Workshop Programme
for period 26  30 March 2007
The Theory of Highly Oscillatory Problems
26  30 March 2007
Timetable
Monday 26 March  
08:3010:00  Registration  
10:0011:00  Berry, M (Bristol)  
Three recent results on asymptotics of oscillations  Sem 1  
The results are separate, and apparently paradoxical, and have implications for physics. First, when two exponentials compete, their interference can be dominated by the contribution with smaller exponent. Second, repeated differentiation of almost all functions in a wide class generates trigonometric oscillations (‘almost all functions tend to cosx’). Third, it is possible to find bandlimited functions that oscillate arbitrarily faster than their fastest Fourier component (‘superoscillations’). 

11:0011:30  Coffee  
11:3012:15  Engquist, B (Texas at Austin)  
Heterogeneous multiscale methods  Sem 1  
Continuum simulations of solids or fluids for which some atomistic information is needed are typical example of multiscale problems with very large ranges of scales. For such problems it is necessary to restrict the simulations on the microscale to a smaller subset of the full computational domain. The heterogeneous multiscale method is a framework for developing and analyzing numerical methods that couple computations from very different scales. Local microscale simulations on small domains supply missing data to a macroscale simulation on the full domain. Examples are local molecular dynamics computations that produce data to a continuum macroscale model, or a highly oscillatory dynamical system for which a local estimate of resonances is enough to supply data for a smoother evolution of averages. We will focus on dynamical systems. 

12:3013:30  Lunch at Wolfson Court  
14:1515:00  MacKay, RS (Warwick)  
Hamiltonian slow manifolds with internal oscillation  Sem 1  
Many Hamiltonian systems possess special families of solutions which can be described approximately as slowly drifting periodic orbits. Examples include the gravitational threebody problem, the interaction of two identical charged particles in a magnetic field, and the propagation and interaction of discrete breathers (timeperiodic spatially localised motions in Hamiltonian networks of units). Theory will be presented for how to improve a zeroth order manifold of approximate solutions to rth order for any r > 0, meaning one that contains all nearby periodic orbits of nearby period and has error of order the rth power of the drift field (even with a small constant for r=1). In the normally hyperbolic case an exactly invariant nearby submanifold can be constructed. If there are normally elliptic directions, however, this is impossible in general but the above rth order approximations can be achieved provided that the normal frequencies avoid all multiples of that of the approximately periodic motion and are fast compared with the drift. An effective Hamitlonian is derived to describe the drift of the orbits. Applications to the above fields will be given. An introduction has been published as R.S.MacKay, Slow Manifolds, in Energy localisation and transfer, eds T.Dauxois, A.LitvakHinenzon, R.S.MacKay, A.Spanoudaki (World Sci Publ Co, 2004), 149192. 

15:0015:30  Tea  
15:3016:15  Ainsworth, M (Strathclyde)  
Dispersive and dissipative behaviour of Galerkin approximation using high order polynomials  Sem 1  
We consider the dispersive properties of Galerkin finite element methods for wave propagation. The dispersive properties of conforming finite element scheme are analysed in the setting of the Helmholtz equation and an explicit form the discrete dispersion relation is obtained for elements of arbitrary order. It is shown that the numerical dispersion displays three different types of behaviour depending on the order of the polynomials used relative to the meshsize and the wave number. Quantitative estimates are obtained for the behaviour and rates of decay of the dispersion error in the differing regimes. 

