Authors:Quan Zheng; Xin Zhao; Yufeng Liu Pages: 1 - 16 Abstract: Publication date: January 2017 Source:Applied Numerical Mathematics, Volume 111 Author(s): Quan Zheng, Xin Zhao, Yufeng Liu This paper studies a finite difference method for one-dimensional nonhomogeneous Burgers' equation on the infinite domain. Two exact nonlinear artificial boundary conditions are applied on two artificial boundaries to limit the original problem onto a bounded computational domain. A function transformation makes both Burgers' equation and artificial boundary conditions linear. Consequently, a novel finite difference scheme is developed by using the method of reduction of order for the obtained equation and artificial boundary conditions. The stability and the convergence with order 3/2 in time and 2 in space in an energy norm are proved for this method for Burgers' equation. Different examples illustrate the unconditional stability and the accuracy of the proposed method.

Authors:Yubo Yang; Peng Zhu Pages: 36 - 48 Abstract: Publication date: January 2017 Source:Applied Numerical Mathematics, Volume 111 Author(s): Yubo Yang, Peng Zhu In this paper, we introduce discontinuous Galerkin methods with interior penalties, both the NIPG and SIPG method for solving 2D singularly perturbed convection–diffusion problems. On the modified graded meshes with the standard Lagrange Q k -elements ( k = 1 , 2 ), we show optimal order error estimates in the ε-weighted energy norm uniformly, up to a logarithmic factor, in the singular perturbation parameter ε. We prove that the convergence rate in the ε-weighted energy norm is O ( log k + 1 ( 1 ε ) N k ) , where the total number of the mesh points is O ( N 2 ) . For k ≥ 3 , our methods can be extended directly, provided the higher order regularities of the solution u are derived. Finally, numerical experiments support our theoretical results.

Authors:Jason Albright; Yekaterina Epshteyn; Michael Medvinsky; Qing Xia Pages: 64 - 91 Abstract: Publication date: January 2017 Source:Applied Numerical Mathematics, Volume 111 Author(s): Jason Albright, Yekaterina Epshteyn, Michael Medvinsky, Qing Xia Numerical approximations and computational modeling of problems from Biology and Materials Science often deal with partial differential equations with varying coefficients and domains with irregular geometry. The challenge here is to design an efficient and accurate numerical method that can resolve properties of solutions in different domains/subdomains, while handling the arbitrary geometries of the domains. In this work, we consider 2D elliptic models with material interfaces and develop efficient high-order accurate methods based on Difference Potentials for such problems.

Authors:Ji-Feng Bao; Chong Li; Wei-Ping Shen; Jen-Chih Yao; Sy-Ming Guu Pages: 92 - 110 Abstract: Publication date: January 2017 Source:Applied Numerical Mathematics, Volume 111 Author(s): Ji-Feng Bao, Chong Li, Wei-Ping Shen, Jen-Chih Yao, Sy-Ming Guu We propose several approximate Gauss–Newton methods, i.e., the truncated, perturbed, and truncated-perturbed GN methods, for solving underdetermined nonlinear least squares problems. Under the assumption that the Fréchet derivatives are Lipschitz continuous and of full row rank, Kantorovich-type convergence criteria of the truncated GN method are established and local convergence theorems are presented with the radii of convergence balls obtained. As consequences of the convergence results for the truncated GN method, convergence theorems of the perturbed and truncated-perturbed GN methods are also presented. Finally, numerical experiments are presented where the comparisons with the standard inexact Gauss–Newton method and the inexact trust-region method for bound-constrained least squares problems [23] are made.

Authors:Zheng Ma; Yong Zhang; Zhennan Zhou Pages: 144 - 159 Abstract: Publication date: January 2017 Source:Applied Numerical Mathematics, Volume 111 Author(s): Zheng Ma, Yong Zhang, Zhennan Zhou In this paper, we propose a new time splitting Fourier spectral method for the semi-classical Schrödinger equation with vector potentials. Compared with the results in [21], our method achieves spectral accuracy in space by interpolating the Fourier series via the NonUniform Fast Fourier Transform (NUFFT) algorithm in the convection step. The NUFFT algorithm helps maintain high spatial accuracy of Fourier method, and at the same time improve the efficiency from O ( N 2 ) (of direct computation) to O ( N log N ) operations, where N is the total number of grid points. The kinetic step and potential step are solved by analytical solution with pseudo-spectral approximation, and, therefore, we obtain spectral accuracy in space for the whole method. We prove that the method is unconditionally stable, and we show improved error estimates for both the wave function and physical observables, which agree with the results in [3] for vanishing potential cases and are superior to those in [21]. Extensive one and two dimensional numerical studies are presented to verify the properties of the proposed method, and simulations of 3D problems are demonstrated to show its potential for future practical applications.

Authors:Liangliang Sun; Ting Wei Pages: 160 - 180 Abstract: Publication date: January 2017 Source:Applied Numerical Mathematics, Volume 111 Author(s): Liangliang Sun, Ting Wei This paper is devoted to identify the zeroth-order coefficient in a time-fractional diffusion equation from two boundary measurement data in one-dimensional case. The existence and uniqueness of two kinds of weak solutions for the direct problem with Neumann boundary condition are proved. We provide the uniqueness for recovering the zeroth-order coefficient and fractional order simultaneously by the Laplace transformation and Gel'fand–Levitan theory. The identification of the zeroth-order coefficient is formulated into a variational problem by the Tikhonov regularization. The existence, stability and convergence of the solution for the variational problem are provided. We deduce an adjoint problem and then use a conjugate gradient method to solve the variational problem. Two numerical examples are provided to show the effectiveness of the proposed method.

