Parsimonious Potential Energy Surface Expansions Using Dictionary Learning with Multipass Greedy Selection.

Potential energy surfaces fit with basis set expansions have been shown to provide accurate representations of electronic energies and have enabled a variety of high-accuracy dynamics, kinetics, and spectroscopy applications. The number of terms in these expansions scales poorly with system size, a drawback that challenges their use for systems with more than ∼10 atoms. A solution is presented here using dictionary learning. Subsets of the full set of conventional basis functions are optimized using a newly developed multipass greedy regression method inspired by forward and backward selection methods from the statistics, signal processing, and machine learning literatures. The optimized representations have accuracies comparable to the full set but are 1 or more orders of magnitude smaller, and notably, the number of terms in the optimized multipass greedy expansions scales approximately linearly with the number of atoms.

Permutationally Invariant Polynomial Expansions with Unrestricted Complexity.

A general strategy is presented for constructing and validating permutationally invariant polynomial (PIP) expansions for chemical systems of any stoichiometry. Demonstrations are made for three categories of gas-phase dynamics and kinetics: collisional energy-transfer trajectories for predicting pressure-dependent kinetics, three-body collisions for describing transient van der Waals adducts relevant to atmospheric chemistry, and nonthermal reactivity via quasiclassical trajectories. In total, 30 systems are considered with up to 15 atoms and 39 degrees of freedom. Permutational invariance is enforced in PIP expansions with as many as 13 million terms and 13 permutationally distinct atom types by taking advantage of petascale computational resources. The quality of the PIP expansions is demonstrated through the systematic convergence of in-sample and out-of-sample errors with respect to both the number of training data and the order of the expansion, and these errors are shown to predict errors in the dynamics for both reactive and nonreactive applications. The parallelized code distributed as part of this work enables the automation of PIP generation for complex systems with multiple channels and flexible user-defined symmetry constraints and for automatically removing unphysical unconnected terms from the basis set expansions, all of which are required for simulating complex reactive systems.

Watching a hydroperoxyalkyl radical (•QOOH) dissociate.

A prototypical hydroperoxyalkyl radical (•QOOH) intermediate, transiently formed in the oxidation of volatile organic compounds, was directly observed through its infrared fingerprint and energy-dependent unimolecular decay to hydroxyl radical and cyclic ether products. Direct time-domain measurements of •QOOH unimolecular dissociation rates over a wide range of energies were found to be in accord with those predicted theoretically using state-of-the-art electronic structure characterizations of the transition state barrier region. Unimolecular decay was enhanced by substantial heavy-atom tunneling involving O-O elongation and C-C-O angle contraction along the reaction pathway. Master equation modeling yielded a fully a priori prediction of the pressure-dependent thermal unimolecular dissociation rates for the •QOOH intermediate-again increased by heavy-atom tunneling-which are required for global models of atmospheric and combustion chemistry.

The end of ice I.

The appearance of ice I in the smallest possible clusters and the nature of its phase coexistence with liquid water could not thus far be unraveled. The experimental and theoretical infrared spectroscopic and free-energy results of this work show the emergence of the characteristic hydrogen-bonding pattern of ice I in clusters containing only around 90 water molecules. The onset of crystallization is accompanied by an increase of surface oscillator intensity with decreasing surface-to-volume ratio, a spectral indicator of nanoscale crystallinity of water. In the size range from 90 to 150 water molecules, we observe mixtures of largely crystalline and purely amorphous clusters. Our analysis suggests that the liquid-ice I transition in clusters loses its sharp 1st-order character at the end of the crystalline-size regime and occurs over a range of temperatures through heterophasic oscillations in time, a process without analog in bulk water.

Water structure at the interface of alcohol monolayers as determined by molecular dynamics simulations and computational vibrational sum-frequency generation spectroscopy.

We investigate the structure of water at the interface of three long-chain alcohol monolayers differing in alkyl chain length through molecular dynamics simulations combined with modeling of vibrational sum-frequency generation (vSFG) spectra. The effects of alkyl chain parity on interfacial water are examined through extensive analysis of structural properties, hydrogen bonding motifs, and spectral features. Besides providing molecular-level insights into the structure of interfacial water, this study also demonstrates that, by enabling comparisons with experimental vSFG spectra, computational spectroscopy may be used to test and validate force fields commonly used in biomolecular simulations. The results presented here may serve as benchmarks for further investigations to characterize ice nucleation induced by alcohol monolayers.

Neat Water-Vapor Interface: Proton Continuum and the Nonresonant Background.