16:1517:00  Dobrokhotov, S (Russian Academy of Sciences)  
Generalization of the Maslov Theory for Localized Asymptotics and Tsunami waves  Sem 1  
We suggest a new asymptotic representation for the solutions to the multidimensional wave equations with variable velocity with localized initial data. This representation is the generalization of the Maslov canonical operator based also on a simple relationship between fast decaying and fast oscillating solutions, and on boundary layer ideas. It establishes the connection between initial localized perturbations and wave profiles near the wave fronts including the neighborhood of backtracking (focal or turning) and self intersection points. We show that wave profiles are related with a form of initial sources and also with the Lagrangian manifolds organized by the rays and wavefronts. In particular we discuss the influence of such topological characteristics like the Maslov and Morse indices to metamorphosis of the profiles after crossing the focal points. We apply these formulas to the problem of a propagation of tsunami waves in the frame of socalled ``piston model''. Finally we suggest a fast asymptoticallynumerical algorithm for simulation of tsunami wave over nonuniform bottom. Different scenarios of the distribution of the waves are considered, the wave profiles of the front are obtained in connection with the different shapes of the source and with the diverse rays generating the fronts. It is possible to use suggested algorithm to predict in real time the zones of the beaches where the amplitude of the tsunami wave has dangerous high values. In this connection we also discuss the following questions: the problems of the regularization of the wave field near focal points; illpossed problems appearing in the geometry of the wavefronts; the inverse problem connected with the possibility of reconstruction of the source via the measurement of the tsunami wave profile on the shelf etc. This work was done together with S.SekerzhZenkovich, B.Tirozzi, B.Volkov and was partially supported by RFBR grant N 050100968 and Agreement Between University "La Sapienza", Rome and Institute for Problems in Mechanics RAS, Moscow. Bibliography [1] S.Yu. Dobrokhotov, S.Ya SekerzhZenkovich, B. Tirozzi, T.Ya. Tudorovskiy, The description of tsunami waves propagation based on the Maslov canonical operator, Doklady Mathematics, 2006, v.74, N 1, pp. 592596 [2] S.Yu. Dobrokhotov, S.Ya SekerzhZenkovich, B. Tirozzi, T.Ya. Tudorovskiy, Asymptotic theory of tsunami waves: geometrical aspects and the generalized Maslov representation, Publications of Kyoto Research Mathematical Institute, Vol.4, page 118153, 2006, ISSN 18802818. [3] S.Dobrokhotov, S.SekerzhZenkovich, B.Tirozzi, B.Volkov Explicit asymptotics for tsunami waves in framework of the piston model, Russ. Journ. Earth Sciences, 2006, v.8, ES403, pp.112 [4] S.Dobrokhotov, S.Sinitsyn, B.Tirozzi, Asymptotics of Localized Solutions of the OneDimensional Wave Equation with Variable Velocity. I. The Cauchy Problem, Russ. Jour.Math.Phys., v.14, N1, 2007, pp.2856 

17:0017:30  Klein, C (MPI)  
Numerical study of oscillatory regimes in the KadomtsevPetviashvili equation  Sem 1  
The aim of this talk is the accurate numerical study of the KP equation. In particular we are concerned with the small dispersion limit of this model, where no comprehensive analytical description exists so far. To this end we first study a similar highly oscillatory regime for asymptotically small solutions, which can be described via the DaveyStewartson system. In a second step we investigate numerically the small dispersion limit of the KP model in the case of large amplitudes. Similarities and differences to the much better studied Kortewegde Vries situation are discussed as well as the dependence of the limit on the additional transverse coordinate. Related Links 

17:3018:30  Welcome Wine Reception  
18:4519:30  Dinner at Wolfson Court (Residents only) 
Tuesday 27 March  
Chair: A Iserles  
09:0009:45  Lubich, C (Tuebingen)  
Numerical oscillations over long times  Sem 1  
09:4510:30  Hairer, E (Geneve)  
Conservation of energy and actions in numerical discretizations of nonlinear wave equations  Sem 1  
For numerical discretizations of nonlinearly perturbed wave equations the longtime nearconservation of energy, momentum, and harmonic actions is studied. The proofs are based on the technique of modulated Fourier expansions in time. Rigorous statements on the longtime conservation properties are shown under suitable numerical nonresonance conditions and under a CFL condition. The time step need not be small compared to the inverse of the largest frequency in the spacediscretized system. This is joint work with Christian Lubich and David Cohen. 