Authors:Maryam Kamranian; Mehdi Dehghan; Mehdi Tatari Pages: 181 - 196 Abstract: Publication date: January 2017 Source:Applied Numerical Mathematics, Volume 111 Author(s): Maryam Kamranian, Mehdi Dehghan, Mehdi Tatari A new adaptive moving least squares (MLS) method with variable radius of influence is presented to improve the accuracy of Meshless Local Petrov–Galerkin (MLPG) methods and to minimize the computational cost for the numerical solution of singularly perturbed boundary value problems. An error indicator based on a posteriori error estimation, accurately captures the regions of the domain with insufficient resolution and adaptively determines the new nodes location. The effectiveness of the new method is demonstrated on some singularly perturbed problems involving boundary layers.

Authors:A.H. Bhrawy; M.A. Zaky Pages: 197 - 218 Abstract: Publication date: January 2017 Source:Applied Numerical Mathematics, Volume 111 Author(s): A.H. Bhrawy, M.A. Zaky Current discretizations of variable-order fractional (V-OF) differential equations lead to numerical solutions of low order of accuracy. This paper explores a high order numerical scheme for multi-dimensional V-OF Schrödinger equations. We derive new operational matrices for the V-OF derivatives of Caputo and Riemann–Liouville type of the shifted Jacobi polynomials (SJPs). These allow us to establish an efficient approximate formula for the Riesz fractional derivative. An operational approach of the Jacobi collocation approach for the approximate solution of the V-OF nonlinear Schrödinger equations. The main characteristic behind this approach is to investigate a space–time spectral approximation for spatial and temporal discretizations. The proposed spectral scheme, both in temporal and spatial discretizations, is successfully developed to handle the two-dimensional V-OF Schrödinger equation. Numerical results indicating the spectral accuracy and effectiveness of this algorithm are presented.

Authors:F. Guillén-González; M.V. Redondo-Neble Pages: 219 - 245 Abstract: Publication date: January 2017 Source:Applied Numerical Mathematics, Volume 111 Author(s): F. Guillén-González, M.V. Redondo-Neble This paper is devoted to the numerical analysis of a first order fractional-step time-scheme (via decomposition of the viscosity) and “inf-sup” stable finite-element spatial approximations applied to the Primitive Equations of the Ocean. The aim of the paper is twofold. First, we prove that the scheme is unconditionally stable and convergent towards weak solutions of the Primitive Equations. Second, optimal error estimates for velocity and pressure are provided of order O ( k + h l ) for l = 1 or l = 2 considering either first or second order finite-element approximations (k and h being the time step and the mesh size, respectively). In both cases, these error estimates are obtained under the same constraint k ≤ C h 2 .

Authors:Heping Ma; Yonghui Qin; Qiuli Ou Pages: 246 - 259 Abstract: Publication date: January 2017 Source:Applied Numerical Mathematics, Volume 111 Author(s): Heping Ma, Yonghui Qin, Qiuli Ou The multidomain Legendre–Galerkin Chebyshev-collocation method is considered to solve one-dimensional linear evolution equations with two nonhomogeneous jump conditions. The scheme treats the first jump condition essentially and the second one naturally. We adopt appropriate base functions to deal with interfaces. The proposed method can be implemented in parallel. Error analysis shows that the approach has an optimal convergence rate. The proposed method is also applied to computing the one-dimensional Maxwell equation and the one-dimensional two phase Stefan problem, respectively. Numerical examples are given to confirm the theoretical analysis.

Authors:Franco Dassi; Luca Formaggia; Stefano Zonca Pages: 1 - 13 Abstract: Publication date: December 2016 Source:Applied Numerical Mathematics, Volume 110 Author(s): Franco Dassi, Luca Formaggia, Stefano Zonca Standard 3D mesh generation algorithms may produce a low quality tetrahedral mesh, i.e., a mesh where the tetrahedra have very small dihedral angles. In this paper, we propose a series of operations to recover these badly-shaped tetrahedra. In particular, we will focus on the shape of these undesired mesh elements by proposing a novel method to distinguish and classify them. For each of these configurations, we apply a suitable sequence of operations to get a higher mesh quality. Finally, we employ a random algorithm to avoid locks and loops in the procedure. The reliability of the proposed mesh optimization algorithm is numerically proved with several examples.

Authors:Changbum Chun; Beny Neta Pages: 14 - 25 Abstract: Publication date: Available online 3 August 2016 Source:Applied Numerical Mathematics Author(s): Changbum Chun, Beny Neta In this paper we analyze Murakami's family of fifth order methods for the solution of nonlinear equations. We show how to find the best performer by using a measure of closeness of the extraneous fixed points to the imaginary axis. We demonstrate the performance of these members as compared to the two members originally suggested by Murakami. We found several members for which the extraneous fixed points are on the imaginary axis, only one of these has 6 such points (compared to 8 for the other members). We show that this member is the best performer.

Authors:John D. Towers Pages: 26 - 40 Abstract: Publication date: Available online 12 August 2016 Source:Applied Numerical Mathematics Author(s): John D. Towers This paper presents a finite volume scheme for a scalar one-dimensional fluid-particle interaction model. When devising a finite volume scheme for this model, one difficulty that arises is how to deal with the moving source term in the PDE while maintaining a fixed grid. The fixed grid requirement comes from the ultimate goal of accommodating two or more particles. The finite volume scheme that we propose addresses the moving source term in a novel way. We use a modified computational stencil, with the lower part of the stencil shifted during those time steps when the particle crosses a mesh point. We then employ an altered convective flux to compensate the stencil shifts. The resulting scheme uses a fixed grid, preserves total momentum, and enforces several stability properties in the single-particle case. The single-particle scheme is easily extended to multiple particles by a splitting method.