Whether the surface of neat water is "acidic" or "basic" remains an active and controversial field of research. Most of the experimental evidence supporting the preferential adsorption of H3O+ ions stems from nonlinear optical spectroscopy methods typically carried out at extreme pH conditions (pH < 1). Here, we use vibrational sum frequency spectroscopy (VSFS) to target the "proton continuum", an unexplored frequency range characteristic of hydrated protons and hydroxide ions. The VSFS spectra of neat water show a broad and nonzero signal intensity between 1700 and 3000 cm-1 in the three different polarization combinations examined. By comparing the SF response of water with that from dilute HCl and NaOH aqueous solutions, we conclude the intensity does not originate from either adsorbed H3O+ or OH- ions. Contributions from the nonresonant background are then critically considered by comparing the experimental results with many-body molecular dynamics (MB-MD) simulated spectra.

Molecular-Level Interpretation of Vibrational Spectra of Ordered Ice Phases.

We build on results from our previous investigation into ice Ih using a combination of classical many-body molecular dynamics (MB-MD) and normal mode (NM) calculations to obtain molecular-level information on the spectroscopic signatures in the OH stretching region for all seven of the known ordered crystalline ice phases. The classical MB-MD spectra are shown to capture the important spectral features by comparing with experimental Raman spectra. This motivates the use of the classical simulations in understanding the spectral features of the various ordered ice phases in molecular terms. This is achieved through NM analysis to first demonstrate that the MB-MD spectra can be well recovered through the transition dipole moments and polarizability tensors calculated from each NM. From the normal mode calculations, measures of the amount of symmetric and antisymmetric stretching are calculated for each ice, as well as an approximation of how localized each mode is. These metrics aid in viewing the ice phases on a continuous spectrum determined by their density. As in ice Ih, it is found that most of the other ordered ice phases have highly delocalized modes and their spectral features cannot, in general, be described in terms of molecular normal modes. The lone exception is ice VIII, the densest crystalline ice phase. Despite being found only at high pressure, the symmetry index shows a clear separation of symmetric and antisymmetric stretching modes, giving rise to two distinct features.

Preordering of water is not needed for ice recognition by hyperactive antifreeze proteins.

Antifreeze proteins (AFPs) inhibit ice growth in organisms living in cold environments. Hyperactive insect AFPs are particularly effective, binding ice through "anchored clathrate" motifs. It has been hypothesized that the binding of hyperactive AFPs to ice is facilitated by preordering of water at the ice-binding site (IBS) of the protein in solution. The antifreeze protein TmAFP displays the best matching of its binding site to ice, making it the optimal candidate to develop ice-like order in solution. Here we use multiresolution simulations to unravel the mechanism by which TmAFP recognizes and binds ice. We find that water at the IBS of the antifreeze protein in solution does not acquire ice-like or anchored clathrate-like order. Ice recognition occurs by slow diffusion of the protein to achieve the proper orientation with respect to the ice surface, followed by fast collective organization of the hydration water at the IBS to form an anchored clathrate motif that latches the protein to the ice surface. The simulations suggest that anchored clathrate order could develop on the large ice-binding surfaces of aggregates of ice-nucleating proteins (INP). We compute the infrared and Raman spectra of water in the anchored clathrate motif. The signatures of the OH stretch of water in the anchored clathrate motif can be distinguished from those of bulk liquid in the Raman spectra, but not in the infrared spectra. We thus suggest that Raman spectroscopy may be used to probe the anchored clathrate order at the ice-binding surface of INP aggregates.

Temperature Dependence of the Air/Water Interface Revealed by Polarization Sensitive Sum-Frequency Generation Spectroscopy.

The temperature dependence of the vibrational sum-frequency generation (vSFG) spectra of the air/water interface is investigated using many-body molecular dynamics (MB-MD) simulations performed with the MB-pol potential energy function. The vSFG spectra calculated for different polarization combinations are then analyzed in terms of molecular autocorrelation and cross-correlation contributions. To provide molecular-level insights into interfacial hydrogen-bonding topologies, which give rise to specific spectroscopic features, the vSFG spectra are further investigated by separating contributions associated with water molecules donating zero, one, or two hydrogen bonds to neighboring water molecules. This analysis suggests that the low frequency shoulder of the free OH peak which appears at ∼3600 cm-1 is primarily due to intermolecular couplings between both singly and doubly hydrogen-bonded molecules.

Orientational Distribution of Free O-H Groups of Interfacial Water is Exponential.

