Showing papers on "Numerical analysis published in 2011"

Anderson Acceleration for Fixed-Point Iterations

Abstract: This paper concerns an acceleration method for fixed-point iterations that originated in work of D. G. Anderson [J. Assoc. Comput. Mach., 12 (1965), pp. 547-560], which we accordingly call Anderson acceleration here. This method has enjoyed considerable success and wide usage in electronic structure computations, where it is known as Anderson mixing; however, it seems to have been untried or underexploited in many other important applications. Moreover, while other acceleration methods have been extensively studied by the mathematics and numerical analysis communities, this method has received relatively little attention from these communities over the years. A recent paper by H. Fang and Y. Saad [Numer. Linear Algebra Appl., 16 (2009), pp. 197-221] has clarified a remarkable relationship of Anderson acceleration to quasi-Newton (secant updating) methods and extended it to define a broader Anderson family of acceleration methods. In this paper, our goals are to shed additional light on Anderson acceleration and to draw further attention to its usefulness as a general tool. We first show that, on linear problems, Anderson acceleration without truncation is “essentially equivalent” in a certain sense to the generalized minimal residual (GMRES) method. We also show that the Type 1 variant in the Fang-Saad Anderson family is similarly essentially equivalent to the Arnoldi (full orthogonalization) method. We then discuss practical considerations for implementing Anderson acceleration and illustrate its performance through numerical experiments involving a variety of applications.

Numerical Methods for High-Speed Flows

Abstract: We review numerical methods for direct numerical simulation (DNS) and large-eddy simulation (LES) of turbulent compressible flow in the presence of shock waves. Ideal numerical methods should be accurate and free from numerical dissipation in smooth parts of the flow, and at the same time they must robustly capture shock waves without significant Gibbs ringing, which may lead to nonlinear instability. Adapting to these conflicting goals leads to the design of strongly nonlinear numerical schemes that depend on the geometrical properties of the solution. For low-dissipation methods for smooth flows, numerical stability can be based on physical conservation principles for kinetic energy and/or entropy. Shock-capturing requires the addition of artificial dissipation, in more or less explicit form, as a surrogate for physical viscosity, to obtain nonoscillatory transitions. Methods suitable for both smooth and shocked flows are discussed, and the potential for hybridization is highlighted. Examples of the application of advanced algorithms to DNS/LES of turbulent, compressible flows are presented.

The Probability That a Numerical, Analysis Problem Is Difficult

Abstract: Numerous problems in numerical analysis, including matrix inversion, eigenvalue calculations and polynomial zerofinding, share the following property: The difficulty of solving a given problem is large when the distance from that problem to the nearest "ill-posed" one is small. For example, the closer a matrix is to the set of noninvertible matrices, the larger its condition number with respect to inversion. We show that the sets of ill-posed problems for matrix inversion, eigenproblems, and polynomial zerofinding all have a common algebraic and geometric structure which lets us compute the probability distribution of the distance from a "random" problem to the set. From this probability distribution we derive, for example, the distribution of the condition number of a random matrix. We examine the relevance of this theory to the analysis and construction of numerical algorithms destined to be run in finite precision arithmetic.

On preconditioned MHSS iteration methods for complex symmetric linear systems

Abstract: We propose a preconditioned variant of the modified HSS (MHSS) iteration method for solving a class of complex symmetric systems of linear equations. Under suitable conditions, we prove the convergence of the preconditioned MHSS (PMHSS) iteration method and discuss the spectral properties of the PMHSS-preconditioned matrix. Numerical implementations show that the resulting PMHSS preconditioner leads to fast convergence when it is used to precondition Krylov subspace iteration methods such as GMRES and its restarted variants. In particular, both the stationary PMHSS iteration and PMHSS-preconditioned GMRES show meshsize-independent and parameter-insensitive convergence behavior for the tested numerical examples.