Authors:Francesco Fambri; Michael Dumbser Pages: 41 - 74 Abstract: Publication date: Available online 3 August 2016 Source:Applied Numerical Mathematics Author(s): Francesco Fambri, Michael Dumbser In this paper two new families of arbitrary high order accurate spectral discontinuous Galerkin (DG) finite element methods are derived on staggered Cartesian grids for the solution of the incompressible Navier–Stokes (NS) equations in two and three space dimensions. The discrete solutions of pressure and velocity are expressed in the form of piecewise polynomials along different meshes. While the pressure is defined on the control volumes of the main grid, the velocity components are defined on edge-based dual control volumes, leading to a spatially staggered mesh. Thanks to the use of a nodal basis on a tensor-product domain, all discrete operators can be written efficiently as a combination of simple one-dimensional operators in a dimension-by-dimension fashion. In the first family, high order of accuracy is achieved only in space, while a simple semi-implicit time discretization is derived by introducing an implicitness factor θ ∈ [ 0.5 , 1 ] for the pressure gradient in the momentum equation. The real advantages of the staggering arise after substituting the discrete momentum equation into the weak form of the continuity equation. In fact, the resulting linear system for the pressure is symmetric and positive definite and either block penta-diagonal (in 2D) or block hepta-diagonal (in 3D). As a consequence, the pressure system can be solved very efficiently by means of a classical matrix-free conjugate gradient method. From our numerical experiments we find that the pressure system appears to be reasonably well-conditioned, since in all test cases shown in this paper the use of a preconditioner was not necessary. This is a rather unique feature among existing implicit DG schemes for the Navier–Stokes equations. In order to avoid a stability restriction due to the viscous terms, the latter are discretized implicitly using again a staggered mesh approach, where the viscous stress tensor is also defined on the dual mesh. The second family of staggered DG schemes proposed in this paper achieves high order of accuracy also in time by expressing the numerical solution in terms of piecewise space–time polynomials. In order to circumvent the low order of accuracy of the adopted fractional stepping, a simple iterative Picard procedure is introduced, which leads to a space–time pressure-correction algorithm. In this manner, the symmetry and positive definiteness of the pressure system are not compromised. The resulting algorithm is stable, computationally very efficient, and at the same time arbitrary high order accurate in both space and time. These features are typically not easy to obtain all at the same time for a numerical method applied to the incompressible Navier–Stokes equations. The new numerical method has been thoroughly validated for approximation polynomials of degree up to N = 11 , using a large set of non-trivial test problems in two and three space dimensions, for which either analytical, numerical or experimental reference solutions exist.

Authors:V. Nijimbere; L.J. Campbell Pages: 75 - 92 Abstract: Publication date: Available online 12 August 2016 Source:Applied Numerical Mathematics Author(s): V. Nijimbere, L.J. Campbell This paper examines the development of a time-dependent nonreflecting boundary condition (or radiation condition) for use in simulations of the propagation of internal gravity waves in a two-dimensional geophysical fluid flow configuration. First, a linear radiation condition, originally derived by Campbell and Maslowe, is implemented in some linear test cases. It involves the computation of a Laplace convolution integral which is nonlocal in time and thus requires values of the dependent variable at all previous time levels. An approximation for the integral is implemented here to reduce the expense of the computation and the results obtained are shown to be more accurate than those obtained using steady boundary conditions. For larger amplitude waves, nonlinear equations are required and the application of the linear radiation condition gives rise to instabilities. A new nonlinear time-dependent nonreflecting boundary condition is introduced which takes into account wave mean flow interactions in the vicinity of the outflow boundary by including a component corresponding to the vertical divergence of the horizontal momentum flux. This prevents the development of numerical instabilities and gives more accurate results in a nonlinear test problem than the results obtained using the linear radiation condition.

Authors:Sebastian Franz Pages: 93 - 109 Abstract: Publication date: Available online 20 August 2016 Source:Applied Numerical Mathematics Author(s): Sebastian Franz It is well known that continuous Galerkin methods lack stability for singularly perturbed convection-diffusion problems. One approach to overcome this behaviour is to use discontinuous Galerkin methods instead. Unfortunately, this increases the number of degrees of freedom and thus the computational costs. We analyse discontinuous Galerkin methods of anisotropic polynomial order and discrete discontinuous spaces. By enforcing continuity in the vertices of a mesh, the number of unknowns can be reduced while the convergence order in the dG-norm is still sustained. Numerical experiments for several polynomial elements and finite element spaces support our theoretical results.

Authors:Haiyun Dong; Maojun Li Pages: 110 - 127 Abstract: Publication date: December 2016 Source:Applied Numerical Mathematics, Volume 110 Author(s): Haiyun Dong, Maojun Li In this paper, we present a class of high order reconstructed central discontinuous Galerkin-finite element methods for the fully nonlinear weakly dispersive Green–Naghdi model, which describes a large spectrum of shallow water waves. In the proposed methods, we first reformulate the Green–Naghdi model into conservation laws coupled with an elliptic equation, and then discretize the conservation laws with reconstructed central discontinuous Galerkin methods and the elliptic equation with continuous FE methods. The reconstructed central discontinuous Galerkin methods can be viewed as a class of fast central discontinuous Galerkin methods, in which we replace the standard formula for the numerical solution defined on the dual mesh in the central discontinuous Galerkin method with a projection equation in the L 2 sense. The proposed methods reduce the computational cost of the traditional methods by nearly half but still maintain the formal high order accuracy. We study the L 2 stability and an L 2 a priori error estimate for smooth solutions of the reconstructed central discontinuous Galerkin method for linear hyperbolic equation. Numerical tests are presented to illustrate the accuracy and computational efficiency of the proposed method.

