This article surveys a general approach to error control and adaptive mesh design in Galerkin finite element methods that is based on duality principles as used in optimal control. Most of the existing work on a posteriori error analysis deals with error estimation in global norms like the ‘energy norm’ or the L2 norm, involving usually unknown ‘stability constants’. However, in most applications, the error in a global norm does not provide useful bounds for the errors in the quantities of real physical interest. Further, their sensitivity to local error sources is not properly represented by global stability constants. These deficiencies are overcome by employing duality techniques, as is common in a priori error analysis of finite element methods, and replacing the global stability constants by computationally obtained local sensitivity factors. Combining this with Galerkin orthogonality, a posteriori estimates can be derived directly for the error in the target quantity. In these estimates local residuals of the computed solution are multiplied by weights which measure the dependence of the error on the local residuals. Those, in turn, can be controlled by locally refining or coarsening the computational mesh. The weights are obtained by approximately solving a linear adjoint problem. The resulting a posteriori error estimates provide the basis of a feedback process for successively constructing economical meshes and corresponding error bounds tailored to the particular goal of the computation. This approach, called the ‘dual-weighted-residual method’, is introduced initially within an abstract functional analytic setting, and is then developed in detail for several model situations featuring the characteristic properties of elliptic, parabolic and hyperbolic problems. After having discussed the basic properties of duality-based adaptivity, we demonstrate the potential of this approach by presenting a selection of results obtained for practical test cases. These include problems from viscous fluid flow, chemically reactive flow, elasto-plasticity, radiative transfer, and optimal control. Throughout the paper, open theoretical and practical problems are stated together with references to the relevant literature.

The objective of this article is to lay down the proper mathematical foundations of the two-dimensional theory of linearly elastic shells. To this end, it provides, without any recourse to any a priori assumptions of a geometrical or mechanical nature, a mathematical justification of two-dimensional linear shell theories, by means of asymptotic methods, with the thickness as the ‘small’ parameter.

A major virtue of this approach is that it naturally leads to precise mathematical definitions of linearly elastic ‘membrane’ and ‘flexural’ shells. Another noteworthy feature is that it highlights in particular the role played by two fundamental tensors, each associated with a displacement field of the middle surface, the linearized change of metric and linearized change of curvature tensors.

More specifically, under fundamentally distinct sets of assumptions bearing on the geometry of the middle surface, on the boundary conditions, and on the order of magnitude of the applied forces, it is shown that the three-dimensional displacements, once properly scaled, converge (in H1, or in L2, or in ad hoc completions) as the thickness approaches zero towards a ‘two-dimensional’ limit that satisfies either the linear two-dimensional equations of a ‘membrane’ shell (themselves divided into two subclasses) or the linear two-dimensional equations of a ‘flexural’ shell. Note that this asymptotic analysis automatically provides in each case the ‘limit’ two-dimensional equations, together with the function space over which they are well-posed.

The linear two-dimensional shell equations that are most commonly used in numerical simulations, namely Koiter's equations, Naghdi's equations, and ‘shallow’ shell equations, are then carefully described, mathematically analysed, and likewise justified by means of asymptotic analyses.

The existence and uniqueness of solutions to each one of these linear two-dimensional shell equations are also established by means of crucial inequalities of Korn's type on surfaces, which are proved in detail at the beginning of the article.

This article serves as a mathematical basis for the numerically oriented companion article by Dominique Chapelle, also in this issue of Acta Numerica.

This article, a companion to the article by Philippe G. Ciarlet on the mathematical modelling of shells also in this issue of Acta Numerica, focuses on numerical issues raised by the analysis of shells.

Finite element procedures are widely used in engineering practice to analyse the behaviour of shell structures. However, the concept of ‘shell finite element’ is still somewhat fuzzy, as it may correspond to very different ideas and techniques in various actual implementations. In particular, a significant distinction can be made between shell elements that are obtained via the discretization of shell models, and shell elements – such as the general shell elements – derived from 3D formulations using some kinematic assumptions, without the use of any shell theory. Our first objective in this paper is to give a unified perspective of these two families of shell elements. This is expected to be very useful as it paves the way for further thorough mathematical analyses of shell elements. A particularly important motivation for this is the understanding and treatment of the deficiencies associated with the analysis of thin shells (among which is the locking phenomenon). We then survey these deficiencies, in the framework of the asymptotic behaviour of shell models. We conclude the article by giving some detailed guidelines to numerically assess the performance of shell finite elements when faced with these pathological phenomena, which is essential for the design of improved procedures.

The development of Krylov subspace methods for the solution of operator equations has shown that two basic construction principles underlie the most commonly used algorithms: the orthogonal residual (OR) and minimal residual (MR) approaches. It is shown that these can both be formulated as techniques for solving an approximation problem on a sequence of nested subspaces of a Hilbert space, an abstract problem not necessarily related to an operator equation. Essentially all Krylov subspace algorithms result when these subspaces form a Krylov sequence. The well-known relations among the iterates and residuals of MR/OR pairs are shown to hold also in this rather general setting. We further show that a common error analysis for these methods involving the canonical angles between subspaces allows many of the known residual and error bounds to be derived in a simple and consistent manner. An application of this analysis to compact perturbations of the identity shows that MR/OR pairs of Krylov subspace methods converge q-superlinearly when applied to such operator equations.

Methods for knowledge discovery in data bases (KDD) have been studied for more than a decade. New methods are required owing to the size and complexity of data collections in administration, business and science. They include procedures for data query and extraction, for data cleaning, data analysis, and methods of knowledge representation. The part of KDD dealing with the analysis of the data has been termed data mining. Common data mining tasks include the induction of association rules, the discovery of functional relationships (classification and regression) and the exploration of groups of similar data objects in clustering. This review provides a discussion of and pointers to efficient algorithms for the common data mining tasks in a mathematical framework. Because of the size and complexity of the data sets, efficient algorithms and often crude approximations play an important role.

This paper gives a review of integration algorithms for finite dimensional mechanical systems that are based on discrete variational principles. The variational technique gives a unified treatment of many symplectic schemes, including those of higher order, as well as a natural treatment of the discrete Noether theorem. The approach also allows us to include forces, dissipation and constraints in a natural way. Amongst the many specific schemes treated as examples, the Verlet, SHAKE, RATTLE, Newmark, and the symplectic partitioned Runge–Kutta schemes are presented.

Optimization problems in which the variable is not a vector but a symmetric matrix which is required to be positive semidefinite have been intensely studied in the last ten years. Part of the reason for the interest stems from the applicability of such problems to such diverse areas as designing the strongest column, checking the stability of a differential inclusion, and obtaining tight bounds for hard combinatorial optimization problems. Part also derives from great advances in our ability to solve such problems efficiently in theory and in practice (perhaps ‘or’ would be more appropriate: the most effective computational methods are not always provably efficient in theory, and vice versa). Here we describe this class of optimization problems, give a number of examples demonstrating its significance, outline its duality theory, and discuss algorithms for solving such problems.