Quantum Monte Carlo technique clusters

Another branch of computational quantum mechanics, quantum Monte Carlo, is described in Chapter 3 by Professor James B. Anderson. Quantum Monte Carlo techniques, such as variational, diffusion, and Green s function, are explained, along with applications to atoms, molecules, clusters, liquids, and solids. Quantum Monte Carlo is not as widely used as other approaches to solving the Schrodinger equation for the electronic structure of a system, and the programs for running these calculations are not as user friendly as those based on the Hartree-Fock approach. This chapter sheds much needed light on the topic. 

Static charge-density susceptibilities have been computed ab initio by Li et al (38). The frequency-dependent susceptibility x(r, r cd) can be calculated within density functional theory, using methods developed by Ando (39 Zang-will and Soven (40 Gross and Kohn (4I and van Gisbergen, Snijders, and Baerends (42). In ab initio work, x(r, r co) can be determined by use of time-dependent perturbation techniques, pseudo-state methods (43-49), quantum Monte Carlo calculations (50-52), or by explicit construction of the linear response function in coupled cluster theory (53). Then the imaginary-frequency susceptibility can be obtained by analytic continuation from the susceptibility at real frequencies, or by a direct replacement co ico, where possible (for example, in pseudo-state expressions). 

In this review, almost all of the simulations we have described use only classical mechanics to describe the nuclear motion of the reaction system. However, a more accurate analysis of many reactions, including some of the ones that have already been simulated via purely classical mechanics, will ultimately require some infusion of quantum mechanical methods. This infusion has already taken place in several different types of reaction dynamics electron transfer in solution, > i> 2 HI photodissociation in rare gas clusters and solids,i i 22 >2 ° I2 photodissociation in Ar fluid,and the dynamics of electron solvation.22-24 Since calculation of the quantum dynamics of a full solvent is at present too time-consuming, all of these calculations involve a quantum solute in a classical solvent. (For a system where the solvent is treated quantum mechanically, see the quantum Monte Carlo treatment of an electron transfer reaction in water by Bader et al. O) As more complex reaaions are investigated, the techniques used in these studies will need to be extended to take into account effects involving electron dynamics such as curve crossing, the interaction of multiple electronic surfaces and other breakdowns of the Born-Oppenheimer approximation, the effect of solvent and solute polarization, and ultimately the actual detailed dynamics of the time evolution of the electronic degrees of freedom. 

Beyond the clusters, to microscopically model a reaction in solution, we need to include a very big number of solvent molecules in the system to represent the bulk. The problem stems from the fact that it is computationally impossible, with our current capabilities, to locate the transition state structure of the reaction on the complete quantum mechanical potential energy hypersurface, if all the degrees of freedom are explicitly included. Moreover, the effect of thermal statistical averaging should be incorporated. Then, classical mechanical computer simulation techniques (Monte Carlo or Molecular Dynamics) appear to be the most suitable procedures to attack the above problems. In short, and applied to the computer simulation of chemical reactions in solution, the Monte Carlo [18-21] technique is a numerical method in the frame of the classical Statistical Mechanics, which allows to generate a set of system configurations... 

Semiclassical techniques like the instanton approach [211] can be applied to tunneling splittings. Finally, one can exploit the close correspondence between the classical and the quantum treatment of a harmonic oscillator and treat the nuclear dynamics classically. From the classical trajectories, correlation functions can be extracted and transformed into spectra. The particular charm of this method rests in the option to carry out the dynamics on the fly, using Born Oppenheimer or fictitious Car Parrinello dynamics [212]. Furthermore, multiple minima on the hypersurface can be treated together as they are accessed by thermal excitation. This makes these methods particularly useful for liquid state or other thermally excited system simulations. Nevertheless, molecular dynamics and Monte Carlo simulations can also provide insights into cold gas-phase cluster formation [213], if a reliable force field is available [189]. 

It is clear from the above that the continuum model can simulate only those aspects of the solvent which are somewhat independent from hydrophobicity, hydrophyUicity, generally the first solvation shell, and specific interactions with the solute. The physical problem is a general one namely, it relates to the validity to use quantities, correctly described and defined at the macroscopic level, in the discrete description of matter at the atomic level. For such study, one needs explicit consideration of the solvent, for example the molecules of water. This can be done either at the quantum-mechanical level, as in cluster computations. Another approach is to simulate the system at the molecular dynamics (or Monte Carlo) level these techniques allow us to consider... 

Whereas selective diffusion can be better investigated using classical dynamic or Monte Carlo simulations, or experimental techniques, quantum chemical calculations are required to analyze molecular reactivity. Quantum chemical dynamic simulations provide with information with a too limited time scale range (of the order of several himdreds of ps) to be of use in diffusion studies which require time scale of the order of ns to s. However, they constitute good tools to study the behavior of reactants and products adsorbed in the proximity of the active site, prior to the reaction. Concerning reaction pathways analysis, static quantum chemistry calculations with molecular cluster models, allowing estimates of transition states geometries and properties, have been used for years. The application to solids is more recent. 

Another approach to investigate the hydrophobic effect is the ab initio quantum mechanical technique." " It is based on first principles (the Schrodinger equation), and this constitutes its main advantage compared to molecular dynamics and Monte Carlo approaches, which are based on classical potentials. At the present time, the ab initio quantum mechanical methods have limitations connected to the complexity and size of the molecular clusters considered." Nevertheless, these methods have been often used to accurately predict the structure and energy of a system of two molecules (dimers), " such as the system methane/water." " However, the structure and energy of a... 

Previous work on the thermodynamic properties of clusters used a number of schemes to evaluate the partition function required in Eq. (3.14). In the normal-mode method, " described in the Introduction, the partition function is constructed from the standard partitioning of a polyatomic gas into classical translational and rotational terms and quantum vibrational contributions. In Monte Carlo studies it is usual to employ a state integration technique.In the state integration method Eq. (3.5) is integrated with respect to temperature to obtain... 

Finally, we stress that the quantum chemical method presented here has the advantage over DFT-based techniques that it also furnishes wavefunctions that can be used to perform computations of spectra, and therefore have a better contact with the experiment. Another advantage of this approach is that, unlike the diffusion Monte-Carlo method, it can coherently be applied to studies of fermion and mixed boson/fermion doped clusters. An example can be found in our recent work on the Raman spectra of (He)w-Br2(X) clusters [27,28]. 

---