The orientational distribution of free O-H (O-D) groups at the H_{2}O- (D_{2}O-)air interface is investigated using combined molecular dynamics (MD) simulations and sum-frequency generation (SFG) experiments. The average angle of the free O-H groups, relative to the surface normal, is found to be ∼63°, substantially larger than previous estimates of 30°-40°. This discrepancy can be traced to erroneously assumed Gaussian or stepwise orientational distributions of free O-H groups. Instead, the MD simulation and SFG measurement reveal a broad and exponentially decaying orientational distribution. The broad orientational distribution indicates the presence of the free O-H group pointing down to the bulk. We ascribe the origin of such free O-H groups to the presence of capillary waves on the water surface.

Molecular Origin of the Vibrational Structure of Ice Ih.

An unambiguous assignment of the vibrational spectra of ice Ih remains a matter of debate. This study demonstrates that an accurate representation of many-body interactions between water molecules, combined with an explicit treatment of nuclear quantum effects through many-body molecular dynamics (MB-MD), leads to a unified interpretation of the vibrational spectra of ice Ih in terms of the structure and dynamics of the underlying hydrogen-bond network. All features of the infrared and Raman spectra in the OH stretching region can be unambiguously assigned by taking into account both the symmetry and the delocalized nature of the lattice vibrations as well as the local electrostatic environment experienced by each water molecule within the crystal. The high level of agreement with experiment raises prospects for predictive MB-MD simulations that, complementing analogous measurements, will provide molecular-level insights into fundamental processes taking place in bulk ice and on ice surfaces under different thermodynamic conditions.

Temperature-dependent vibrational spectra and structure of liquid water from classical and quantum simulations with the MB-pol potential energy function.

The structure of liquid water as a function of temperature is investigated through the modeling of infrared and Raman spectra along with structural order parameters calculated from classical and quantum molecular dynamics simulations with the MB-pol many-body potential energy function. The magnitude of nuclear quantum effects is also monitored by comparing the vibrational spectra obtained from classical and centroid molecular dynamics, both in intensities and peak positions. The observed changes in spectral activities are shown to reflect changes in the underlying structure of the hydrogen-bond network and are found to be particularly sensitive to many-body effects in the representation of the electrostatic interactions. Overall, good agreement is found with the experimental spectra, which provides further evidence for the accuracy of MB-pol in predicting the properties of water.

On the accuracy of the MB-pol many-body potential for water: Interaction energies, vibrational frequencies, and classical thermodynamic and dynamical properties from clusters to liquid water and ice.

The MB-pol many-body potential has recently emerged as an accurate molecular model for water simulations from the gas to the condensed phase. In this study, the accuracy of MB-pol is systematically assessed across the three phases of water through extensive comparisons with experimental data and high-level ab initio calculations. Individual many-body contributions to the interaction energies as well as vibrational spectra of water clusters calculated with MB-pol are in excellent agreement with reference data obtained at the coupled cluster level. Several structural, thermodynamic, and dynamical properties of the liquid phase at atmospheric pressure are investigated through classical molecular dynamics simulations as a function of temperature. The structural properties of the liquid phase are in nearly quantitative agreement with X-ray diffraction data available over the temperature range from 268 to 368 K. The analysis of other thermodynamic and dynamical quantities emphasizes the importance of explicitly including nuclear quantum effects in the simulations, especially at low temperature, for a physically correct description of the properties of liquid water. Furthermore, both densities and lattice energies of several ice phases are also correctly reproduced by MB-pol. Following a recent study of DFT models for water, a score is assigned to each computed property, which demonstrates the high and, in many respects, unprecedented accuracy of MB-pol in representing all three phases of water.

Thermal weights for semiclassical vibrational response functions.

Semiclassical approximations to response functions can allow the calculation of linear and nonlinear spectroscopic observables from classical dynamics. Evaluating a canonical response function requires the related tasks of determining thermal weights for initial states and computing the dynamics of these states. A class of approximations for vibrational response functions employs classical trajectories at quantized values of action variables and represents the effects of the radiation-matter interaction by discontinuous transitions. Here, we evaluate choices for a thermal weight function which are consistent with this dynamical approximation. Weight functions associated with different semiclassical approximations are compared, and two forms are constructed which yield the correct linear response function for a harmonic potential at any temperature and are also correct for anharmonic potentials in the classical mechanical limit of high temperature. Approximations to the vibrational linear response function with quantized classical trajectories and proposed thermal weight functions are assessed for ensembles of one-dimensional anharmonic oscillators. This approach is shown to perform well for an anharmonic potential that is not locally harmonic over a temperature range encompassing the quantum limit of a two-level system and the limit of classical dynamics.