10:3011:00  Cohen, D (NTNU)  
Highly oscillatory Hamiltonian systems with nonconstant mass matrix  Sem 1  
We will present a class of numerical methods (based on the trigonometric methods) for such Hamiltonian problems. We will then present a frequency expansion of the numerical solution: the modulated Fourier expansion. The system that determines the coefficients of this expansion has two formal invariants which are related to the total energy and the oscillatory energy of the original system. This allows us to prove the nearconservation of the total and the oscillatory energy for the numerical schemes over very long time intervals. Related Links


11:0011:30  Coffee  
11:3012:15  Reich, S (Potsdam)  
Highly oscillatory PDEs, slow manifolds and regularized PDE formations  Sem 1  
The main motivation of my talk is provided by geophysical fluid dynamics. The underlying Euler or NavierStokes equations display oscillatory wave dynamics on a wide range of temporal and spatial scales. Simplified models are often based on the idea of balance and the concept of a slow manifold. Examples are provided by hydrostatic and geostrophic balance. One would also like to exploit these concepts on a computational level. However, slow manifolds are idealized objects that do not fully characterize the complex fluid behavior. I will describe a novel regularization technique that makes use of balance and slow manifolds in an adaptive manner. The regularization approach is based on a reinterpretation of the (linearly) implicit midpoint rule as an explicit timestepping method applied to a regularized set of Euler equations. Adaptivity can be achieved by means of a predictorcorrector interpretation of the regularization. 

12:3013:30  Lunch at Wolfson Court  
14:1515:00  Sjostrand, J (Ecole Polytechnique)  
Tunnel effect for KramersFokkerPlanck type operators  Sem 1  
For a large class of KramersFokkerPlanck type operators, we determine in the semiclassical (here the low temperature) limit the full asymptotic expansion of the splitting between the lowest eigenvalue (0) and the next one. In a previous work with Herau and C. Stolk we did so in the case when a certain potential (or exponent of a Maxwellian) has preciely one local minimum and then the splitting is "large". In this new work we treat the case when the potential has two local minima. Then the splitting is exponentially small and related to a tunnel effect between the minima. Our most direct source of inspiration has been works by HerauF.Nier, B.HelfferNier, but our methods are quite different. Joint work with F Herau and M Hitrik. 

15:0015:30  Tea  
15:3016:15  Mahalov, A (Arizona State)  
Fast singular oscillating limits of the 3D NavierStokes and Euler equations: Global regularity and threedimensional Euler dynamics  Sem 1  
In the first part of the talk, we study highly oscillatory problems for the incompressible 3D NavierStokes and prove existence on infinite time intervals of regular strong solutions; smoothness assumptions for initial data are the same as in local existence theorems. There are no conditional assumptions on the properties of solutions at later times, nor are the global solutions close to any 2D manifold. The global existence is proven using techniques of fast singular oscillating limits, Lemmas on restricted convolutions and the LittlewoodPaley dyadic decomposition. In the second part of the talk, we analyze regularity and dynamics of highly oscillatory problems for the 3D Euler equations. Detailed proofs can be found in the following references: Asymptotic Analysis, vol. 15, No. 2, p. 103150, 1997 (with A. Babin and B. Nicolaenko). Indiana University Mathematics Journal, vol. 48, No. 3, p. 11331176, 1999 (with A. Babin and B. Nicolaenko). Indiana University Mathematics Journal, vol. 50, p. 135, 2001 (with A. Babin and B. Nicolaenko). Russ. Mathematical Surveys, vol. 58, No. 2 (350), p. 287318, 2003 (with B. Nicolaenko). Methods and Applications of Analysis, vol. 11, No. 4, p. 605633, 2004 (with B. Nicolaenko, C. Bardos and F. Golse). Hokkaido Mathematical Journal, vol. 35, No. 2, p. 321364, 2006 (with Y. Giga, K. Inui and S. Matsui). Annals of Math. Studies, Princeton University Press (Editors: J. Bourgain and C. Kenig), In Press (with Y. Giga and B. Nicolaenko). 

