Error estimates for stochastic differential games: the adverse stopping case

^** Email: frederic.bonnans{at}inria.fr^*** Email: stefania.maroso{at}inria.fr^**** Email: zidani{at}ensta.fr We obtain error bounds for monotone approximation schemes ofa particular Isaacs equation. This is an extension of the theoryfor estimating errors for the Hamilton–Jacobi–Bellmanequation. To obtain the upper error bound, we consider the ‘Krylovregularization’ of the Isaacs equation to build an approximatesub-solution of the scheme. To get the lower error bound, weextend the method of Barles & Jakobsen (2005, SIAM J. Numer.Anal.) which consists in introducing a switching system whosesolutions are local super-solutions of the Isaacs equation.     

Strong stabilization of the system of linear elasticity by a Dirichlet boundary feedback

In this paper, by using the Nagy–Foias–Foguel theoryof decomposition of continuous semigroups of contractions, we prove that the system of linear elasticity is strongly stabilizable by a Dirichlet boundary feedback. We also give a concise proofof a theorem of Dafermos about the stability of thermoelasticity.     

Finite element approximation of the Cahn-Hilliard equation with concentration dependent mobility

We consider the Cahn-Hilliard equation with a logarithmic free energy and non-degenerate concentration dependent mobility. In particular we prove that there exists a unique solution for sufficiently smooth initial data. Further, we prove an error bound for a fully practical piecewise linear finite element approximation in one and two space dimensions. Finally some numerical experiments are presented.

     

Lax-Wendroff-type schemes of arbitrary order in several space dimensions

^{The second-order accurate Lax–Wendroff scheme is basedon the first three terms of a Taylor expansion in time in whichthe time derivatives are replaced by space derivatives usingthe governing evolution equations. The space derivatives arethen approximated by central differencing. In this paper, weextend this idea and truncate the Taylor expansion at an arbitraryorder. One main building block is the so-called Cauchy–Kovalevskayaprocedure to replace all the time derivatives by space derivativeswhich can be formulated for a general system of linear equationswith arbitrary order and in two- or three-space dimensions.The linear case is the main focus of this paper because theproposed high-order schemes are good candidates for the approximationof linear wave motion over long distances and times with importantapplications in aeroacoustics and electromagnetics. The stabilityand the efficiency of Lax–Wendroff-type schemes are examined.The numerical results are compared with a standard scheme foraeroacoustical applications with respect to their quality andthe computational effort. The extensions of the schemes to generalgrids, nonconstant and nonlinear cases are alsoaddressed.}     

Global existence and blow-up for the solutions to nonlinear parabolic-elliptic system modelling chemotaxis

In this paper we study the global in-time and blow-up solutionsfor the simplified Keller–Segel system modelling chemotaxis.We prove that there is a critical number which determines theoccurrence of blowup in the two-dimensional case for 1 <p < 2. In three- or higher-dimensional cases, we show thatthe radial symmetrical solution will blow up if 1 < p <N/N–2 (N 3) for non-negative initial value.     

Semi-discrete central-upwind schemes with reduced dissipation for Hamilton-Jacobi equations

We introduce a new family of Godunov-type semi-discrete centralschemes for multidimensional Hamilton–Jacobi equations.These schemes are a less dissipative generalization of the central-upwindschemes that have been recently proposed in Kurganov, Noelleand Petrova (2001, SIAM J. Sci. Comput., 23, pp. 707–740).We provide the details of the new family of methods in one,two, and three space dimensions, and then verify their expectedlow-dissipative property in a variety of examples.     

Pattern formation in a 2D simple chemical system with general orders of autocatalysis and decay

In this paper, we investigate pattern formation in a coupledsystem of reaction–diffusion equations in two spatialdimensions. These equations arise as a model of isothermal chemicalautocatalysis with termination in which the orders of autocatalysisand termination, m and n, respectively, are such that 1 <n < m. We build on the preliminary work by Leach & Wei(2003, Physica D, 180, 185–209) for this coupled systemin one spatial dimension, by presenting rigorous stability analysisand detailed numerical simulations for the coupled system intwo spatial dimensions. We demonstrate that spotty patternsare observed over a wide parameter range.     

