drives MD simulations. The replica ensemble is trained quickly, and its standard deviation provides atomic-resolution uncertainties. We apply this method to a highly reactive silicidation process of Si(001) overlaid with Ni thin films and confirm that the replica ensemble can spatially and temporally trace simulation errors at atomic resolution, which in turn guides the augmentation of the training set. The refined NNP completes a 3.6 ns simulation without any noticeable problems. By suggesting an efficient and atomic-resolution uncertainty indicator, this work will contribute to achieving reliable MD simulations by NNPs.