Authors:Alexandru Mihai Bica; Mircea Curila; Sorin Curila Pages: 128 - 147 Abstract: Publication date: December 2016 Source:Applied Numerical Mathematics, Volume 110 Author(s): Alexandru Mihai Bica, Mircea Curila, Sorin Curila A new iterative numerical method to solve two-point boundary value problems associated to functional differential equations of even order is proposed. The method uses a cubic spline interpolation procedure activated at each iterative step. The convergence of the method is proved and it is tested on some numerical experiments. The notion of numerical stability with respect to the choice of the first iteration is introduced proving that the proposed method is numerically stable in this sense.

Authors:Tamara Kogan; Luba Sapir; Amir Sapir; Ariel Sapir Pages: 148 - 158 Abstract: Publication date: Available online 30 August 2016 Source:Applied Numerical Mathematics Author(s): Tamara Kogan, Luba Sapir, Amir Sapir, Ariel Sapir This paper presents a class of stationary iterative processes with convergence order equal to the growth rate of generalized Fibonacci sequences. We prove that the informational and computational efficiency of the processes of our class tend to 2 from below. The paper illustrates a connection of the methods of the class with the nonstationary iterative method suggested by our previous paper, whose efficiency index equals to 2. We prove that the efficiency of the nonstationary iterative method, measured by Ostrowski-Traub criteria, is maximal among all iterative processes of order 2.

Authors:Mohan K. Kadalbajoo; Alpesh Kumar; Lok Pati Tripathi Pages: 159 - 173 Abstract: Publication date: December 2016 Source:Applied Numerical Mathematics, Volume 110 Author(s): Mohan K. Kadalbajoo, Alpesh Kumar, Lok Pati Tripathi In this article, we present a radial basis function based implicit explicit numerical method to solve the partial integro-differential equation which describes the nature of the option price under jump diffusion model. The governing equation is time semi discrtized by using the implicit–explicit backward difference method of order two (IMEX-BDF2) followed by radial basis function based finite difference (RBF-FD) method. The numerical scheme derived for European option is extended for American option by using operator splitting method. Numerical results for put and call option under Merton and Kou models are given to illustrate the efficiency and accuracy of the present method. The stability of time semi discretized scheme is also proved.

Authors:Jueyu Wang; Detong Zhu Pages: 174 - 189 Abstract: Publication date: Available online 30 August 2016 Source:Applied Numerical Mathematics Author(s): Jueyu Wang, Detong Zhu In this paper, we propose an inexact-Newton via GMRES (generalized minimal residual) subspace method without line search technique for solving symmetric nonlinear equations. The iterative direction is obtained by solving the Newton equation of the system of nonlinear equations with the GMRES algorithm. The global convergence and local superlinear convergence rate of the proposed algorithm are established under some reasonable conditions. Finally, the numerical results are reported to show the effectiveness of the proposed algorithm.

Authors:Michael V. Klibanov; Loc H. Nguyen; Kejia Pan Pages: 190 - 203 Abstract: Publication date: December 2016 Source:Applied Numerical Mathematics, Volume 110 Author(s): Michael V. Klibanov, Loc H. Nguyen, Kejia Pan Inverse scattering problems without the phase information arise in imaging of nanostructures, whose sizes are hundreds of nanometers, as well as in imaging of biological cells. The governing equation is the 3-D generalized Helmholtz equation with the unknown coefficient, which represents the spatially distributed dielectric constant. It is assumed in the classical inverse scattering problem that both the modulus and the phase of the complex valued scattered wave field are measured outside of a scatterer. Unlike this, it is assumed here that only the modulus of the complex valued scattered wave field is measured on a certain interval of frequencies. The phase is not measured. In this paper a substantially modified reconstruction procedure of [25] is developed and numerically implemented. Ranges of parameters, which are realistic for imaging of nanostructures, are used in numerical examples. Note that numerical studies were not carried out in [25].

Authors:Arvet Pedas; Enn Tamme; Mikk Vikerpuur Pages: 204 - 214 Abstract: Publication date: December 2016 Source:Applied Numerical Mathematics, Volume 110 Author(s): Arvet Pedas, Enn Tamme, Mikk Vikerpuur We consider a class of boundary value problems for linear fractional weakly singular integro-differential equations which involve Caputo-type derivatives. Using an integral equation reformulation of the boundary value problem, we first study the regularity of the exact solution. Based on the obtained regularity properties and spline collocation techniques, the numerical solution of the boundary value problem by suitable non-polynomial approximations is discussed. Optimal global convergence estimates are derived and a super-convergence result for a special choice of grid and collocation parameters is given. A numerical illustration is also presented.

Authors:M. Kordy; E. Cherkaev; P. Wannamaker Pages: 1 - 18 Abstract: Publication date: Available online 11 June 2016 Source:Applied Numerical Mathematics Author(s): M. Kordy, E. Cherkaev, P. Wannamaker This work develops a model order reduction method for a numerical solution of an inverse multi-frequency eddy current problem using a rational interpolation of the transfer function in the complex plane. We use the Pade interpolation in the complex frequency plane; this allows us to speed up the calculation of the frequency-dependent Jacobian in the inversion procedure without loosing accuracy. Interpolating frequencies are chosen adaptively to reduce the maximal approximation error. We use the error indicator that is equivalent to a seminorm of the residual. The efficiency of the developed approach is demonstrated by applying it to the inverse magnetotelluric problem, which is a geophysical electromagnetic remote sensing method used in mineral, geothermal, and groundwater exploration. In this application, the transfer function values are needed for shifts in a purely imaginary interval. Thus we consider the interpolating shifts in the same interval as well as in a purely real interval, containing the spectrum of the operator. Numerical tests show an excellent performance of the proposed methods characterized by a significant reduction of computational time without loss of accuracy of the calculated Jacobian.

