Ratio control variate method for efficiently determining high-dimensional model representations

The High-Dimensional Model Representation (HDMR) technique is a family of approaches to efficiently interpolate high-dimensional functions. RS(Random Sampling)-HDMR is a practical form of HDMR based on randomly sampling the overall function, and utilizing orthonormal polynomial expansions to approximate the RS-HDMR component functions. The determination of the expansion coefficients for the component functions employs Monte Carlo integration, which controls the accuracy of the RS-HDMR interpolation. The control variate method is an established approach to improve the accuracy of Monte Carlo integration. However, this method is often not practical for an arbitrary function f(x) because there is no general way to find the analytical control variate function h(x), which needs to be very similar to f(x). In this article, we show that truncated RS-HDMR expansions can be used as h(x) for arbitrary f(x), and a more stable iterative ratio control variate method was developed for the determination of the expansion coefficients for the RS-HDMR component functions. As the asymptotic error (standard deviation) of the estimator given by the ratio control variate method is proportional to 1/N(sample size), it is more efficient than the direct Monte Carlo integration, whose error is proportional to 1/square root(N). In an illustration of a four-dimensional atmospheric model a few hundred random samples are sufficient to construct an RS-HDMR expansion by the ratio control variate method with an accuracy comparable to that obtained by direct Monte Carlo integration with thousands of samples.     

A combined pressure-controlled scanning calorimetry and Monte Carlo determination of the Joule-Thomson inversion curve. Application to methane

A combination of a pressure-controlled scanning calorimetry (PCSC) and Monte Carlo simulations (MCS) is presented for an unequivocal determination of the Joule-Thomson inversion curve (JTIC) with high accuracy over wide ranges of pressure and temperature. The MCS performed with the fluctuation method are fast and easy to operate, but the results can vary significantly depend on the set of primary molecular data needed for the calculations. The PCSC is an experimental and more laborious technique, but supplies data of high quality. Thus, it can be used to check the MCS data and to verify the molecular parameters used for the calculations. Such a combined procedure was used in the present study for determination of the JTIC for methane, for which a correlation equation was established valid from 302.9 to 586.5 K. A combination of a direct experimental technique with molecular simulations permits also to better understand the complex behavior of the Joule-Thomson inversion phenomenon over wide ranges of pressure and temperature.     

High-dimensional model representations generated from low order terms--lp-RS-HDMR

High-dimensional model representation (HDMR) is a general set of quantitative model assessment and analysis tools for improving the efficiency of deducing high dimensional input-output system behavior. RS-HDMR is a particular form of HDMR based on random sampling (RS) of the input variables. The component functions in an HDMR expansion are optimal choices tailored to the n-variate function f(x) being represented over the desired domain of the n-dimensional vector x. The high-order terms (usually larger than second order, or equivalently beyond cooperativity between pairs of variables) in the expansion are often negligible. When it is necessary to go beyond the first and the second order RS-HDMR, this article introduces a modified low-order term product (lp)-RS-HDMR method to approximately represent the high-order RS-HDMR component functions as products of low-order functions. Using this method the high-order truncated RS-HDMR expansions may be constructed without directly computing the original high-order terms. The mathematical foundations of lp-RS-HDMR are presented along with an illustration of its utility in an atmospheric chemical kinetics model.     

Monte Carlo approach to the decay rate of a metastable system with an arbitrarily shaped barrier

A path integral Monte Carlo method based on the fast-Fourier transform technique combined with the important sampling method is proposed to calculate the decay rate of a metastable quantum system with an arbitrary shape of a potential barrier. The contribution of all fluctuation actions is included which can be used to check the accuracy of the usual steepest-descent approximation, namely, the perturbation expansion of potential. The analytical approximation is found to produce the decay rate of a particle in a cubic potential being about 20% larger than the Monte Carlo data at the crossover temperature. This disagreement increases with increasing complexity of the potential shape. We also demonstrate via Langevin simulation that the postsaddle potential influences strongly upon the classical escape rate.     

自优化剩余函数量子Monte Carlo方法

提出剩余函数量子Monte Carlo的一个新算法，这是一个自优化和自改善的过程.与以前的算法相比，本算法中的试探函数的优化是在剩余函数方法中同步进行的，而不是在变分Monte Carlo计算之前.为了优化试探函数，使用一种改进了的速降法，这是一个步长能够自动调节，超线性收敛的优化技术.在这个算法中,还使用了一种新的相关函数，它满足电子与电子以及电子与核奇点条件.此方法已被用于计算H2、LiH、Li2、H2O分子的基态以及CH2的X 3B1态、1 1A1态和2 1A1态的能量值.     

