On the stability of some second order numerical methods for weak approximation of Itô SDEs

In this paper, we first investigate the stability of two weak second order methods introduced by Debrabant and Rößler (Appl Numer Math 59:582–594, 2009) and Platen (Math Comput Simulation 38:69–76, 1995). We then propose a new weak second order predictor-corrector method, with an improved stability properties, based on the Rößler’s method as the predictor and the implicit method of Platen as the corrector. The stability functions of these methods, applied to a scalar linear test equation with multiplicative noise, are determined and their regions of stability are then compared with the corresponding stability regions of the test equation. Furthermore, we also investigate mean square stability (MS-stability) of these methods applied to a linear Itô 2-dimensional stochastic differential test equation. Numerical examples will be presented to support the theoretical results.     

A moment inequality of the Marcinkiewicz–Zygmund type for some weakly dependent random fields

The goal of this note is to give a new moment inequality for sums of some weakly dependent random fields. Our result extends moment bounds given by Wu (2007) or Liu and Lin (2009) for causal autoregressive processes and follows by using recursive applications of the Burkhölder inequality for martingales.     

Concentration-compactness principles for Moser–Trudinger inequalities: new results and proofs

We are concerned with the best exponent in Concentration-Compactness principles for the borderline case of the Sobolev inequality. We present a new approach, which both yields a rigorous proof of the relevant principle in the standard case when functions vanishing on the boundary are considered, and enables us to deal with functions with unrestricted boundary values.     

Explicit pseudo two-step exponential Runge–Kutta methods for the numerical integration of first-order differential equations

Numerical Algorithms - This paper is devoted to the explicit pseudo two-step exponential Runge–Kutta (EPTSERK) methods for the numerical integration of first-order ordinary differential...     

Existence and multiplicity results for some nonlinear problems with singular ?-Laplacian

Using Leray-Schauder degree theory we obtain various existence and multiplicity results for nonlinear boundary value problems

     

Maximum principles and isoperimetric inequalities for some Monge–Ampère-type problems

In this note we derive a maximum principle for an appropriate functional combination of u(x)$u(\mathbf{x})$ and |∇u|²${|\mathrm{\nabla }u|}^2$, where u(x)$u(\mathbf{x})$ is a strictly convex classical solution to a general class of Monge–Ampère equations. This maximum principle is then employed to establish some isoperimetric inequalities of interest in the theory of surfaces of constant Gauss curvature in R^N+1${\mathbb{R}}^{N+1}$.     

Modified Runge–Kutta Verner methods for the numerical solution of initial and boundary-value problems with engineering applications

Two new modified Runge–Kutta methods with minimal phase-lag are developed for the numerical solution of Ordinary Differential Equations with engineering applications. These methods are based on the well-known Runge–Kutta method of Verner RK6(5)9b (see J.H. Verner, some Runge–Kutta formula pairs, SIAM J. Numer. Anal 28 (1991) 496–511) of order six. Numerical and theoretical results in some problems of the plate deflection theory show that this new approach is more efficient compared with the well-known classical sixth order Runge–Kutta Verner method.     

A new family of Runge–Kutta type methods for the numerical integration of perturbed oscillators

Summary. Our task in this paper is to present a new family of methods of the Runge–Kutta type for the numerical integration of perturbed oscillators. The key property is that those algorithms are able to integrate exactly, without truncation error, harmonic oscillators, and that, for perturbed problems the local error contains the perturbation parameter as a factor. Some numerical examples show the excellent behaviour when they compete with Runge–Kutta–Nystr?m type methods. Received June 12, 1997 / Revised version received July 9, 1998     

On convergence of numerical methods for variational–hemivariational inequalities under minimal solution regularity

Hemivariational inequalities have been successfully employed for mathematical and numerical studies of application problems involving nonsmooth, nonmonotone and multivalued relations. In recent years, error estimates have been derived for numerical solutions of hemivariational inequalities under additional solution regularity assumptions. Since the solution regularity properties have not been rigorously proved for hemivariational inequalities, it is important to explore the convergence of numerical solutions of hemivariational inequalities without assuming additional solution regularity. In this paper, we present a general convergence result enhancing existing results in the literature.     

Decomposition methods for polyhedral approximation of the Edgeworth–Pareto hull

In the framework of multiobjective optimization (MOO) techniques, there is a group of methods that attract growing attention. These methods are based on the approximation of the Edgeworth–Pareto hull (EPH), that is, the largest set (in the sense of inclusion) that has the same Pareto frontier as the feasible objective set [1, 2]. The block separable structure of MOO problems is used in this paper to improve the efficiency of EPH approximation methods in the case when EPH is convex. In this case, polyhedral approximation can be applied to EPH. Its practical importance was shown, for example, in [3, 4]. In this paper, a two-level system is considered that consists of an upper (coordinating) level and a finite number of blocks interacting through the upper level. It is assumed that the optimization objectives are related only to the variables of the upper level. The approximation method is based on the polyhedral approximation of EPH for single blocks, followed by the application of these results for approximating the EPH for the whole MOO problem.     

Asymptotics of the approximation of continuous and differentiable functions by the singular integrals of de la Vallée Poussin

The complete asymptotic developments in powers of 1/n are derived for quantities characterizing approximation by singular integrals of de la Vallée Poussin $$V_n (f:x) = \frac{1}{{\Delta _n }}\int_{ - \pi }^\pi {f(x + t)} \cos ^{2n} \frac{t}{2}dt;\Delta _n = \int_{ - \pi }^\pi {\cos ^{2n} \frac{t}{2}dt}$$ of the function classes Lipa, 0w ^(r), r?1 an integer.     

