Using a classical potential as an efficient importance function for sampling from an ab initio potential

Abstract

In this paper the ab initio potential of mean force for the formic acid-water system is calculated in a Monte Carlo simulation using a classical fluctuating charge molecular mechanics potential to guide Monte Carlo updates. The ab initio energies in the simulation are calculated using density-functional theory ͑DFT͒ methods recently developed by Salahub et al. ͓J. Chem. Phys. 107, 6770 ͑1997͔͒ to describe hydrogen-bonded systems. Importance sampling methods are used to investigate structural changes and it is demonstrated that using a molecular mechanics importance function can improve the efficiency of a DFT simulation by several orders of magnitude. Monte Carlo simulation of the system in a canonical ensemble at Tϭ300 K reveals two chemical processes at intermediate time scales: The rotation of the H 2 O bonded to HCOOH, which takes place on a time scale of 3 ps, and the dissociation of the complex which occurs in 24 ps. It is shown that these are the only important structural ''reactions'' in the formic acid-water cluster which take place on a time scale shorter than the double transfer of the proton.

Bulk Liquid Water at Ambient Temperature and Pressure from MP2 Theory

“…NpT, in a relatively straightforward manner. On the other hand, an efficient sampling of phase space in MC requires smart and system dependent trial moves [30][31][32][33][34] , making the application of the method more intricate than molecular dynamics where configurational sampling follows a general principle. Within the framework of MC, McGrath and coworkers 27 reported the first results from first-principles simulations of liquid water in the isobaric-isothermal ensemble at ambient pressure.…”

“…The MC efficiency is improved with the presampling of moves. 30,32 In this method an additional inexpensive (approximated) potential is introduced and used to generate a short sequence of MC moves which is entirely accepted or rejected based on the difference between the exact and approximated potential functions. The employed approximate potential is a classical but refitted force field based on Ref.…”

Simulating enzyme reactions: Challenges and perspectives

“…Several such schemes have been developed, including geometry optimization methods, 65,80 in which extensive minimizations in the MM subspace are coupled to less frequent geometry updates of the QM region, and methods for the calculation of free energies, in which the sampling of the QM part of the potential energy surface is enhanced by the use of approximate replacement potentials, of either EVB- 36,81 or of fluctuating charge-type. 82 To end this section, a new class of methods will be considered that are having a large impact in the investigation of reaction processes, although they do not yet appear to have been applied to enzyme systems. Traditionally, calculations of the rate of a condensed-phase reaction have employed a transition-state or activated-dynamics approach (see, e.g., ref.…”

An efficient Monte Carlo method for calculating ab initio transition state theory reaction rates in solution

“…A number of techniques for computing free energy profiles along reaction coordinates have been proposed in the literature, including umbrella sampling, 8 thermodynamic integration in conjunction with the blue-moon ensemble method, 9 projection methods, 10 variable transformation approaches 11 and guiding potentials. 12, 13 The use of molecular mechanics-based guiding potentials was proposed simultaneously and independently by Iftimie et al 12 and Vondele et al 13 and implemented in a Monte Carlo and a molecular dynamics framework, respectively. The basic idea of the method consists of using a "fast" molecular mechanics potential to guide a computationally intensive ab initio simulation.…”

“…The Monte Carlo version of the "guiding" approach in reference [12] was called the molecular mechanics based importance function method (MMBIF). It was demonstrated that the utilization of a reasonably accurate molecular mechanics potential as an importance function decreases the correlation of an ab initio Monte Carlo calculation by two orders of magnitude.…”