Authors:C.F. Bracciali; A. Sri Ranga; A. Swaminathan Pages: 19 - 40 Abstract: Publication date: Available online 11 June 2016 Source:Applied Numerical Mathematics Author(s): C.F. Bracciali, A. Sri Ranga, A. Swaminathan When a nontrivial measure μ on the unit circle satisfies the symmetry d μ ( e i ( 2 π − θ ) ) = − d μ ( e i θ ) then the associated orthogonal polynomials on the unit circle, say Φ n , are all real. In this case, in 1986, Delsarte and Genin have shown that the two sequences of para-orthogonal polynomials { z Φ n ( z ) + Φ n ⁎ ( z ) } and { z Φ n ( z ) − Φ n ⁎ ( z ) } , where Φ n ⁎ ( z ) = z n Φ n ( 1 / z ‾ ) ‾ , satisfy three term recurrence formulas and have also explored some further consequences of these sequences of polynomials such as their connections to sequences of orthogonal polynomials on the interval [ − 1 , 1 ] . The same authors, in 1988, have also provided a means to extend these results to cover any nontrivial measure on the unit circle. However, only recently the extension associated with the para-orthogonal polynomials z Φ n ( z ) − Φ n ⁎ ( z ) was thoroughly explored, especially from the point of view of three term recurrence and chain sequences. The main objective of the present article is to provide the theory surrounding the extension associated with the para-orthogonal polynomials z Φ n ( z ) + Φ n ⁎ ( z ) for any nontrivial measure on the unit circle. As an important application of the theory, a characterization for the existence of the integral ∫ 0 2 π e i θ − w − 2 d μ ( e i θ ) , where w is such that w = 1 , is given in terms of the coefficients α n − 1 = − Φ n ( 0 ) ‾ ... PubDate: 2016-06-16T18:07:53Z DOI: 10.1016/j.apnum.2016.05.008 Issue No:Vol. 109 (2016)

Authors:Yali Gao; Liquan Mei Pages: 41 - 60 Abstract: Publication date: November 2016 Source:Applied Numerical Mathematics, Volume 109 Author(s): Yali Gao, Liquan Mei In this paper, implicit–explicit multistep Galerkin methods are studied for two-dimensional nonlinear Schrödinger equations and coupled nonlinear Schrödinger equations. The spatial discretization is based on Galerkin method using linear and quadratic basis functions on triangular and rectangular finite elements. And the implicit–explicit multistep method is used for temporal discretization. Linear and nonlinear numerical tests are presented to verify the validity and efficiency of the numerical methods. The numerical results record that the optimal order of the error in L 2 and L ∞ norm can be reached.

Authors:J.B. Collins; P.A. Gremaud Pages: 61 - 72 Abstract: Publication date: November 2016 Source:Applied Numerical Mathematics, Volume 109 Author(s): J.B. Collins, P.A. Gremaud In this paper we analyze the convergence of the domain decomposition method applied to transport problems on networks. In particular, we derive estimates for the number of required iterations for linear problems. These estimates can be used to determine when the implementation of domain decomposition methods would be beneficial for this type of problems.

Authors:Po-Hsien Lin; S.-T. John Yu Pages: 73 - 84 Abstract: Publication date: Available online 7 June 2016 Source:Applied Numerical Mathematics Author(s): Po-Hsien Lin, S.-T. John Yu A set of model equations are proposed to simulate waves generated by unsteady, low-speed, nearly incompressible air and water flows. The equations include the continuity and momentum equations with pressure and velocity as the unknowns. Compressibility effect associated with waves motion is directly tracked by time-accurate calculation of pressure fluctuations. The corresponding density changes are modeled by using the bulk modulus of the medium. The three-dimensional equations are shown to be hyperbolic by analyzing eigenvalues and eigenvectors of the composite Jacobian matrix of the equations. Specifically, the matrix is shown to be diagonalizable and have a real spectrum. Moreover, an analytical form of the Riemann invariants of the one-dimensional equations are derived. To validate the model equations, the space-time Conservation Element and Solution Element (CESE) method and the SOLVCON code are employed to solve the two-dimensional equations. Aeolian tones generated by air and water flows passing a cylinder and over an open cavity are simulated. Numerical results compare well with previously reported data.

Authors:Thomas Wolf; Heiko K.F. Panzer; Boris Lohmann Pages: 85 - 95 Abstract: Publication date: November 2016 Source:Applied Numerical Mathematics, Volume 109 Author(s): Thomas Wolf, Heiko K.F. Panzer, Boris Lohmann A new version of the alternating directions implicit (ADI) iteration for the solution of large-scale Lyapunov equations is introduced. It generalizes the hitherto existing iteration, by incorporating tangential directions in the way they are already available for rational Krylov subspaces. Additionally, first strategies to adaptively select shifts and tangential directions in each iteration are presented. Numerical examples emphasize the potential of the new results.

Authors:Jingjun Zhao; Rui Zhan; Alexander Ostermann Pages: 96 - 108 Abstract: Publication date: November 2016 Source:Applied Numerical Mathematics, Volume 109 Author(s): Jingjun Zhao, Rui Zhan, Alexander Ostermann The aim of this paper is to analyze stability properties of explicit exponential integrators for three kinds of delay differential equations. First, linear autonomous delay differential equations are studied, and sufficient conditions for P- and GP-contractivity of explicit exponential Runge–Kutta methods are given. Second, for linear nonautonomous test equations, PN- and GPN-stability of a particular Magnus integrator is investigated. It is shown that the Magnus integrator is GPN-stable and convergent of order two. Finally, for semilinear delay differential equations, RN- and GRN-stability of explicit exponential Runge–Kutta methods is studied and sufficient conditions for GRN-stability are derived. Some examples of P-, GP-, RN- and GRN-stable exponential integrators are given, and numerical experiments that illustrate the theoretical results are included.