16:1517:00  Quispel, R (La Trobe)  
Some remarks concerning exponential integrators  Sem 1  
TBA 

17:0017:30  Babych, N (Bath)  
Spectral results for membrane with perturbed stiffness and mass density  Sem 1  
We study the spectrum of a nonhomogeneous membrane that consists of two parts with strongly different stiffness and mass density. The small parameter describes the quotient of stiffness coefficients. The Mth power of the parameter is comparable to the ratio of densities. We show that the asymptotic behaviour of eigenvalues and eigenfunctions depends on rate M. We distinguish three cases M<1, M=1 and M>1. The strong resolvent convergence of perturbed operators leads to loss of the completeness of limit eigenfunction system in both cases when M is different from 1. Therefore the limit operators describe only a part of the prelimit membrane vibrations. With an eye to close this gap we use the WKBasymptotic expansions with a quantized small parameter to prove the existence of other kind of eigenvibrations, namely high frequency vibrations. In the critical case M=1 the limit operator is a nonselfadjoint one, nevertheless the perturbed operators are selfadjoint in a certain topology. The multiplicity of spectrum and structure of root spaces are investigated. Complete asymptotic expansions for eigenelements are constructed and justified in each case of the perturbations. 

18:4519:30  Dinner at Wolfson Court (Residents only) 
Wednesday 28 March  
Chair: B Engquist  
09:0009:45  Schuette, C (Free University of Berlin)  
Nanomechanics of biomolecules: whether oscillatory details are important or not  Sem 1  
09:4510:30  Hochbruck, M (HeinrichHeineUniversitaet, Dusseldorf)  
Exponential integrators for oscillatory secondorder differential equations  Sem 1  
In this talk, we analyse a family of exponential integrators for secondorder differential equations in which highfrequency oscillations in the solution are generated by a linear part. We characterise methods which allow secondorder error bounds by presenting a unified error analysis for the whole family of methods. A major advantage of our analysis is that it does not require bounds for pointwise products of matrices and therefore, generalises to abstract differential equations, where the linear part is an unbounded operator with infinitely many large eigenvalues directly. This is joint work with Volker Grimm 

10:3011:00  Grava, T (SISSA)  
Small dispersion limit of the KdV and semiclassical limit of focusing NLS  Sem 1  
We compare numerically the small dispersion limit of the Korteweg de Vries equation with the asymptotic formulas obtained in the works of LaxLevermore and DeifVenakides and Zhou. We compare numerically the semiclassical limit of the focusing nonlinear Schrodinger equation with the asymptotic formula obtained in the works of KamvissisMcLaughlinMiller and TovbisVenakidesZhou. As a results we outline the regions of the (x,t) where the above asymptotic formulas give a satisfactory description of the respective equations. 

11:0011:30  Coffee  
11:3012:15  Teufel, S (Tuebingen)  
Beyond the adiabatic approximation: exponentially small coupling terms  Sem 1  
For multilevel timedependent quantum systems one can construct superadiabatic representations in which the coupling between separated levels is exponentially small in the adiabatic limit. We explicitly determine the asymptotic behavior of the exponentially small coupling term for generic twostate systems with realsymmetric Hamiltonian. The superadiabatic coupling term takes a universal form and depends only on the location and the strength of the complex singularities of the adiabatic coupling function. First order perturbation theory for the Hamiltonian in the superadiabatic representation then allows to describe the timedevelopment of exponentially small adiabatic transitions and thus to rigorously confirm Michael Berry’s predictions on the universal form of adiabatic transition histories (Joint work with V. Betz from Warwick). Related Links