Algorithms for finding hidden oscillations in nonlinear systems. The Aizerman and Kalman conjectures and Chua's circuits

Abstract: An algorithm for searching hidden oscillations in dynamic systems is developed to help solve the Aizerman's, Kalman's and Markus-Yamabe's conjectures well-known in control theory. The first step of the algorithm consists in applying modified harmonic linearization methods. A strict mathematical substantiation of these methods is given using special Poincare maps. Subsequent steps of the proposed algorithms rely on the modern applied theory of bifurcations and numerical methods of solving differential equations. These algorithms help find and localize hidden strange attractors (i.e., such that a basin of attraction of which does not contain neighborhoods of equilibria), as well as hidden periodic oscillations. One of these algorithms is used here to discover, for the first time, a hidden strange attractor in the dynamic system describing a nonlinear Chua's circuit, viz. an electronic circuit with nonlinear feedback.

Numerical Methods for Two-phase Incompressible Flows

Abstract: Introduction.- Part I One-phase incompressible flows.- Mathematical models.- Finite element discretization.- Time integration.-

Exploring Monte Carlo Methods

Abstract: Exploring Monte Carlo Methods is a basic text that describes the numerical methods that have come to be known as "Monte Carlo." The book treats the subject generically through the first eight chapters and, thus, should be of use to anyone who wants to learn to use Monte Carlo. The next two chapters focus on applications in nuclear engineering, which are illustrative of uses in other fields. Five appendices are included, which provide useful information on probability distributions, general-purpose Monte Carlo codes for radiation transport, and other matters. The famous "Buffon’s needle problem" provides a unifying theme as it is repeatedly used to illustrate many features of Monte Carlo methods. This book provides the basic detail necessary to learn how to apply Monte Carlo methods and thus should be useful as a text book for undergraduate or graduate courses in numerical methods. It is written so that interested readers with only an understanding of calculus and differential equations can learn Monte Carlo on their own. Coverage of topics such as variance reduction, pseudo-random number generation, Markov chain Monte Carlo, inverse Monte Carlo, and linear operator equations will make the book useful even to experienced Monte Carlo practitioners. Provides a concise treatment of generic Monte Carlo methods Proofs for each chapter Appendixes include Certain mathematical functions; Bose Einstein functions, Fermi Dirac functions, Watson functions

A probabilistic numerical method for fully nonlinear parabolic PDEs

Abstract: We consider the probabilistic numerical scheme for fully nonlinear PDEs suggested in \cite{cstv}, and show that it can be introduced naturally as a combination of Monte Carlo and finite differences scheme without appealing to the theory of backward stochastic differential equations. Our first main result provides the convergence of the discrete-time approximation and derives a bound on the discretization error in terms of the time step. An explicit implementable scheme requires to approximate the conditional expectation operators involved in the discretization. This induces a further Monte Carlo error. Our second main result is to prove the convergence of the latter approximation scheme, and to derive an upper bound on the approximation error. Numerical experiments are performed for the approximation of the solution of the mean curvature flow equation in dimensions two and three, and for two and five-dimensional (plus time) fully-nonlinear Hamilton-Jacobi-Bellman equations a! rising in the theory of portfolio optimization in financial mathematics.

Analysis of HDG Methods for Stokes Flow

Abstract: In this paper, we analyze a hybridizable discontinuous Galerkin method for numerically solving the Stokes equations. The method uses polynomials of degree k for all the components of the approximate solution of the gradient-velocity-pressure formulation. The novelty of the analysis is the use of a new projection tailored to the very structure of the numerical traces of the method. It renders the analysis of the projection of the errors very concise and allows us to see that the projection of the error in the velocity superconverges. As a consequence, we prove that the approximations of the velocity gradient, the velocity and the pressure converge with the optimal order of convergence of k+1 in L 2 for any k > 0. Moreover, taking advantage of the superconvergence properties of the velocity, we introduce a new element-by-element postprocessing to obtain a new velocity approximation which is exactly divergence-free, H(div)-conforming, and converges with order k+2 for k > 1 and with order 1 for k = 0. Numerical experiments are presented which validate the theoretical results.

Combination synchronization of three classic chaotic systems using active backstepping design.