Measurement uncertainty of thermodynamic data

Thermodynamic quantities of chemical reactions are commonly derived from experimental data obtained by chemical analysis. The accuracy of the evaluated thermodynamic quantities is limited by the measurement uncertainty of the analytical techniques applied. Straightforward transfer of metrological rules established for determination of single analytes to the more complex process of evaluating values of thermodynamic quantities is not possible. Computer-intensive statistical methods and Monte Carlo techniques are shown to enable integration of existing metrological concepts. An initial stage of the integration of both concepts is presented, taking solubility data for Am(III) in carbonate media as an illustrative example. A cause and effect diagram is created as a means of identification of sources of uncertainty. The uncertainties are used in a resampling-based Monte Carlo study to produce a probability distribution of the value of a quantity.     

Thermodynamic characterization of fluids confined in heterogeneous pores by monte carlo simulations in the grand canonical and the isobaric-isothermal ensembles

Materials presenting nanoscale porosity are able to condense gases in their structure. This "capillary condensation" phenomenon has been studied for more than one century. Theoretical models help to understand experimental results but fail in explaining all experimental features. Most of the time, the difficulties in making quantitative or even qualitative predictions are due to the geometric complexity of the porous materials, such as large pore size distribution, chemical heterogeneities, or pore interconnections. Numerical calculations (lattice gas models or molecular simulations) are of considerable interest to calculate the adsorption properties of a fluid confined in a porous model with characteristic sizes up to several tens of nanometers. For instance, the grand canonical Monte Carlo method allows one to compute the average amount of fluid adsorbed in the porous model as a function of the temperature and the chemical potential of the fluid. However, the grand potential, necessary for a complete characterization of the system, is not a direct output of the algorithm. It is shown in this paper that the use of the isobaric-isothermal (NPT) ensemble allows one to circumvent this problem; that is, it is possible to get in one single Monte Carlo run the absolute grand potential for any given thermodynamic state of the fluid. A simplified thermodynamic integration scheme is then used to evaluate the grand potential over the whole isotherm branch passing through this initially given point. Since the usual NPT technique is a priori limited to homogeneous pores, it is proposed, for the first time, to generalize this procedure to a pore presenting a chemical heterogeneity along its axis. The new method gives the same results as the previous for homogeneous pores and allows new predictions for chemically heterogeneous pores. Comparison with the full integration scheme shows that the proposed direct calculation is faster since it avoids multiple Monte Carlo runs and more precise because it avoids the possible cumulative errors of the integration procedure.     

Forward--backward semiclassical dynamics with information-guided noise reduction for a molecule in solution

The forward--backward semiclassical dynamics (FBSD) methodology is used to obtain expressions for time correlation functions of a system (atom or molecule) in solution. We use information-guided noise reduction (IGNoR) [Makri, N. Chem. Phys. Lett. 2004, 400, 446] to minimize the statistical error associated with the Monte Carlo integration of oscillatory functions. This is possible by reformulating the correlation function in terms of an oscillatory solvent-dependent contribution whose integral can be obtained analytically and a slowly varying function obtained via a grid-based iterative evaluation of solute properties. Knowledge of the exact integral of the oscillatory function, combined with correlated statistics, leads to partial cancellation of the Monte Carlo error. Application on a one-dimensional solute-solvent model shows a substantial improvement of convergence in the IGNoR-enhanced FBSD correlation function for a fixed number of Monte Carlo samples. The reduction of statistical error achieved by using the IGNoR methodology becomes more significant as the number of solvent particles increases.     

Toward an accurate and efficient semiclassical surface hopping procedure for nonadiabatic problems

The derivation of a semiclassical surface hopping procedure from a formally exact solution of the Schrodinger equation is discussed. The fact that the derivation proceeds from an exact solution guarantees that all phase terms are completely and accurately included. Numerical evidence shows the method to be highly accurate. A Monte Carlo implementation of this method is considered, and recent work to significantly improve the statistical accuracy of the Monte Carlo approach is discussed.     

Numerical study of the accuracy and efficiency of various approaches for Monte Carlo surface hopping calculations

A one-dimensional, two-state model problem with two well-separated avoided crossing points is employed to test the efficiency and accuracy of a semiclassical surface hopping technique. The use of a one-dimensional model allows for the accurate numerical evaluation of both fully quantum-mechanical and semiclassical transition probabilities. The calculations demonstrate that the surface hopping procedure employed accounts for the interference between different hopping trajectories very well and provides highly accurate transition probabilities. It is, in general, not computationally feasible to completely sum over all hopping trajectories in the semiclassical calculations for multidimensional problems. In this case, a Monte Carlo procedure for selecting important trajectories can be employed. However, the cancellation due to the different phases associated with different trajectories limits the accuracy and efficiency of the Monte Carlo procedure. Various approaches for improving the accuracy and efficiency of Monte Carlo surface hopping procedures are investigated. These methods are found to significantly reduce the statistical sampling errors in the calculations, thereby increasing the accuracy of the transition probabilities obtained with a fixed number of trajectories sampled.     

