Performance assessment of exponential Rosenbrock method for large systems of ODE

This paper studies time integration methods for large stiff systems of ordinary differential equations (ODEs) of the form u'(t) = g(u(t)). For such problems, implicit methods generally outperform explicit methods, since the time step is usually less restricted by stability constraints. Recently, however, explicit so-called exponential integrators have become popular for stiff problems due to their favourable stability properties. These methods use matrix-vector products involving exponential-like functions of the Jacobian matrix, which can be approximated using Krylov subspace methods that require only matrix-vector products with the Jacobian. In this paper, we implement exponential integrators of second, third and fourth order and demonstrate that they are competitive with well-established approaches based on the backward differentiation formulas and a preconditioned Newton-Krylov solution strategy.

Efficient construction of nonorthogonal localized molecular orbitals in large systems.

Localized molecular orbitals (LMOs) are much more compact representations of electronic degrees of freedom than canonical molecular orbitals (CMOs). The most compact representation is provided by nonorthogonal localized molecular orbitals (NOLMOs), which are linearly independent but are not orthogonal. Both LMOs and NOLMOs are thus useful for linear-scaling calculations of electronic structures for large systems. Recently, NOLMOs have been successfully applied to linear-scaling calculations with density functional theory (DFT) and to reformulating time-dependent density functional theory (TDDFT) for calculations of excited states and spectroscopy. However, a challenge remains as NOLMO construction from CMOs is still inefficient for large systems. In this work, we develop an efficient method to accelerate the NOLMO construction by using predefined centroids of the NOLMO and thereby removing the nonlinear equality constraints in the original method ( J. Chem. Phys. 2004 , 120 , 9458 and J. Chem. Phys. 2000 , 112 , 4 ). Thus, NOLMO construction becomes an unconstrained optimization. Its efficiency is demonstrated for the selected saturated and conjugated molecules. Our method for fast NOLMO construction should lead to efficient DFT and NOLMO-TDDFT applications to large systems.

An Exploration of the challenges associated with software logging in large systems

Thesis (Master, Computing) -- Queen's University, 2016-05-29 18:11:34.114

Variational mean-field algorithm for efficient inference in large systems of stochastic differential equations

This work introduces a Gaussian variational mean-field approximation for inference in dynamical systems which can be modeled by ordinary stochastic differential equations. This new approach allows one to express the variational free energy as a functional of the marginal moments of the approximating Gaussian process. A restriction of the moment equations to piecewise polynomial functions, over time, dramatically reduces the complexity of approximate inference for stochastic differential equation models and makes it comparable to that of discrete time hidden Markov models. The algorithm is demonstrated on state and parameter estimation for nonlinear problems with up to 1000 dimensional state vectors and compares the results empirically with various well-known inference methodologies.

An Exploration of the challenges associated with software logging in large systems

Over the past few years, logging has evolved from from simple printf statements to more complex and widely used logging libraries. Today logging information is used to support various development activities such as fixing bugs, analyzing the results of load tests, monitoring performance and transferring knowledge. Recent research has examined how to improve logging practices by informing developers what to log and where to log. Furthermore, the strong dependence on logging has led to the development of logging libraries that have reduced the intricacies of logging, which has resulted in an abundance of log information. Two recent challenges have emerged as modern software systems start to treat logging as a core aspect of their software. In particular, 1) infrastructural challenges have emerged due to the plethora of logging libraries available today and 2) processing challenges have emerged due to the large number of log processing tools that ingest logs and produce useful information from them. In this thesis, we explore these two challenges. We first explore the infrastructural challenges that arise due to the plethora of logging libraries available today. As systems evolve, their logging infrastructure has to evolve (commonly this is done by migrating to new logging libraries). We explore logging library migrations within Apache Software Foundation (ASF) projects. We i find that close to 14% of the pro jects within the ASF migrate their logging libraries at least once. For processing challenges, we explore the different factors which can affect the likelihood of a logging statement changing in the future in four open source systems namely ActiveMQ, Camel, Cloudstack and Liferay. Such changes are likely to negatively impact the log processing tools that must be updated to accommodate such changes. We find that 20%-45% of the logging statements within the four systems are changed at least once. We construct random forest classifiers and Cox models to determine the likelihood of both just-introduced and long-lived logging statements changing in the future. We find that file ownership, developer experience, log density and SLOC are important factors in determining the stability of logging statements.

Decentralized dynamic compensators for large multivariable systems

In this paper, we consider the synthesis of decentralized dynamic compensators for large systems. The eliminant approach is used to obtain sufficient conditions for the existence of proper, stable, decentralized observer-controllers for stabilizing a large system. An illustrative example is given.