12:3013:30  Lunch at Wolfson Court  
14:0014:30  Wulff, C (Surrey)  
Approximate energy conservation for symplectic time semidiscretizations of semilinear wave equations  Sem 1  
We prove that Astable symplectic RungeKutta time semidiscretizations applied to semilinear wave equations with periodic boundary conditions, analytic nonlinearity and analytic initial data conserve a modified energy p to an exponentially small error. This modified energy is O(h^p)close to the original energy where p is the order of the method and h the timestepsize. Standard backward error analysis can not be applied because of the occurrence of unbounded operators in the construction of the modified vectorfield. This loss of regularity in the construction can be taken care of by projecting the PDE to a finitedimensional space and by coupling the number of excited modes as well as the number of terms in the expansion of the vectorfield with the stepsize. This way we obtain exponential estimates of the form O(\exp(C/ h^(1/2) )). A similar technique has been used for averaging of rapidly forced Hamiltonian PDEs by [Matthies and Scheel, 2003]. As a sideproduct, we also provide a convergence analysis of RungeKutta methods in Hilbert spaces. Related Links


14:3015:00  Olver, S (Cambridge)  
Numerical approximation of multivariate highly oscillatory integrals  Sem 1  
The aim of this talk is to describe several methods for numerically approximating the integral of a multivariate highly oscillatory function. We begin with a review of the asymptotic and Filontype methods developed by Iserles and Nørsett. Using a method developed by Levin as a point of departure we will construct a new method that uses the same information as a Filontype method, and obtains the same asymptotic order, while not requiring moments. This allows us to integrate over nonsimplicial domains, and with complicated oscillators. We also present a method for approximating oscillatory integrals with stationary points. 

15:0015:30  Tea  
15:3016:15  Boutet de Monvel, A (Paris 7)  
Asymptotics for the CamassaHolm equation  Sem 1  
I will present recent results on asymptotic behaviors for the Camassa–Holm (CH) equation ut  utxx + 2?ux + 3uux = 2uxuxx + uuxxx on the line, ? being a nonnegative parameter. Firstly, I will describe the longtime asymptotic behavior of the solution u? (x, t), ? > 0 of the initialvalue problem with fast decaying initial data u0(x). It appears that u? (x, t) behaves differently in different sectors of the (x, t)halfplane. Then I will analyse the behavior of u? (x, t) as ? 0. The methods are inverse scattering in a matrix RiemannHilbert approach and Deift and Zhou’s nonlinear steepest descent method. Work in collaboration with Dmitry Shepelsky. 

16:1517:00  ChandlerWilde, SN (Reading)  
Norms and condition numbers of oscillatory integral operators in acoustic scattering  Sem 1  
In this talk we discuss domain and boundary integral operators arising in the theory and numerical treatment by integral equation methods of the Helmholtz equation or time harmonic Maxwell equations. These integral operators are increasingly oscillatory as the wave number k increases (k proportional to the frequency of the time harmonic incident field). An interesting theoretical question, also of practical significance, is the dependence of the norms of these integral operators and their inverses on k. We investigate this question, in particular for classical single and doublelayer potential operators for the Helmholtz equation on the boundary of bounded Lipschitz domains. The results and techniques used depend on the domain geometry. In certain 2D cases (for example where the boundary is a starlike polygon) bounds which are sharp in their dependence on k can be obtained, but there are many open problems for more general geometries and higher dimension. Related Links


17:0017:30  Wirosoetisno, D (Durham)  
Exponential asymptotics for the primitive equations  Sem 1  
Following the work of Matthies (2001), it is shown how Gevrey (exponential) regularity of the solution and a classical method used together to prove an exponentiallyaccurate approximation result for a singular perturbation problem with a small parameter. The model considered is the viscous primitive equations of the ocean, although the method is applicable more generally. 