Authors:Matthias Schlottbom Pages: 109 - 122 Abstract: Publication date: November 2016 Source:Applied Numerical Mathematics, Volume 109 Author(s): Matthias Schlottbom We use a diffuse interface method for solving Poisson's equation with a Dirichlet condition on an embedded curved interface. The resulting diffuse interface problem is identified as a standard Dirichlet problem on approximating regular domains. We estimate the errors introduced by these domain perturbations, and prove convergence and convergence rates in the H 1 -norm, the L 2 -norm and the L ∞ -norm in terms of the width of the diffuse layer. For an efficient numerical solution we consider the finite element method for which another domain perturbation is introduced. These perturbed domains are polygonal and non-convex in general. We prove convergence and convergences rates in the H 1 -norm and the L 2 -norm in terms of the layer width and the mesh size. In particular, for the L 2 -norm estimates we present a problem adapted duality technique, which crucially makes use of the error estimates derived for the regularly perturbed domains. Our results are illustrated by numerical experiments, which also show that the derived estimates are sharp.

Authors:Philipp Bader; David I. McLaren; G.R.W. Quispel; Marcus Webb Pages: 123 - 137 Abstract: Publication date: November 2016 Source:Applied Numerical Mathematics, Volume 109 Author(s): Philipp Bader, David I. McLaren, G.R.W. Quispel, Marcus Webb It is a classical theorem of Liouville that Hamiltonian systems preserve volume in phase space. Any symplectic Runge–Kutta method will respect this property for such systems, but it has been shown by Iserles, Quispel and Tse and independently by Chartier and Murua that no B-Series method can be volume preserving for all volume preserving vector fields. In this paper, we show that despite this result, symplectic Runge–Kutta methods can be volume preserving for a much larger class of vector fields than Hamiltonian systems, and discuss how some Runge–Kutta methods can preserve a modified measure exactly.

Authors:Wei-ping Shen; Chong Li; Xiao-qing Jin; Jen-chih Yao Pages: 138 - 156 Abstract: Publication date: November 2016 Source:Applied Numerical Mathematics, Volume 109 Author(s): Wei-ping Shen, Chong Li, Xiao-qing Jin, Jen-chih Yao We consider the convergence problem of some Newton-type methods for solving the inverse singular value problem with multiple and positive singular values. Under the nonsingularity assumption of the relative generalized Jacobian matrices at the solution c ⁎ , a convergence analysis for the multiple and positive case is provided and the superlinear or quadratical convergence properties are proved. Moreover, numerical experiments are given in the last section and comparisons are made.

Authors:M.I.M. Copetti; M. Aouadi Pages: 157 - 183 Abstract: Publication date: November 2016 Source:Applied Numerical Mathematics, Volume 109 Author(s): M.I.M. Copetti, M. Aouadi The problem of thermoviscoelastic quasi-static contact between a rod and a rigid obstacle, when the diffusion effect is taken into account, is modeled and analyzed. The contact is modeled by the Signorini's condition and the stress–strain constitutive equation is of the Kelvin–Voigt type. In the quasi-static case, the governing equations correspond to the coupling of an elliptic and two parabolic equations. It poses some new mathematical difficulties due to the nonlinear boundary conditions. The existence of solutions is proved as the limit of solutions to a penalized problem. Moreover, we show that the weak solution converges to zero exponentially as time goes to infinity. Finally, we give some computational results where the influence of diffusion and viscosity are illustrated in contact.

Authors:María G. Armentano; Verónica Moreno Pages: 184 - 207 Abstract: Publication date: November 2016 Source:Applied Numerical Mathematics, Volume 109 Author(s): María G. Armentano, Verónica Moreno The goal of this work is to introduce a local and a global interpolator in Jacobi-weighted spaces, with optimal order of approximation in the context of the p-version of finite element methods. Then, an a posteriori error indicator of the residual type is proposed for a model problem in two dimensions and, in the mathematical framework of the Jacobi-weighted spaces, the equivalence between the estimator and the error is obtained on appropriate weighted norm.

Authors:Mehdi Dehghan; Mostafa Abbaszadeh Pages: 208 - 234 Abstract: Publication date: November 2016 Source:Applied Numerical Mathematics, Volume 109 Author(s): Mehdi Dehghan, Mostafa Abbaszadeh The element free Galerkin technique is a meshless method based on the variational weak form in which the test and trial functions are the shape functions of moving least squares approximation. Since the shape functions of moving least squares approximation do not have the Kronecker property thus the Dirichlet boundary condition can not be applied directly and also in this case obtaining an error estimate is not simple. The main aim of the current paper is to propose an error estimate for the extracted numerical scheme from the element free Galerkin method. To this end, we select the fractional cable equation with Dirichlet boundary condition. Firstly, we obtain a time-discrete scheme based on a finite difference formula with convergence order O ( τ 1 + min { α , β } ) , then we use the meshless element free Galerkin method, to discrete the space direction and obtain a full-discrete scheme. Also, for calculating the appeared integrals over the boundary and the domain of problem the Gauss–Legendre quadrature rule has been used. In the next, we change the main problem with Dirichlet boundary condition to a new problem with Robin boundary condition. Then, we show that the new technique is unconditionally stable and convergent using the energy method. We show convergence orders of the time discrete scheme and the full discrete scheme are O ( τ 1 + min { α , β } ) and O ( r p + 1 + τ 1 + min { α , β } ) , respectively. So, we can say that the main aim of this paper is as follows, (1) Transferring the main problem with Dirichlet boundary condition (old problem) to a problem with Robin boundary condition (new problem), (2) Showing that with special condition (when σ → + ∞ ) the solution of the new problem is convergent to the solution of the old problem, (3) Obtaining an error estimate for the new problem. Numerical examples confirm the efficiency and accuracy of the proposed scheme.