An improved error bound for a finite element approximation of a model for phase separation of a multi-component alloy with a concentration dependent mobility matrix

Summary. Using a slightly different discretization scheme in time and adapting the approach in Nochetto et al. (1998) for analysing the time discretization error in the backward Euler method, we improve on the error bounds derived in (i) Barrett and Blowley (1998) and (ii) Barrett and Blowey (1999c) for a fully practical piecewise linear finite element approximation of a model for phase separation of a multi-component alloy with a concentration dependent mobility matrix and (i) a logarithmic free energy, and (ii) a non-smooth free energy (the deep quench limit); respectively. Moreover, the improved error bound in the deep quench limit is optimal. Numerical experiments with three components illustrating the above error bounds are also presented. Received June 28, 1999 / Revised version received December 3, 1999 / Published online November 8, 2000     

Sharp error estimates for discretizations of the 1D convection-diffusion equation with Dirac initial data

This paper derives sharp estimates of the error arising fromexplicit and implicit approximations of the constant-coefficient1D convection–diffusion equation with Dirac initial data.The error analysis is based on Fourier analysis and asymptoticapproximation of the integrals resulting from the inverse Fouriertransform. This research is motivated by applications in computationalfinance and the desire to prove convergence of approximationsto adjoint partial differential equations.     

Error estimates for a semi-implicit numerical scheme solving the Landau-Lifshitz equation with an exchange field

^** Email: Ivan.Cimrak{at}ugent.be We study the Landau–Lifshitz (LL) equation describingthe evolution of spin fields in continuum ferromagnets. We considerthe 3D case when the effective field arising in the LL equationincludes exchange interaction, the most challenging case. Thissetting corresponds to the pure isotropic case without a demagnetizingfield. We derive some regularity results for the exact solutionto the LL with Neumann-type boundary conditions. We modify thenumerical scheme studied by A. Prohl in two dimensions and weprove error estimates for this scheme in three dimensions.     

Asymptotic expansions of the generalized Epstein-Hubbell integral

The generalized Epstein–Hubbell integral recently introducedby Kalla & Tuan (Comput. Math. Applic. 32, 1996) is consideredfor values of the variable k close to its upper limit k = 1.Distributional approach is used for deriving two convergentexpansions of this integral in increasing powers of 1 –k². For certain values of the parameters, one of these expansionsinvolves also a logarithmic term in the asymptotic variable1 – k². Coefficients of these expansions are given interms of the Appell function and its derivative. All the expansionsare accompanied by an error bound at any order of the approximation.Numerical experiments show that this bound is considerably accurate.     

The Normal Holonomy Group of Kahler Submanifolds

We study the (restricted) holonomy group Hol() of the normalconnection (shortened to normal holonomy group) of a Kählersubmanifold of a complex space form. We prove that if the normalholonomy group acts irreducibly on the normal space then itis linear isomorphic to the holonomy group of an irreducibleHermitian symmetric space. In particular, it is a compact groupand the complex structure J belongs to its Lie algebra. We prove that the normal holonomy group acts irreducibly ifthe submanifold is full (that is, it is not contained in a totallygeodesic proper Kähler submanifold) and the second fundamentalform at some point has no kernel. For example, a Kähler–Einsteinsubmanifold of CPⁿ has this property. We define a new invariant µ of a Kähler submanifoldof a complex space form. For non-full submanifolds, the invariantµ measures the deviation of J from belonging to the normalholonomy algebra. For a Kähler–Einstein submanifold,the invariant µ is a rational function of the Einsteinconstant. By using the invariant µ, we prove that thenormal holonomy group of a not necessarily full Kähler–Einsteinsubmanifold of CPⁿ is compact, and we give a list of possibleholonomy groups. The approach is based on a definition of the holonomy algebrahol(P) of an arbitrary curvature tensor field P on a vectorbundle with a connection and on a De Rham type decompositiontheorem for hol(P). 2000 Mathematics Subject Classification53C40 (primary), 53B25 (secondary).     

A Petrov-Galerkin method with quadrature for elliptic boundary value problems

We propose and analyse a fully discrete Petrov–Galerkinmethod with quadrature, for solving second-order, variable coefficient,elliptic boundary value problems on rectangular domains. Inour scheme, the trial space consists of C² splines of degreer 3, the test space consists of C⁰ splines of degree r –2, and we use composite (r – 1)-point Gauss quadrature.We show existence and uniqueness of the approximate solutionand establish optimal order error bounds in H², H¹ and L² norms.     