18:4519:30  Dinner at Wolfson Court (Residents only) 
Thursday 29 March  
Chair: E Hairer  
09:0009:45  Zuazua, E (Autonoma)  
Finite difference approximation of homogenization problems for elliptic equations  Sem 1  
In this talk, the problem of the approximation by finite differences of solutions to elliptic problems with rapidly oscillating coefficients and periodic boundary conditions will be discussed. The meshsize is denoted by $h$ while $\ee$ denotes the period of the rapidly oscillating coefficient. Using Bloch wave decompositions, we analyze the case where the ratio $h/\ee$ is rational. We show that if $h/\ee$ is kept fixed, being a rational number, even when $h,\ee\to 0$, the limit of the numerical solution does not coincide with the homogenized one obtained when passing to the limit as $\ee\to 0$ in the continuous problem. Explicit error estimates are given showing that, as the ratio $h/\ee$ approximates an irrational number, solutions of the finite difference approximation converge to the solutions of the homogenized elliptic equation. We consider both the 1d and the multidimensional case. Our analysis yields a quantitative version of previous results on numerical homogenization by Avellaneda, Hou and Papanicolaou, among others. This is a joint work with Rafael Orive 

09:4510:30  Venakides, S (Duke)  
Rigorous leading order asymptotic solutions of the semiclassical focusing NLS equation  Sem 1  
Solving an integrable nonlinear differential equation reduces to solving the matrix RiemannHilbert problem (RHP) that arises from the invese scattering procedure. Typically, this means finding a matrix m(z) that is analytic in the closed complex plane except on a given oriented contour on which m(z) has a jump discontinuity: the value of m(z) left of the contour equals its value on the right multiplied by a given "jump matrix'' V(z). The spacetime variables appear as parameters in the jump matrix; the complex variable z is the spectral variable of the linear differential operator that effects the linearization of the nonlinear system through the Lax pair associated with the system. We derive an explicit formula for the leading asymptotic (oscillatory) solution of the semiclassical focusing nonlinear Schroedinger equation for a class of initial data and prove an error estimate. We extend the analysis when time is large, the asymptotic formula is then expressed in terms of elementary functions. We first outline the notion of a Lax pair and how the RHP arises naturally.We then describe the steepest descent procedure that leads to an explicitly solvable asymptotic RHP through jump matrix factorizations and contour deformations. We also outline the gfunciton mechanism and describe how it determines the successful deformations. 

10:3011:00  Malham, SJA (HeriotWatt)  
Numerically evaluating the Evans function by Magnus integration  Sem 1  
We use Magnus methods to compute the Evans function for spectral problems as arise when determining the linear stability of travelling wave solutions to reactiondiffusion and related partial differential equations. In a typical application scenario, we need to repeatedly sample the solution to a system of linear nonautonomous ordinary differential equations for different values of one or more parameters as we detect and locate the zeros of the Evans function in the right half of the complex plane. In this situation, a substantial portion of the computational effortthe numerical evaluation of the iterated integrals which appear in the Magnus seriescan be performed independent of the parameters and hence needs to be done only once. More importantly, for any given tolerance Magnus integrators possess lower bounds on the step size which are uniform across large regions of parameter space and which can be estimated a priori. We demonstrate, analytically as well as through numerical experiment, that these features render Magnus integrators extremely robust and, depending on the regime of interest, efficient in comparison with standard ODE solvers. 

11:0011:30  Coffee  
11:3012:15  Efendiev, Y (Texas A and M)  
Multiscale numerical methods using limited global information  Sem 1  
In this talk, I will describe multiscale numerical methods for flows in heterogeneous porous media. The main idea of these methods is to construct local basis functions that can capture the small scale information when they are coupled via some global formulation. I will discuss the use of local boundary conditions, oversampling methods and the use of global information in constructing basis functions. Applications of these methods to stochastic equations will be also discussed. This is joint work with J. Aarnes, T. Hou, L. Jiang, V. Ginting. 

12:3013:30  Lunch at Wolfson Court  
14:0014:30  Matthews, P (Nottingham)  
Exponential integrators and functions of the matrix exponential  Sem 1  
Exponential integrators are the most efficient class of methods for the timestepping of stiff, semilinear, oscillatory PDEs such as the KdV equation. They solve the stiff, linear part of the PDE exactly. In the case of periodic boundary conditions, a Fourier spectral method can be used, so the linear part is diagonal and the methods can be applied straightforwardly. For other spatial discretizations, functions of the matrix exponential are required, which are susceptible to rounding errors. Several methods for evaluating these functions will be discussed. 

