Searcharxiv⌕ Search

arXiv subjects

L. R. Pratt

Publications and source records attributed to L. R. Pratt.

25 records · Page 2Linked to original sources

Gaussian Models for the Statistical Thermodynamics of Liquid Water

A gaussian distribution of binding energies, but conditioned to exploit generally available information on packing in liquids, provides a statistical-thermodynamic theory of liquid water that is structurally non-committal, molecularly realistic, and surprisingly accurate. Neglect of fluctuation contributions to this gaussian model yields a mean-field theory that produces useless results. A refinement that accounts for sharper-than-gaussian behavior at high binding energies recognizes contributions from a discrete number of water molecules and permits a natural matching of numerically exact results. These gaussian models, which can be understood as vigorous simplifications of quasi-chemical theories, are applicable to aqueous environments where the utility of structural models based on geometrical considerations of water hydrogen bonding have not been established.

physics.bio-ph↗

Molecular Realism in Default Models for Information Theories of Hydrophobic Effects

This letter considers several physical arguments about contributions to hydrophobic hydration of inert gases, constructs default models to test them within information theories, and gives information theory predictions using those default models with moment information drawn from simulation of liquid water. Tested physical features include: packing or steric effects, the role of attractive forces that lower the solvent pressure, and the roughly tetrahedral coordination of water molecules in liquid water. Packing effects (hard sphere default model) and packing effects plus attractive forces (Lennard-Jones default model) are ineffective in improving the prediction of hydrophobic hydration free energies of inert gases over the previously used Gibbs and flat default models. However, a conceptually simple cluster Poisson model that incorporates tetrahedral coordination structure in the default model is one of the better performers for these predictions. These results provide a partial rationalization of the remarkable performance of the flat default model with two moments in previous applications. The cluster Poisson default model thus will be the subject of further refinement.

physics.chem-ph↗

Hydrophobic Effects on a Molecular Scale

A theoretical approach is developed to quantify hydrophobic hydration and interactions on a molecular scale, with the goal of gaining insight into the molecular origins of hydrophobic effects. The model is based on the fundamental relation between the probability for cavity formation in bulk water resulting from molecular-scale density fluctuations, and the hydration free energy of the simplest hydrophobic solute, a hard particle. This probability is estimated using an information theory (IT) approach, incorporating experimentally available properties of bulk water -- the density and radial distribution function. The IT approach reproduces the simplest hydrophobic effects: hydration of spherical nonpolar solutes, the potential of mean force between methane molecules, and solvent contributions to the torsional equilibrium of butane. Applications of this approach to study temperature and pressure effects provide new insights into the thermodynamics and kinetics of protein folding. The IT model relates the hydrophobic-entropy convergence observed in protein unfolding experiments to the macroscopic isothermal compressibility of water. A novel explanation for pressure denaturation of proteins follows from an analysis of the pressure stability of hydrophobic aggregates, suggesting that water penetrates the hydrophobic core of proteins at high pressures. This resolves a long-standing puzzle, whether pressure denaturation contradicts the hydrophobic-core model of protein stability. Finally, issues of ``dewetting'' of molecularly large nonpolar solutes are discussed in the context of a recently developed perturbation theory approach.

physics.chem-ph↗

Construction of Simulation Wavefunctions for Aqueous Species: D3O+

This paper investigates Monte Carlo techniques for construction of compact wavefunctions for the internal atomic motion of the D3O+ ion. The polarization force field models of Stillinger, et al and of Ojamae, et al. were used. Initial pair product wavefunctions were obtained from the asymptotic high temperature many-body density matrix after contraction to atom pairs using Metropolis Monte Carlo. Subsequent characterization shows these pair product wavefunctions to be well optimized for atom pair correlations despite that fact that the predicted zero point energies are too high. The pair product wavefunctions are suitable to use within variational Monte Carlo, including excited states, and density matrix Monte Carlo calculations. Together with the pair product wavefunctions, the traditional variational theorem permits identification of wavefunction features with significant potential for further optimization. The most important explicit correlation variable found for the D3O+ ion was the vector triple product {\bf r}$_{OD1}\cdot$({\bf r}$_{OD2}\times${\bf r}$_{OD3}$). Variational Monte Carlo with 9 of such explicitly correlated functions yielded a ground state wavefunction with an error of 5-6% in the zero point energy.

physics.class-ph↗

Boundary Integral Methods for the Poisson Equation of Continuum Dielectric Solvation Models

This paper tests a dielectric model for variation of hydration free energy with geometry of complex solutes in water. It works out some basic aspects of the theory of boundary integral methods for these problems. One aspect of the algorithmic discussion lays the basis for multigrid methods of solution, methods that are likely to be necessary for similarly accurate numerical solution of these models for much larger solutes. Other aspects of the algorithmic work show how macroscopic surfaces such as solution interfaces and membranes may be incorporated and also show how these methods can be transferred directly to periodic boundary conditions. This dielectric model is found to give interesting and helpful results for the variation in solvation free energy with solute geometry. However, it typically significantly over-stabilizes classic attractive ion-pairing configurations. On the basis of the examples and algorithmic considerations, we make some observations about extension of this continuum model incrementally to reintroduce molecular detail of the solvation structure.

chem-ph↗

A Heuristic Molecular Model of Hydrophobic Interactions

Hydrophobic interactions provide driving forces for protein folding, membrane formation, and oil-water separation. Motivated by information theory, the poorly understood nonpolar solute interactions in water are investigated. A simple heuristic model of hydrophobic effects in terms of density fluctuations is developed. This model accounts quantitatively for the central hydrophobic phenomena of cavity formation and association of inert gas solutes; it therefore clarifies the underlying physics of hydrophobic effects and permits important applications to conformational equilibria of nonpolar solutes and hydrophobic residues in biopolymers.

chem-ph↗

Mixed Direct-Iterative Methods for Boundary Integral Formulations of Dielectric Solvation Models

This paper describes a mixed direct-iterative method for boundary integral formulations of dielectric solvation models. We give an example for which a direct solution at thermal accuracy is nontrivial and for which Gauss-Seidel iteration diverges in rare but reproducible cases. This difficulty is analyzed by obtaining the eigenvalues and the spectral radius of the iteration matrix. This establishes that the nonconvergence is due to inaccuracies of the asymptotic approximations for the matrix elements for accidentally close boundary element pairs on different spheres. This difficulty is cured by checking for boundary element pairs closer than the typical spatial extent of the boundary elements and for those pairs performing an `in-line' Monte Carlo integration to evaluate the required matrix elements. This difficulty are not expected and have not been observed when only a direct solution is sought. Finally, we give an example application of these methods to deprotonation of monosilicic acid in water.

chem-ph↗