Authors:Y. Mesri; M. Khalloufi; E. Hachem Pages: 235 - 249 Abstract: Publication date: November 2016 Source:Applied Numerical Mathematics, Volume 109 Author(s): Y. Mesri, M. Khalloufi, E. Hachem In this paper we derive a multi-dimensional mesh adaptation method which produces optimal meshes for quadratic functions, positive semi-definite. The method generates anisotropic adaptive meshes as quasi-uniform ones in some metric space, with the tensor metric being computed based on interpolation error estimates. It does not depend, a priori, on the PDEs at hand in contrast to residual methods. The estimated error is then used to steer local modifications of the mesh in order to reach a prescribed level of error in L p -norm or a prescribed number of elements. The L p -norm of the estimated error is then minimized in order to get an optimal mesh. Numerical examples in 2D and 3D for analytic challenging problems and an application to a Computational Fluid Dynamics problem are presented and discussed in order to show how the proposed method recovers optimal convergence rates as well as to demonstrate its computational performance.

Authors:S. McKee; Jose A. Cuminato Abstract: Publication date: Available online 13 October 2016 Source:Applied Numerical Mathematics Author(s): S. McKee, Jose A. Cuminato A new integration technique, which is suitable for integrands with multiple weak singularities, is introduced. Local truncation errors are given. This scheme, when applied to the Beta function, is shown to emerge naturally from discrete fractional integration. To illustrate the effectiveness of the integration method a numerical example is provided, with somewhat unexpected convergence results.

Authors:Filomena D. d' Almeida; Rosário Fernandes Abstract: Publication date: Available online 15 October 2016 Source:Applied Numerical Mathematics Author(s): Filomena D. d' Almeida, Rosário Fernandes For the solution of a weakly singular Fredholm integral equation of the 2nd kind defined on a Banach space, for instance L 1 ( [ a , b ] ) , the classical projection methods with the discretization of the approximating operator on a finite dimensional subspace usually use a basis of this subspace built with grids on [ a , b ] . This may require a large dimension of the subspace. One way to overcome this problem is to include more information in the approximating operator or to compose one classical method with one step of iterative refinement. This is the case of Kulkarni method or iterated Kantorovich method. Here we compare these methods in terms of accuracy and arithmetic workload. A theorem stating comparable error bounds for these methods, under very weak assumptions on the kernel, the solution and the space where the problem is set, is given.

Authors:V. Baron; Y. Coudière; P. Sochala Abstract: Publication date: Available online 17 October 2016 Source:Applied Numerical Mathematics Author(s): V. Baron, Y. Coudière, P. Sochala We derive some a posteriori error estimates for the Richards equation. This parabolic equation is nonlinear in space and in time, thus its resolution requires fixed-point iterations within each time step. We measure the approximation error with the dual norm of the residual. A computable upper bound of this error consists of several estimators involving adequate reconstructions based on the degrees of freedom of the scheme. The space and time reconstructions are specified for a two-step backward differentiation formula and a discrete duality finite volume scheme. Our strategy to decrease the computational cost relies on an aggregation of the estimators in three components: space discretization, time discretization, and linearization. We propose an algorithm to stop the fixed-point iterations after the linearization error becomes negligible, and to choose the time step in order to balance the time and space errors. We analyze the influence of the parameters of this algorithm on three test cases and quantify the gain obtained in comparison with a classical simulation.

Authors:Laura Gori; Francesca Pitolli Abstract: Publication date: Available online 11 October 2016 Source:Applied Numerical Mathematics Author(s): Laura Gori, Francesca Pitolli We construct a new class of approximating functions that are M-refinable and provide shape preserving approximations. The refinable functions in the class are smooth, compactly supported, centrally symmetric and totally positive. Moreover, their refinable masks are associated with convergent subdivision schemes. The presence of one or more shape parameters gives a great flexibility in the applications. Some examples for dilation M = 4 and M = 5 are also given.

Authors:Zhongqiang Zhang; Heping Ma Abstract: Publication date: Available online 8 October 2016 Source:Applied Numerical Mathematics Author(s): Zhongqiang Zhang, Heping Ma We introduce a class of explicit balanced schemes for stochastic differential equations with coefficients of superlinearly growth satisfying a global monotone condition. The first scheme is a balanced Euler scheme and is of order half in the mean-square sense whereas it is of order one under additive noise. The second scheme is a balanced Milstein scheme, which is of order one in the mean-square sense. Some numerical results are presented.

Authors:Jingyang Guo; Jae-Hun Jung Abstract: Publication date: Available online 11 October 2016 Source:Applied Numerical Mathematics Author(s): Jingyang Guo, Jae-Hun Jung Essentially non-oscillatory (ENO) and weighted ENO (WENO) methods are efficient high order numerical methods for solving hyperbolic conservation laws designed to reduce the Gibbs oscillations. The original ENO and WENO methods are based on the polynomial interpolation and the overall convergence rate is uniquely determined by the total number of interpolation points involved for the approximation. In this paper, we propose non-polynomial ENO and WENO finite volume methods in order to enhance the local accuracy and convergence. The infinitely smooth radial basis functions (RBFs) are adopted as a non-polynomial interpolation basis. Particularly we use the multi-quadratic and Gaussian RBFs. The non-polynomial interpolation such as the RBF interpolation offers the flexibility to control the local error by optimizing the free parameter. Then we show that the non-polynomial interpolation can be represented as a perturbation of the polynomial interpolation. To guarantee the essentially non-oscillatory property, the monotone polynomial interpolation method is introduced as a switching method to the polynomial reconstruction adaptively near the non-smooth area. The numerical results show that the developed non-polynomial ENO and WENO methods with the monotone polynomial interpolation method enhance the local accuracy and give sharper solution profile.