14:3015:00  Hoang, VH (Cambridge)  
High dimensional finite element methods for elliptic problems with highly oscillating coefficients  Sem 1  
Elliptic homogenization problems in a domain $\Omega \subset \R^d$ with $n+1$ separated scales are reduced to elliptic onescale problems in dimension $(n+1)d$. These onescale problems are discretized by a sparse tensor product finite element method (FEM). We prove that this sparse FEM has accuracy, work and memory requirement comparable to standard FEM for single scale problems in $\Omega$ while it gives numerical approximations of the correct homogenized limit as well as of all first order correctors, throughout the physical domain with performance independent of the physical problem's scale parameters. Numerical examples for model diffusion problems with two and three scales confirm our results. 

15:0015:30  Tea  
15:3016:15  Liu, H (Texas at Austin)  
A multiscale method for stiff ordinary differential equations with resonance  Sem 1  
We introduce a multiscale method to compute the effective behavior of a class of stiff and highly oscillatory ODEs. The oscillations may be in resonance with one another and thereby generate some hidden slow dynamics. Our method relies on correctly tracking a set of slow variables whose dynamics is effectively closed, and is sufficient to approximate the effective behavior of the ODEs. This set of variables is found by our numerical methods. We demonstrate our algorithms by a few examples that include a commonly studied problem of Fermi, Pasta, and Ulam (FPU). 

16:1517:00  Smyshlyaev, VP (Bath)  
Localization and propagation in high contrast highly oscillatory media via 'nonclassical' homogenization  Sem 1  
We discuss fundamental mathematical tools for analysis of localization and propagation effects in highly oscillatory media with high contrasts. With the underlying two small parameters of the oscillations and the contrast, there is a "critical" scaling when the phenomena at the micro and macro scales are coupled in a nontrivial way, with "unusual" effects observed in an asymptotically explicit way. The related mathematical tool is that of a "nonclassical" (highcontrast) homogenization, accounting for "highfrequency" oscillations, as opposes to the "classical" homogenization whose scope is limited by dealing in effect with low frequencies only. Those tools include "nonclassical" twoscale asymptotic expansions, twoscale operator and spectral convergence, and twoscale compactness (with the latter building on, among others, recent deep ideas of V.V. Zhikov [1]). We illustrate this on the problem of wave localization in high contrast periodic media with a defect [2] (a problem relevant to photonic crystal fibres). We also discuss the use of these techniques to the problem of "slowing down" of wave packets in high contrast highlyoscillatory media (the socalled "slow light" effect, with relevance to coupled resonances and metastability), and other prospects. Joint work with Ilia V. Kamotski. [1] V.V. Zhikov, On an extension of the method of twoscale convergence and its applications, (Russian) Mat. Sbornik 191 (2000), 3172; English translation in Sbornik Math. 191 (2000), 9731014; V.V. Zhikov, Gaps in the spectrum of some elliptic operators in divergent form with periodic coefficients. (Russian) Algebra i Analiz 16 (2004), 3458; [2] I.V. Kamotski and V.P. Smyshlyaev, Localised eigenstates due to defects in high contrast periodic media via homogenisation. BICS preprint 3/06. http://www.bath.ac.uk/mathsci/bics/preprints/BICS06_3.pdf (2006) Related Links


17:0017:30  Matthies, K (Bath)  
Exponential estimates in averaging and homogenisation  Sem 1  
Many partial differential equations with rapid spatial or temporal scales have effective descriptions which can be derived by homogenisation or averaging. In this talk we deal with examples, where quantitative estimates of the error is possible for higher order homogenisation and averaging. In particular, we provide theorems, which allow homogenisation and averaging beyond all orders by giving exponential estimates of appropriately averaged and homogenised descriptions. Methods include iterated averaging transformations, optimal truncation of asymptotic expansions and highly regular solutions (Gevrey regularity). Prototypical examples are reactiondiffusion equations with heterogeneous reaction terms or rapid external forcing, nonlinear Schr\"odinger equations describing dispersion management, and secondorder linear elliptic equations. Related Links