Abstract: In this paper, an active backstepping design is proposed to achieve combination synchronization between three different chaotic systems: Lorenz system, Chen's system, and Lu system. The proposed method is a systematic design approach and consists in a recursive procedure that interlaces the choice of a Lyapunov function with the design of active control. Numerical simulations are shown to verify the feasibility and effectiveness of the proposed control technique.

Mathematical and computational methods for semiclassical Schrödinger equations

Abstract: We consider time-dependent (linear and nonlinear) Schrodinger equations in a semiclassical scaling. These equations form a canonical class of (nonlinear) dispersive models whose solutions exhibit high-frequency oscillations. The design of efficient numerical methods which produce an accurate approxima- tion of the solutions, or at least of the associated physical observables, is a formidable mathematical challenge. In this article we shall review the basic analytical methods for dealing with such equations, including WKB asymp- totics, Wigner measure techniques and Gaussian beams. Moreover, we shall give an overview of the current state of the art of numerical methods (most of which are based on the described analytical techniques) for the Schrodinger equation in the semiclassical regime.

The secret to successful solute-transport modeling.

Abstract: Modeling subsurface solute transport is difficult-more so than modeling heads and flows. The classical governing equation does not always adequately represent what we see at the field scale. In such cases, commonly used numerical models are solving the wrong equation. Also, the transport equation is hyperbolic where advection is dominant, and parabolic where hydrodynamic dispersion is dominant. No single numerical method works well for all conditions, and for any given complex field problem, where seepage velocity is highly variable, no one method will be optimal everywhere. Although we normally expect a numerically accurate solution to the governing groundwater-flow equation, errors in concentrations from numerical dispersion and/or oscillations may be large in some cases. The accuracy and efficiency of the numerical solution to the solute-transport equation are more sensitive to the numerical method chosen than for typical groundwater-flow problems. However, numerical errors can be kept within acceptable limits if sufficient computational effort is expended. But impractically long simulation times may promote a tendency to ignore or accept numerical errors. One approach to effective solute-transport modeling is to keep the model relatively simple and use it to test and improve conceptual understanding of the system and the problem at hand. It should not be expected that all concentrations observed in the field can be reproduced. Given a knowledgeable analyst, a reasonable description of a hydrogeologic framework, and the availability of solute-concentration data, the secret to successful solute-transport modeling may simply be to lower expectations.

A numerical technique for solving fractional optimal control problems

Abstract: This paper presents a numerical method for solving a class of fractional optimal control problems (FOCPs). The fractional derivative in these problems is in the Caputo sense. The method is based upon the Legendre orthonormal polynomial basis. The operational matrices of fractional Riemann-Liouville integration and multiplication, along with the Lagrange multiplier method for the constrained extremum are considered. By this method, the given optimization problem reduces to the problem of solving a system of algebraic equations. By solving this system, we achieve the solution of the FOCP. Illustrative examples are included to demonstrate the validity and applicability of the new technique.

Comparing Numerical Methods for Isothermal Magnetized Supersonic Turbulence

Abstract: Many astrophysical applications involve magnetized turbulent flows with shock waves. Ab initio star formation simulations require a robust representation of supersonic turbulence in molecular clouds on a wide range of scales imposing stringent demands on the quality of numerical algorithms. We employ simulations of supersonic super-Alfvenic turbulence decay as a benchmark test problem to assess and compare the performance of nine popular astrophysical MHD methods actively used to model star formation. The set of nine codes includes: ENZO, FLASH, KT-MHD, LL-MHD, PLUTO, PPML, RAMSES, STAGGER, and ZEUS. These applications employ a variety of numerical approaches, including both split and unsplit, finite difference and finite volume, divergence preserving and divergence cleaning, a variety of Riemann solvers, and a range of spatial reconstruction and time integration techniques. We present a comprehensive set of statistical measures designed to quantify the effects of numerical dissipation in these MHD solvers. We compare power spectra for basic fields to determine the effective spectral bandwidth of the methods and rank them based on their relative effective Reynolds numbers. We also compare numerical dissipation for solenoidal and dilatational velocity components to check for possible impacts of the numerics on small-scale density statistics. Finally, we discuss the convergence of various characteristics for the turbulence decay test and the impact of various components of numerical schemes on the accuracy of solutions. The nine codes gave qualitatively the same results, implying that they are all performing reasonably well and are useful for scientific applications. We show that the best performing codes employ a consistently high order of accuracy for spatial reconstruction of the evolved fields, transverse gradient interpolation, conservation law update step, and Lorentz force computation. The best results are achieved with divergence-free evolution of the magnetic field using the constrained transport method and using little to no explicit artificial viscosity. Codes that fall short in one or more of these areas are still useful, but they must compensate for higher numerical dissipation with higher numerical resolution. This paper is the largest, most comprehensive MHD code comparison on an application-like test problem to date. We hope this work will help developers improve their numerical algorithms while helping users to make informed choices about choosing optimal applications for their specific astrophysical problems.