A Novel Monte-Carlo-Sampling-Based Receiver for Large-Scale Uplink Multiuser MIMO Systems

In this paper, we propose low-complexity algorithms based on Monte Carlo sampling for signal detection and channel estimation on the uplink in large-scale multiuser multiple-input-multiple-output (MIMO) systems with tens to hundreds of antennas at the base station (BS) and a similar number of uplink users. A BS receiver that employs a novel mixed sampling technique (which makes a probabilistic choice between Gibbs sampling and random uniform sampling in each coordinate update) for detection and a Gibbs-sampling-based method for channel estimation is proposed. The algorithm proposed for detection alleviates the stalling problem encountered at high signal-to-noise ratios (SNRs) in conventional Gibbs-sampling-based detection and achieves near-optimal performance in large systems with M-ary quadrature amplitude modulation (M-QAM). A novel ingredient in the detection algorithm that is responsible for achieving near-optimal performance at low complexity is the joint use of a mixed Gibbs sampling (MGS) strategy coupled with a multiple restart (MR) strategy with an efficient restart criterion. Near-optimal detection performance is demonstrated for a large number of BS antennas and users (e. g., 64 and 128 BS antennas and users). The proposed Gibbs-sampling-based channel estimation algorithm refines an initial estimate of the channel obtained during the pilot phase through iterations with the proposed MGS-based detection during the data phase. In time-division duplex systems where channel reciprocity holds, these channel estimates can be used for multiuser MIMO precoding on the downlink. The proposed receiver is shown to achieve good performance and scale well for large dimensions.

A novel MCMC algorithm for near-optimal detection in large-scale uplink mulituser MIMO systems

In this paper, we propose a low-complexity algorithm based on Markov chain Monte Carlo (MCMC) technique for signal detection on the uplink in large scale multiuser multiple input multiple output (MIMO) systems with tens to hundreds of antennas at the base station (BS) and similar number of uplink users. The algorithm employs a randomized sampling method (which makes a probabilistic choice between Gibbs sampling and random sampling in each iteration) for detection. The proposed algorithm alleviates the stalling problem encountered at high SNRs in conventional MCMC algorithm and achieves near-optimal performance in large systems with M-QAM. A novel ingredient in the algorithm that is responsible for achieving near-optimal performance at low complexities is the joint use of a randomized MCMC (R-MCMC) strategy coupled with a multiple restart strategy with an efficient restart criterion. Near-optimal detection performance is demonstrated for large number of BS antennas and users (e.g., 64, 128, 256 BS antennas/users).

Development and Application of Embedding Methods for the Simulation of Large Chemical Systems

The high computational cost of correlated wavefunction theory (WFT) calculations has motivated the development of numerous methods to partition the description of large chemical systems into smaller subsystem calculations. For example, WFT-in-DFT embedding methods facilitate the partitioning of a system into two subsystems: a subsystem A that is treated using an accurate WFT method, and a subsystem B that is treated using a more efficient Kohn-Sham density functional theory (KS-DFT) method. Representation of the interactions between subsystems is non-trivial, and often requires the use of approximate kinetic energy functionals or computationally challenging optimized effective potential calculations; however, it has recently been shown that these challenges can be eliminated through the use of a projection operator. This dissertation describes the development and application of embedding methods that enable accurate and efficient calculation of the properties of large chemical systems.

Chapter 1 introduces a method for efficiently performing projection-based WFT-in-DFT embedding calculations on large systems. This is accomplished by using a truncated basis set representation of the subsystem A wavefunction. We show that naive truncation of the basis set associated with subsystem A can lead to large numerical artifacts, and present an approach for systematically controlling these artifacts.

Chapter 2 describes the application of the projection-based embedding method to investigate the oxidative stability of lithium-ion batteries. We study the oxidation potentials of mixtures of ethylene carbonate (EC) and dimethyl carbonate (DMC) by using the projection-based embedding method to calculate the vertical ionization energy (IE) of individual molecules at the CCSD(T) level of theory, while explicitly accounting for the solvent using DFT. Interestingly, we reveal that large contributions to the solvation properties of DMC originate from quadrupolar interactions, resulting in a much larger solvent reorganization energy than that predicted using simple dielectric continuum models. Demonstration that the solvation properties of EC and DMC are governed by fundamentally different intermolecular interactions provides insight into key aspects of lithium-ion batteries, with relevance to electrolyte decomposition processes, solid-electrolyte interphase formation, and the local solvation environment of lithium cations.

Optimal distributed generation placement in radial distribution systems

Distributed generation (DG) resources are commonly used in the electric systems to obtain minimum line losses, as one of the benefits of DG, in radial distribution systems. Studies have shown the importance of appropriate selection of location and size of DGs. This paper proposes an analytical method for solving optimal distributed generation placement (ODGP) problem to minimize line losses in radial distribution systems using loss sensitivity factor (LSF) based on bus-injection to branch-current (BIBC) matrix. The proposed method is formulated and tested on 12 and 34 bus radial distribution systems. The classical grid search algorithm based on successive load flows is employed to validate the results. The main advantages of the proposed method as compared with the other conventional methods are the robustness and no need to calculate and invert large admittance or Jacobian matrices. Therefore, the simulation time and the amount of computer memory, required for processing data especially for the large systems, decreases.