Quadrature formulas with exponential convergence and calculation of the Fermi–Dirac integrals

A class of functions for which the trapezoidal rule has superpower convergence is described: these are infinitely differentiable functions all of whose odd derivatives take equal values at the left and right endpoints of the integration interval. An heuristic law is revealed; namely, the convergence exponentially depends on the number of nodes, and the exponent equals the ratio of the length of integration interval to the distance from this interval to the nearest pole of the integrand. On the basis of the obtained formulas, a method for calculating the Fermi–Dirac integrals of half-integer order is proposed, which is substantially more economical than all known computational methods. As a byproduct, an asymptotics of the Bernoulli numbers is found.     

High order methods on Shishkin meshes for singular perturbation problems of convection–diffusion type

In this paper we construct and analyze two compact monotone finite difference methods to solve singularly perturbed problems of convection–diffusion type. They are defined as HODIE methods of order two and three, i.e., the coefficients are determined by imposing that the local error be null on a polynomial space. For arbitrary meshes, these methods are not adequate for singularly perturbed problems, but using a Shishkin mesh we can prove that the methods are uniformly convergent of order two and three except for a logarithmic factor. Numerical examples support the theoretical results. This revised version was published online in June 2006 with corrections to the Cover Date.     

Convergence of descent methods for semi-algebraic and tame problems: proximal algorithms, forward–backward splitting, and regularized Gauss–Seidel methods

In view of the minimization of a nonsmooth nonconvex function f, we prove an abstract convergence result for descent methods satisfying a sufficient-decrease assumption, and allowing a relative error tolerance. Our result guarantees the convergence of bounded sequences, under the assumption that the function f satisfies the Kurdyka–?ojasiewicz inequality. This assumption allows to cover a wide range of problems, including nonsmooth semi-algebraic (or more generally tame) minimization. The specialization of our result to different kinds of structured problems provides several new convergence results for inexact versions of the gradient method, the proximal method, the forward–backward splitting algorithm, the gradient projection and some proximal regularization of the Gauss–Seidel method in a nonconvex setting. Our results are illustrated through feasibility problems, or iterative thresholding procedures for compressive sensing.     

On the constants for some fractional Gagliardo–Nirenberg and Sobolev inequalities

We consider the inequalities of Gagliardo–Nirenberg and Sobolev in ${\mathbb{R}}^d$, formulated in terms of the Laplacian $\Delta$ and of the fractional powers $D^n?{\sqrt{?\Delta }}^n$ with real $n?0$; we review known facts and present novel, complementary results in this area. After illustrating the equivalence between these two inequalities and the relations between the corresponding sharp constants and maximizers, we focus the attention on the ${?}^2$ case where, for all sufficiently regular $f:{\mathbb{R}}^d\to \mathbb{C}$, the norm $6D^{j}f{6}_{{?}^r}$ is bounded in terms of $6f{6}_{{?}^2}$ and $6D^{n}f{6}_{{?}^2}$, for $1∕r=1∕2?(?n?j)∕d$, and suitable values of $j,n,?$ (with $j,n$ possibly noninteger). In the special cases $?=1$ and $?=j∕n+d∕2n$ (i.e., $r=+\infty$), related to previous results of Lieb and Ilyin, the sharp constants and the maximizers can be found explicitly; we point out that the maximizers can be expressed in terms of hypergeometric, Fox and Meijer functions. For the general ${?}^2$ case, we present two kinds of upper bounds on the sharp constants: the first kind is suggested by the literature, the second one is an alternative proposal of ours, often more precise than the first one. We also derive two kinds of lower bounds. Combining all the available upper and lower bounds, the sharp constants are confined to quite narrow intervals. Several examples are given, including the numerical values of the previously mentioned bounds.     

Design of a new dynamical core for global atmospheric models based on some efficient numerical methods

A careful study on the integral properties of the primitive hydrostatic balance equations for baroclinic atmosphere is carried out, and a new scheme to design the global adiabatic model of atmospheric dynamics is presented. This scheme includes a method of weighted equal-area mesh and a fully discrete finite difference method with quadratic and linear conservations for solving the primitive equation system. Using this scheme, we established a new dynamical core with adjustable high resolution acceptable to the available     

Non-standard Lagrange–Burman methods for the numerical integration of differential equations

We present new non-standard methods based on Mickens' ideas for constructing numerical schemes for differential equations. By analysing his schemes, we were able to give generalized versions by means of Lagrange–Burman expansions. Our generalization provides a theoretical frame to understand non-standard finite difference methods and also to analyse the errors in the approximations. We apply the new non-standard methods to several examples.     

Review of wavelet methods for the solution of reaction–diffusion problems in science and engineering

Wavelet method is a recently developed tool in applied mathematics. Investigation of various wavelet methods, for its capability of analyzing various dynamic phenomena through waves gained more and more attention in engineering research. Starting from ‘offering good solution to differential equations’ to capturing the nonlinearity in the data distribution, wavelets are used as appropriate tools at various places to provide good mathematical model for scientific phenomena, which are usually modeled through linear or nonlinear differential equations. Review shows that the wavelet method is efficient and powerful in solving wide class of linear and nonlinear reaction–diffusion equations. This review intends to provide the great utility of wavelets to science and engineering problems which owes its origin to 1919. Also, future scope and directions involved in developing wavelet algorithm for solving reaction–diffusion equations are addressed.