20:0018:00  Conference Dinner in the Dining Hall at Magdalene College 
Friday 30 March  
Chair: C Lubich  
09:0009:45  Iserles, A (Cambridge)  
From high oscillation to rapid approximation  Sem 1  
In this talk we address our recent work on the theory of modified Fourier expansions and their generalisation for the approximation of smooth functions in one and several variables. In particular, we debate rapid evaluation of expansion coefficients by means of Filonlike and GaussTuranlike quadrature techniques and various acceleration techniques. 

09:4510:30  Saut, JC (ParisSud)  
On the zeromass constraint for KP type equations  Sem 1  
For a rather general class of equations of KadomtsevPetviashvili (KP) type, we prove that the zeromass constraint (in x) is satisfied at any non zero time even if it is not satisfied at initial time zero. Our results are based on a precise analysis of the fundamental solution of the linear part and of its anti xderivative. 

10:3011:00  Langdon, S (Reading)  
High frequency scattering by convex polygons  Sem 1  
Standard boundary integral methods for high frequency scattering problems, with piecewise polynomial approximation spaces, suffer from the debilitating restriction that the number of degrees of freedom required to achieve a prescribed level of accuracy grows at least linearly with respect to the frequency. For problems of acoustic scattering by sound soft convex polygons, it has recently been demonstrated by ChandlerWilde and Langdon that by including plane wave basis functions supported on a graded mesh in the approximation space, with smaller elements closer to the corners of the polygon, convergence rates that depend only logarithmically on the frequency can be achieved. In this talk we review this approach, and consider some improvements and extensions to more complicated scattering problems. Related Links 

11:0011:30  Coffee  
11:3012:00  SanzSerna, JM (Valladolid)  
Mollified impulse methods revisited  Sem 1  
We introduce a family of impulselike methods for the integration of highly oscillatory secondorder differential equations whose forces can be split into a fast and a slow part. Methods of this family are specified by two weight functions: one is used to average positions and the other to mollify the force. In cases where the fast forces are conservative, the new family inlcudes as particular cases the mollified impulse methods introduced by GarcíaArchilla, Skeel and the present author. On the other hand the methods here extend to nonlinear fast forces a wellknown class of exponential integrators introduced by Hairer and Lubich. A convergence analysis will be presented that provides insight into the role played by the processes of averaging and mollification. A simple condition on the weight functions is shown to be both necessary and sufficient to avoid order reduction. 

12:3013:30  Lunch at Wolfson Court  
14:0014:30  Huybrechs, D  
On the localisation principle of high frequency scattering problems in integral equation discretisations  Sem 1  
The numerical simulation of scattering problems at moderate to high frequencies is a challenging problem. The discretization of any suitable mathematical model usually requires resolving the oscillations, which naturally leads to large and, in the context of integral equations, dense discretization matrices. Yet, high frequency scattering problems have a very local nature. The localization principle states that the reflection, refraction or diffraction of a wave that hits an obstacle is governed mainly by the geometry of the scattering object which is local to the point of contact. This principle is exploited by asymptotic methods, such as geometrical optics and the geometrical theory of diffraction. Suddenly, higher frequencies are desirable and lower frequencies become problematic. In this talk we examine how the localization principle can be exploited numerically in a more classical finite element setting. In particular, we may arrive at a sparse discretization matrix for integral equations by the use of Filontype quadrature rules for oscillatory integrals. We discuss the advantages and limitations of this approach and we examine the asymptotic nature of this sparse representation. Related Links


18:4519:30  Dinner at Wolfson Court (Residents only) 