Modeling of precipitation in reverse micellar systems

A model of the precipitation process in reverse micelles has been developed to calculate the size of fine particles obtained therein. While the method shares several features of particle nucleation and growth common to precipitation in large systems, complexities arise in describing the processes of nucleation, due to the extremely small size of a micelle and of particle growth caused by fusion among the micelles. Occupancy of micelles by solubilized molecules is governed by Poisson statistics, implying most of them are empty and cannot nucleate of its own. The model therefore specifies the minimum number of solubilized molecules required to form a nucleus which is used to calculate the homogeneous nucleation rate. Simultaneously, interaction between micelles is assumed to occur by Brownian collision and instantaneous fusion. Analysis of time scales of various events shows growth of particles to be very fast compared to other phenomena occurring. This implies that nonempty micelles either are supersaturated or contain a single precipitated particle and allows application of deterministic population balance equations to describe the evolution of the system with time. The model successfully predicts the experimental measurements of Kandori ct al.(3) on the size of precipitated CaCO3 particles, obtained by carbonation of reverse micelles containing aqueous Ca(OH)(2) solution.

Sensitivity of polarization fluctuations to the nature of protein-water interactions: Study of biological water in four different protein-water systems

Since the time of Kirkwood, observed deviations in magnitude of the dielectric constant of aqueous protein solution from that of neat water (similar to 80) and slower decay of polarization have been subjects of enormous interest, controversy, and debate. Most of the common proteins have large permanent dipole moments (often more than 100 D) that can influence structure and dynamics of even distant water molecules, thereby affecting collective polarization fluctuation of the solution, which in turn can significantly alter solution's dielectric constant. Therefore, distance dependence of polarization fluctuation can provide important insight into the nature of biological water. We explore these aspects by studying aqueous solutions of four different proteins of different characteristics and varying sizes, chicken villin headpiece subdomain (HP-36), immunoglobulin binding domain protein G (GB1), hen-egg white lysozyme (LYS), and Myoglobin (MYO). We simulate fairly large systems consisting of single protein molecule and 20000-30000 water molecules (varied according to the protein size), providing a concentration in the range of similar to 2-3 mM. We find that the calculated dielectric constant of the system shows a noticeable increment in all the cases compared to that of neat water. Total dipole moment auto time correlation function of water < dM(W) (0)delta M-W (t) > is found to be sensitive to the nature of the protein. Surprisingly, dipole moment of the protein and total dipole moment of the water molecules are found to be only weakly coupled. Shellwise decomposition of water molecules around protein reveals higher density of first layer compared to the succeeding ones. We also calculate heuristic effective dielectric constant of successive layers and find that the layer adjacent to protein has much lower value (similar to 50). However, progressive layers exhibit successive increment of dielectric constant, finally reaching a value close to that of bulk 4-5 layers away. We also calculate shellwise orientational correlation function and tetrahedral order parameter to understand the local dynamics and structural re-arrangement of water. Theoretical analysis providing simple method for calculation of shellwise local dielectric constant and implication of these findings are elaborately discussed in the present work. (C) 2014 AIP Publishing LLC.

Nonlinear oscillations in discrete and continuous systems

The first thesis topic is a perturbation method for resonantly coupled nonlinear oscillators. By successive near-identity transformations of the original equations, one obtains new equations with simple structure that describe the long time evolution of the motion. This technique is related to two-timing in that secular terms are suppressed in the transformation equations. The method has some important advantages. Appropriate time scalings are generated naturally by the method, and don't need to be guessed as in two-timing. Furthermore, by continuing the procedure to higher order, one extends (formally) the time scale of valid approximation. Examples illustrate these claims. Using this method, we investigate resonance in conservative, non-conservative and time dependent problems. Each example is chosen to highlight a certain aspect of the method.

The second thesis topic concerns the coupling of nonlinear chemical oscillators. The first problem is the propagation of chemical waves of an oscillating reaction in a diffusive medium. Using two-timing, we derive a nonlinear equation that determines how spatial variations in the phase of the oscillations evolves in time. This result is the key to understanding the propagation of chemical waves. In particular, we use it to account for certain experimental observations on the Belusov-Zhabotinskii reaction.

Next, we analyse the interaction between a pair of coupled chemical oscillators. This time, we derive an equation for the phase shift, which measures how much the oscillators are out of phase. This result is the key to understanding M. Marek's and I. Stuchl's results on coupled reactor systems. In particular, our model accounts for synchronization and its bifurcation into rhythm splitting.

Finally, we analyse large systems of coupled chemical oscillators. Using a continuum approximation, we demonstrate mechanisms that cause auto-synchronization in such systems.

Variational data assimilation for very large environmental problems

Variational data assimilation is commonly used in environmental forecasting to estimate the current state of the system from a model forecast and observational data. The assimilation problem can be written simply in the form of a nonlinear least squares optimization problem. However the practical solution of the problem in large systems requires many careful choices to be made in the implementation. In this article we present the theory of variational data assimilation and then discuss in detail how it is implemented in practice. Current solutions and open questions are discussed.