Construction of an optimized explicit Runge-Kutta-Nyström method for the numerical solution of oscillatory initial value problems

Abstract: An explicit optimized Runge-Kutta-Nystrom method with four stages and fifth algebraic order is developed The produced method has variable coefficients with zero phase-lag, zero amplification factor and zero first derivative of the amplification factor We provide an analysis of the local truncation error of the new method We also measure the efficiency of the new method in comparison to other numerical methods through the integration of the two-body problem with various eccentricities and three other periodical/oscillatory initial value problems

Model-based optoacoustic inversion with arbitrary-shape detectors.

Abstract: Purpose: Optoacousticimaging enables mapping the optical absorption of biological tissue using optical excitation and acoustic detection. Although most image-reconstruction algorithms are based on the assumption of a detector with an isotropic sensitivity, the geometry of the detector often leads to a response with spatially dependent magnitude and bandwidth. This effect may lead to attenuation or distortion in the recorded signal and, consequently, in the reconstructed image. Methods: Herein, an accurate numerical method for simulating the spatially dependent response of an arbitrary-shape acoustic transducer is presented. The method is based on an analytical solution obtained for a two-dimensional line detector. The calculated response is incorporated in the forward model matrix of an optoacousticimaging setup using temporal convolution, and image reconstruction is performed by inverting the matrix relation. Results: The method was numerically and experimentally demonstrated in two dimensions for both flat and focused transducers and compared to the spatial-convolution method. In forward simulations, the developed method did not suffer from the numerical errors exhibited by the spatial-convolution method. In reconstruction simulations and experiments, the use of both temporal-convolution and spatial-convolution methods lead to an enhancement in resolution compared to a reconstruction with a point detectormodel. However, because of its higher modeling accuracy, the temporal-convolution method achieved a noise figure approximated three times lower than the spatial-convolution method. Conclusions: The demonstrated performance of the spatial-convolution method shows it is a powerful tool for reducing reconstruction artifacts originating from the detector finite size and improving the quality of optoacousticreconstructions. Furthermore, the method may be used for assessing new system designs. Specifically, detectors with nonstandard shapes may be investigated.

A robust preconditioning technique for the extended finite element method

Abstract: The extended finite element method enhances the approximation properties of the finite element space by using additional enrichment functions. But the resulting stiffness matrices can become ill-conditioned. In that case iterative solvers need a large number of iterations to obtain an acceptable solution. In this paper a procedure is described to obtain stiffness matrices whose condition number is close to the one of the finite element matrices without any enrichments. A domain decomposition is employed and the algorithm is very well suited for parallel computations. The method was tested in numerical experiments to show its effectiveness. The experiments have been conducted for structures containing cracks and material interfaces. We show that the corresponding enrichments can result in arbitrarily ill-conditioned matrices. The method proposed here, however, provides well-conditioned matrices and can be applied to any sort of enrichment. The complexity of this approach and its relation to the domain decomposition is discussed. Computation times have been measured for a structure containing multiple cracks. For this structure the computation times could be decreased by a factor of 2.