Authors:G.Yu. Kulikov; M.V. Kulikova Abstract: Publication date: Available online 3 October 2016 Source:Applied Numerical Mathematics Author(s): G.Yu. Kulikov, M.V. Kulikova This paper further advances the idea of accurate Gaussian filtering towards efficient cubature Kalman filters for estimating continuous-time nonlinear stochastic systems with discrete measurements. It implies that the moment differential equations describing evolution of the predicted mean and covariance of the propagated Gaussian density in time are solved accurately, i.e. with negligible error. The latter allows the total error of the cubature Kalman filtering to be reduced significantly and results in a new accurate continuous-discrete cubature Kalman filtering method. At the same time, we revise the earlier developed version of the accurate continuous-discrete extended Kalman filter by amending the involved iteration and relaxing the utilized global error control mechanism. In addition, we build a mixed-type method, which unifies the best features of the accurate continuous-discrete extended and cubature Kalman filters. More precisely, the time updates are done in this state estimator as those in the first filter whereas the measurement updates are conducted with use of the third-degree spherical-radial cubature rule applied for approximating the arisen Gaussian-weighted integrals. All these are examined in severe conditions of tackling a seven-dimensional radar tracking problem, where an aircraft executes a coordinated turn, and compared to the state-of-the-art cubature Kalman filters.

Authors:J.B. Francisco; F.S. Viloche Bazán; M. Weber Mendonça Abstract: Publication date: Available online 5 October 2016 Source:Applied Numerical Mathematics Author(s): J.B. Francisco, F.S. Viloche Bazán, M. Weber Mendonça This paper concerns a non-monotone algorithm for minimizing differentiable functions on closed sets. A general numerical scheme is proposed which combines a regularization/trust-region framework with a non-monotone strategy. Global convergence to stationary points is proved under usual assumptions. Numerical experiments for a particular version of the general algorithm are reported. In addition, a promising numerical scheme for medium/large-scale orthogonal Procrustes problem is also proposed and numerically illustrated.

Authors:Fidalgo Abstract: Publication date: January 2017 Source:Applied Numerical Mathematics, Volume 111 Author(s): U. Fidalgo We use a connection between interpolatory quadrature formulas and Fourier series to find a wide class of convergent schemes of interpolatory quadrature rules. In the process we use techniques coming from Riemann–Hilbert problems for varying measures and convex analysis.

Authors:Alemdar Hasanov; Balgaisha Mukanova Abstract: Publication date: Available online 13 September 2016 Source:Applied Numerical Mathematics Author(s): Alemdar Hasanov, Balgaisha Mukanova Inverse problem of identifying the unknown spacewise dependent source F ( x ) in 1D wave equation u t t = c 2 u x x + F ( x ) G ( t ) + h ( x , t ) , ( x , t ) ∈ ( 0 , 1 ) × ( 0 , T ) , from the Neumann-type measured output g ( t ) : = u x ( 0 , t ) is investigated. Most studies have attempted to reconstruct an unknown spacewise dependent source F ( x ) from the final observation u T ( x ) : = u ( x , T ) . Since a boundary measured data is most feasible from an engineering viewpoint, the identification problem has wide applications, in particular, in electrical networks governed by harmonically varying source for the linear wave equation u t t − u x x = F ( x ) c o s ( ω t ) , where ω > 0 is the frequency and F ( x ) is an unknown source term. In this paper Fourier Collocation Algorithm for reconstructing the spacewise dependent source F ( x ) is developed. This algorithm is based on Fourier expansion of the direct problem solution applied to the minimization problem for Tikhonov functional, by taking then a partial N-sum of the Fourier expansion. Tikhonov regularization is then applied to the obtained discrete ill-posed problem. To obtain high quality reconstruction in large values of the noise level, a numerical filtering algorithm is used for smoothing the noisy data. As an application, we demonstrate the ability of the algorithm on benchmark problems, in particular, on source identification problem in electrical networks governed by mono-frequency source. Numerical results show that the proposed algorithm allows to reconstruct the spacewise dependent source F ( x ) with enough high accuracy, in the presence of high noise levels.

Authors:Martin Bourne; Joab Winkler Abstract: Publication date: Available online 6 September 2016 Source:Applied Numerical Mathematics Author(s): Martin Bourne, Joab R. Winkler, Su Yi This paper considers the computation of the degree t of an approximate greatest common divisor d ( y ) of two Bernstein polynomials f ( y ) and g ( y ) , which are of degrees m and n respectively. The value of t is computed from the QR decomposition of the Sylvester resultant matrix S ( f , g ) and its subresultant matrices S k ( f , g ) , k = 2 , … , min ( m , n ) , where S 1 ( f , g ) = S ( f , g ) . It is shown that the computation of t is significantly more complicated than its equivalent for two power basis polynomials because (a) S k ( f , g ) can be written in several forms that differ in the complexity of the computation of their entries, (b) different forms of S k ( f , g ) may yield different values of t, and (c) the binomial terms in the entries of S k ( f , g ) may cause the ratio of its entry of maximum magnitude to its entry of minimum magnitude to be large, which may lead to numerical problems. It is shown that the QR decomposition and singular value decomposition (SVD) of the Sylvester matrix and its subresultant matrices yield better results than the SVD of the Bézout matrix, and that f ( y ) and g ( y ) must be processed before computations are performed on these resultant and subresultant matrices in order to obtain good results.