Multi-scale analysis. Modeling and simulation. (Q2485247)

This book is a course which introduces the reader into the problematics of multi-scale systems from the point of view of applied mathematics. The author underlines the fact that the mathematician who analyses multi-scale time and space problems has the following possibilities: to simulate the system with very efficient techniques; to separate the low-scale problems from the high-scale problems; to manage differently in the system the high-scale problematics and the low-scale problematics. The theme of the course is presented through several examples, most of them coming from the mechanics of solids, from physics, chemistry (polymeric fluids, chemical kinetics etc). The course contains six chapters, each accompanied by a specific selective bibliography. At the end of the book a complete list of references is given, containing 49 authors, the quotes being from after 1990. In chapter 1, entitled ``Micro-macro models for solids'', a strategy is presented allowing the simulation of deforming solid body by coupling the microscopic description of the matter deformation and strongly deformed regions with the classical description in terms of deformable continuous media. A logic for changing the scale is presented here, through an example which suggests information at macroscopic level starting from the microscopic level. The elements of continuum mechanics are analysed, and the equilibrium elasticity equations are given together with optimal equilibrium equations (Eceter-Lagrange equations). The atomic scale of elastic energy is presented by choosing as example a perfect periodic crystal (simplifying hypothesis). The atomic energy is calculated in nine examples. In this chapter, the deformation in the case of fracture, is also debated. In the area close to the fracture it is impossible to use only the macroscopic scale. The areas around the fracture are treated differently, analysing the deformations. The area of weak gradients (where a macroscopic model is enough) is different from the area of strong gradients, where a microscopic model is necessary. In this chapter, notions of weak topology are defined together with their link to calculus of variations. The introduced notions of weak convergence are followed by exercises. The chapter closes with a bibliography which allows a thorough study of each paragraph. In chapter 2, ``The technique of homogenising'', mathematical analysing techniques are presented, as well as numerical techniques (numerical homogenization). The author starts with a one-dimensional boundary problem. The topology of weak convergence provides mathematical formulations. The canonical example is the equation \[ -\frac{d}{dx}\left(a\left(\frac{x}{\epsilon}\right)\frac {d}{dx}u^\epsilon\right)=f, \] where function \(a\left(\frac{x}{\epsilon}\right)\) is a periodic function of small period \(\epsilon\). First of all, the limit case is considered, when \(\epsilon\to0\). The same method is also used in the limit (boundary) conditions. The two-dimensional case is an extension of the one-dimensional case for lamellated materials. Differently from the most complicated cases, the convergence in two scales is proved only in a particular periodic case. General considerations are exposed, referring to multi-scale methods. In the case when small scales interfere uniformly in the entire domain, the coefficient is periodically oscillating. The interference of small scales is sensitive only along the walls of the domain (example: the boundary layer in the problems of fluid mechanics). In this case, the fluid thickness plays a role only in the wall area. Far away from the wall, the fluid is considered to be perfect. The situation of eliminating the boundary layer is approached through two examples. The strategy consists in replacing the equation of the boundary layer with another one, close (using Taylor formula) and with a new boundary. At the end, a molecular simulation is analysed, in the static and in the dynamic case. Numerical models are exposed. Chapter 3 is entitled ``The molecular simulation''. In this chapter, the molecular systems are shaped by the quantum chemistry. Schrödinger equation is replaced in practice by numerical approximations. Models of the following form are presented: the discretization of Hartree-Fock problem; the discretization of Newtonian dynamics; methods of accelerating the molecular dynamics simulations. The modelling of the phase is analysed by considering a continuous model with its numerical solution. The chapter ends with some basic results on integral equations. The references from the end of the chapter complete the possible questions in this case. Chapter 4, ``Micro-macro models for fluids'', begins with an exposure of some elements of incompressible fluids mechanics, with referring to polymeric fluids. At each point of the fluid the strain tensor is calculated, considering the presence of millions of polymeric chains. The Navier-Stokes equations are valid. Numerical simulations are presented for Couette flow. Some elements of numerical analysis by using discretization schemes for stochastic equations are presented at the end of this chapter. Convergence notions are defined for the simulation of stochastic differential equations, as well as of the stability domains (moderate and asymptotic). Chapter 5, ``Chemical kinetics''. After introducing some notions of chemical kinetics, as an example of multi-scale system in time, efficient numerical schemes are adopted here. The numerical scheme is decomposed in sub-stages for each scale of time. The reduction of the multi-scale system consists in eliminating the shortest time intervals. The model differential equation is a Cauchy problem defined on an interval from \(\mathbb R\), with values in \(\mathbb R^m\). Some methods of solving Euler equations by explicit and implicit schemes are briefly described. The stability and convergence of Cauchy problem's solution are analysed. The stiff Cauchy problem is defined and illustrated with examples for different time intervals. The systems analysed in this chapter are of the following form: \[ \begin{cases} \frac{dy_1}{dt}=-k_1y_1+k_2y_2,\\ \frac{dy_2}{dt}=k_1y_1-k_2y_2+k_3y_3,\\ \frac{dy_3}{dt}=-k_3y_3, \end{cases} \] where \(k_i\) are constants of different reactions. The author gives also some methods of decomposing operators and reducing dynamic systems. In chapter 6, ``Towards an approximation unit'', the author summarizes a synthetic vision of the models considered in this course, but also in worldwide research. A classification of the encountered problems is made, and a unit of approximations done is found. Notions are specified, such as: micro, local, macro, global, very useful to researchers. At the end, the author's reasons referring to the change of scale are exposed (problems which depend on time, problems which start from atomic scale, problems derived from material dynamics, etc.).