The surface finite element method for pattern formation on evolving biological surfaces.

Abstract: In this article we propose models and a numerical method for pattern formation on evolving curved surfaces. We formulate reaction-diffusion equations on evolving surfaces using the material transport formula, surface gradients and diffusive conservation laws. The evolution of the surface is defined by a material surface velocity. The numerical method is based on the evolving surface finite element method. The key idea is based on the approximation of Γ by a triangulated surface Γh consisting of a union of triangles with vertices on Γ. A finite element space of functions is then defined by taking the continuous functions on Γh which are linear affine on each simplex of the polygonal surface. To demonstrate the capability, flexibility, versatility and generality of our methodology we present results for uniform isotropic growth as well as anisotropic growth of the evolution surfaces and growth coupled to the solution of the reaction-diffusion system. The surface finite element method provides a robust numerical method for solving partial differential systems on continuously evolving domains and surfaces with numerous applications in developmental biology, tumour growth and cell movement and deformation.

Computational modeling of passive myocardium

Abstract: This work deals with the computational modeling of passive myocardial tissue within the framework of mixed, non-linear finite element methods. We consider a recently proposed, convex, anisotropic hyperelastic model that accounts for the locally orthotropic micro-structure of cardiac muscle. A coordinate-free representation of anisotropy is incorporated through physically relevant invariants of the Cauchy–Green deformation tensors and structural tensors of the corresponding material symmetry group. This model, which has originally been designed for exactly incompressible deformations, is extended towards entirely three-dimensional inhomogeneous deformations by additively decoupling the strain energy function into volumetric and isochoric parts along with the multiplicative split of the deformation gradient. This decoupled constitutive structure is then embedded in a mixed finite element formulation through a three-field Hu–Washizu functional whose simultaneous variation with respect to the independent pressure, dilatation, and placement fields results in the associated Euler–Lagrange equations, thereby minimizing the potential energy. This weak form is then consistently linearized for uniform-pressure elements within the framework of an implicit finite element method. To demonstrate the performance of the proposed approach, we present a three-dimensional finite element analysis of a generic biventricular heart model, subjected to physiological ventricular pressure. The parameters employed in the numerical analysis are identified by solving an optimization problem based on six simple shear experiments on explanted cardiac tissue. Copyright © 2010 John Wiley & Sons, Ltd.

Optimal pulse design in quantum control: A unified computational method

Abstract: Many key aspects of control of quantum systems involve manipulating a large quantum ensemble exhibiting variation in the value of parameters characterizing the system dynamics. Developing electromagnetic pulses to produce a desired evolution in the presence of such variation is a fundamental and challenging problem in this research area. We present such robust pulse designs as an optimal control problem of a continuum of bilinear systems with a common control function. We map this control problem of infinite dimension to a problem of polynomial approximation employing tools from geometric control theory. We then adopt this new notion and develop a unified computational method for optimal pulse design using ideas from pseudospectral approximations, by which a continuous-time optimal control problem of pulse design can be discretized to a constrained optimization problem with spectral accuracy. Furthermore, this is a highly flexible and efficient numerical method that requires low order of discretization and yields inherently smooth solutions. We demonstrate this method by designing effective broadband π/2 and π pulses with reduced rf energy and pulse duration, which show significant sensitivity enhancement at the edge of the spectrum over conventional pulses in 1D and 2D NMR spectroscopy experiments.

Study of a third grade non-Newtonian fluid flow between two parallel plates using the multi-step differential transform method

Abstract: In this paper, the multi-step differential transform method (MDTM), one of the most effective method, is implemented to compute an approximate solution of the system of nonlinear differential equations governing the problem. It has been attempted to show the reliability and performance of the MDTM in comparison with the numerical method (fourth-order Runge-Kutta) and other analytical methods such as HPM, HAM and DTM in solving this problem. The first differential equation is the plane Couette flow equation which serves as a useful model for many interesting problems in engineering. The second one is the Fully-developed plane Poiseuille flow equation and finally the third one is the plane Couette-Poiseuille flow.