Phase coexistence in heterogeneous porous media: a new extension to Gibbs ensemble Monte Carlo simulation method

The effect of confinement on phase behavior of simple fluids is still an area of intensive research. In between experiment and theory, molecular simulation is a powerful tool to study the effect of confinement in realistic porous materials, containing some disorder. Previous simulation works aiming at establishing the phase diagram of a confined Lennard-Jones-type fluid, concentrated on simple pore geometries (slits or cylinders). The development of the Gibbs ensemble Monte Carlo technique by Panagiotopoulos [Mol. Phys. 61, 813 (1987)], greatly favored the study of such simple geometries for two reasons. First, the technique is very efficient to calculate the phase diagram, since each run (at a given temperature) converges directly to an equilibrium between a gaslike and a liquidlike phase. Second, due to volume exchange procedure between the two phases, at least one invariant direction of space is required for applicability of this method, which is the case for slits or cylinders. Generally, the introduction of some disorder in such simple pores breaks the initial invariance in one of the space directions and prevents to work in the Gibbs ensemble. The simulation techniques for such disordered systems are numerous (grand canonical Monte Carlo, molecular dynamics, histogram reweighting, N-P-T+test method, Gibbs-Duhem integration procedure, etc.). However, the Gibbs ensemble technique, which gives directly the coexistence between phases, was never generalized to such systems. In this work, we focus on two weakly disordered pores for which a modified Gibbs ensemble Monte Carlo technique can be applied. One of the pores is geometrically undulated, whereas the second is cylindrical but presents a chemical variation which gives rise to a modulation of the wall potential. In the first case almost no change in the phase diagram is observed, whereas in the second strong modifications are reported.     

Theory and algorithms for mixed Monte Carlo-stochastic dynamics simulations

The recently introduced mixed MC-SD method is a fundamentally new procedure which essentially eliminates the distinction between Monte Carlo and dynamics. Unlike other methods which utilize forces, Brownian motion or dynamical steps to generate new trial configurations in a Monte Carlo search, mixed MC-SD does stochastic dynamics on the cartesian space of a molecule and Monte Carlo on the torsion space of the molecule simultaneously. After each dynamical step, a random deformation of a rotatable torsion is performed and accepted or rejected according to the Metropolis criteria. The next dynamical step is performed from the most recent configuration and the velocities from the previous dynamical step. The smooth merging of Monte Carlo and dynamics requires the use of the stochastic velocity Verlet integration scheme. Here, the velocity Verlet stochastic dynamics method is derived, and the reasons why it can be joined with Metropolis Monte Carlo in a continuous fashion are explored.     

Density-functional expansion methods: generalization of the auxiliary basis

The formulation of density-functional expansion methods is extended to treat the second and higher-order terms involving the response density and spin densities with an arbitrary single-center auxiliary basis. The two-center atomic orbital products are represented by the auxiliary functions centered about those two atoms, and the mapping coefficients are determined from a local constrained variational procedure. This two-center variational procedure allows the mapping coefficients to be pretabulated and splined as a function of internuclear separation for efficient look up. The splines of mapping coefficients have a range no longer than that of the overlap integrals, and the auxiliary density appears as a single point-multipole expansion to all nonoverlapping atoms, thus allowing for the trivial implementation of a linear-scaling algorithm. The method is tested using Gaussian multipole expansions, and the effect of angular and radial completeness is explored. Several auxiliary basis sets are parametrized and compared to an auxiliary basis analogous to that used in the self-consistent-charge density-functional tight-binding model, and the method is demonstrated to greatly improve the representation of the density response with respect to a reference expansion model that does not use an auxiliary basis.     

Recognition of amino acids in solution

The basic solvation shells of all the amino acids, of use in the study of their recognition in aqueous solutions, are determined by means of a method based on the use of 1/R expansions parameterized on the basis of the results from accurate SCF calculations. The accuracy of the calculations is tested in a more extensive study of the solvation of glycine, for which the results of Monte Carlo calculations are reproduced.     

Super-Monte Carlo: A photon/electron dose calculation algorithm for radiotherapy

Super-Monte Carlo (SMC) is a method of dose calculation for radiotherapy which combines both analytical calculations and Monte Carlo electron transport. Analytical calculations are used where possible, such as the determination of photon interaction density, to decrease computation time. A Monte Carlo method is used for the electron transport in order to obtain high accuracy of results. To further speed computation, Monte Carlo is used once only, to form an electron track kernel (etk). The etk is a dataset containing the lengths and energy deposition of each step of a number of electron tracks. The etk is transported from each incident particle interaction site, from which the dose is calculated. Dose distributions calculated in heterogeneous media show SMC results similar to those of Monte Carlo. For the same statistical uncertainty, SMC takes an order of magnitude less computation time than a full Monte Carlo simulation. SMC has only been implemented for photons and electrons, however the same basic method could be used for the transport of other particles. Current development includes the optimisation of the etks and the code in order to decrease computation time, and also the inclusion of SMC onto a clinical planning system.     