Parameter-free H(div) preconditioning for a mixed finite element formulation of diffusion problems

^** Email: c.powell{at}manchester.ac.uk Mixed finite element formulations of generalised diffusion problemsyield linear systems with ill-conditioned, symmetric and indefinitecoefficient matrices. Preconditioners with optimal work complexitythat do not rely on artificial parameters are essential. Weimplement lowest order Raviart–Thomas elements and analysepractical issues associated with so-called ‘H(div) preconditioning’.Properties of the exact scheme are discussed in Powell &Silvester (2003, SIAM J. Matrix Anal. Appl., 25, 718–738).We extend the discussion, here, to practical implementation,the components of which are any available multilevel solverfor a weighted H(div) operator and a pressure mass matrix. Anew bound is established for the eigenvalue spectrum of thepreconditioned system matrix and extensive numerical resultsare presented.     

A local error analysis of the boundary-concentrated hp-FEM

^** Email: teibner{at}mathematik.tu-chemnitz.de^*** Email: melenk{at}tuwien.ac.at The boundary-concentrated finite-element method (FEM) is a variantof the hp-version of the FEM that is particularly suited forthe numerical treatment of elliptic boundary value problemswith smooth coefficients and boundary conditions with low regularityor non-smooth geometries. In this paper, we consider the caseof the discretization of a Dirichlet problem with the exactsolution u H¹⁺() and investigate the local error in variousnorms. For 2D problems, we show that the error measured in thesenorms is O(N^––ß), where N denotes thedimension of the underlying finite-element space and ß> 0. Furthermore, we present a new Gauss–Lobatto-basedinterpolation operator that is adapted to the case of non-uniformpolynomial degree distributions.     

Finite element approximation of a model for phase separation of a multi-component alloy with a concentration-dependent mobility matrix

We consider a model for phase separation of a multi-componentalloy with a concentration-dependent mobility matrix and logarithmicfree energy. In particular we prove that there exists a uniquesolution for sufficiently smooth initial data. Further, we provean error bound for a fully practical piecewise linear finiteelement approximation in one and two space dimensions. Finallynumerical experiments with three components in one space dimensionare presented.     

Error expansion for a first-order upwind difference scheme applied to a model convection-diffusion problem

A singularly perturbed convection–diffusion problem isconsidered. The problem is discretized using a simple first-orderupwind difference scheme on general meshes. We derive an expansionof the error of the scheme that enables uniform error boundswith respect to the perturbation parameter in the discrete maximumnorm for both a defect correction method and the Richardsonextrapolation technique. This generalizes and simplifies resultsobtained in earlier publications by Fröhner et al.(2001,Numer. Algorithms, 26, 281–299) and by Natividad &Stynes (2003, Appl. Numer. Math., 45, 315–329). Numericalexperiments complement our theoretical results.     

Attractor for lattice system of dissipative Zakharov equation

We consider the asymptotic behavior of solutions of an infinite lattice dynamical system of dissipative Zakharov equation. By introducing new weight inner product and norm in the space and establishing uniform estimate on ＂Tail End＂ of solutions, we overcome some difficulties caused by the lack of Sobolev compact embedding under infinite lattice system, and prove the existence of the global attractor; then by using element decomposition and the covering property of a polyhedron in the finite-dimensional space, we obtain an upper bound for the Kolmogorov ε-entropy of the global attractor; finally, we present the upper semicontinuity of the global attractor.     

Crystal Bases for Quantum Generalized Kac-Moody Algebras

In this paper, we develop the crystal basis theory for quantumgeneralized Kac–Moody algebras. For a quantum generalizedKac–Moody algebra U_q(g), we first introduce the categoryO_int of U_q(g)-modules and prove its semisimplicity. Next, wedefine the notion of crystal bases for U_q(g)-modules in thecategory O_int and for the subalgebra . We then prove the tensor product rule and the existence theoremfor crystal bases. Finally, we construct the global bases forU_q(g)-modules in the category O_int and for the subalgebra . 2000 Mathematics Subject Classification17B37, 17B67.     

On Nordlander’s conjecture in the three-dimensional case

In the present paper we prove, that in the real normed space X, having at least three dimensions, the Nordlander’s conjecture about the modulus of convexity of the space X is true, i.e. from the validity of Day’s inequality for a fixed real number from the interval (0,2), follows that X is an inner product space.