High Dimensional Model Representations Generated from Low Dimensional Data Samples. I. mp-Cut-HDMR

High dimensional model representation (HDMR) is a general set of quantitative model assessment and analysis tools for improving the efficiency of deducing high dimensional input–output system behavior. For a high dimensional system, an output f(x) is commonly a function of many input variables x=|x ₁,x ₂,...,x _n } with n10² or larger. HDMR describes f(x) by a finite hierarchical correlated function expansion in terms of the input variables. Various forms of HDMR can be constructed for different purposes. Cut- and RS-HDMR are two particular HDMR expansions. Since the correlated functions in an HDMR expansion are optimal choices tailored to f(x) over the entire domain of x, the high order terms (usually larger than second order, or beyond pair cooperativity) in the expansion are often negligible. When the approximations given by the first and the second order Cut-HDMR correlated functions are not adequate, this paper presents a monomial based preconditioned HDMR method to represent the higher order terms of a Cut-HDMR expansion by expressions similar to the lower order ones with monomial multipliers. The accuracy of the Cut-HDMR expansion can be significantly improved using preconditioning with a minimal number of additional input–output samples without directly invoking the determination of higher order terms. The mathematical foundations of monomial based preconditioned Cut-HDMR is presented along with an illustration of its applicability to an atmospheric chemical kinetics model.     

Optimized Jastrow-Slater wave functions for ground and excited states: application to the lowest states of ethene

A quantum Monte Carlo method is presented for determining multideterminantal Jastrow-Slater wave functions for which the energy is stationary with respect to the simultaneous optimization of orbitals and configuration interaction coefficients. The approach is within the framework of the so-called energy fluctuation potential method which minimizes the energy in an iterative fashion based on Monte Carlo sampling and a fitting of the local energy fluctuations. The optimization of the orbitals is combined with the optimization of the configuration interaction coefficients through the use of additional single excitations to a set of external orbitals. A new set of orbitals is then obtained from the natural orbitals of this enlarged configuration interaction expansion. For excited states, the approach is extended to treat the average of several states within the same irreducible representation of the pointgroup of the molecule. The relationship of our optimization method with the stochastic reconfiguration technique by Sorella et al. is examined. Finally, the performance of our approach is illustrated with the lowest states of ethene, in particular with the difficult case of the 1(1)B(1u) state.     

Monte Carlo computation of ground‐state energy derivatives

Monte Carlo is a simple technique, which uses random numbers to compute ground‐state energies of small molecules (and quantum systems in general). The results always have a small statistical error, which poses a major obstacle when estimating properties defined as ground‐state‐energy derivatives (such as the molecule's geometry, its vibrational frequencies, polarizabilities, etc.). In this article, we present and demonstrate an approach that makes an accurate Monte–Carlo estimation of such derivatives possible. This is achieved by realizing that the simulation constitutes an autocorrelated stochastic process, whose proper analysis then enables us to estimate various energy derivatives as a combination of total correlation between readily computable quantities. The resulting procedure is a natural extension of the usual Monte Carlo algorithm for computing the ground‐state energy, with relatively small computational overhead. © 2007 Wiley Periodicals, Inc. Int J Quantum Chem, 2008     

Development of a method for the direct analysis of peptide AM336 in monkey cerebrospinal fluid using liquid chromatography/electrospray ionization mass spectrometry with a mixed-function column

A liquid chromatography/mass spectrometry (LC/MS) analytical procedure, using a single column for sample clean-up, enrichment and separation, has been developed for the determination of the peptide AM336 in monkey cerebrospinal fluid (CSF). CSF samples were injected and analyzed using a polymer-coated mixed-function high-performance liquid chromatography (HPLC) column with gradient elution and application of a timed valve-switching event. The mass spectrometer was operated in the positive electrospray ionization (ESI(+)) mode with single ion recording (SIR) at m/z 920. The method was validated, yielding calibration curves with correlation coefficients greater than 0.9892. Assay precision and accuracy were evaluated by direct injection of AM336-fortified CSF samples at three concentration levels. Analyzed concentrations ranged from 99.93 to 113.1% of their respective theoretical concentrations with coefficients of variation below 9.0%. An evaluation of the signal-to-noise (S/N) ratio for a 200 ng/mL calibration standard, considered to be the lower limit of quantitation (LLOQ), resulted in an estimated limit of detection (LOD) of 31.2 ng/mL. Preliminary data suggest the possibility of using this method to analyze AM336 also in plasma samples, pending the successful outcome of additional investigations.