No label defined (Q4337697)
From MaRDI portal
!
This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:
scientific article; zbMATH DE number 1013030
| Language | Label | Description | Also known as |
|---|---|---|---|
| default for all languages | No label defined |
||
| English | No label defined |
scientific article; zbMATH DE number 1013030 |
Statements
26 May 1997
0 references
formal first-order differential operators
0 references
formal adjoints
0 references
Helmholtz type decompositions
0 references
Weyl decompositions
0 references
Lyapunov-Schmidt splitting method
0 references
variational problems of continuum mechanics
0 references
0.8938766
0 references
0.8626417
0 references
0.8569778
0 references
0.85671496
0 references
0.85558784
0 references
This monograph is motivated by a perturbation theoretical perspective on nonlinear partial differential equations and provides an interesting point of view for their discussion. The aim is to study the question of describing the set of solutions of a nonlinear equation that lie in some sense in a sufficiently small neighborhood of a known solution. It is natural to study linearizations of the nonlinear problem in order to analyze neighboring solutions. Following the notation of the monograph, let \(\dot y(x)\) denote the derivative of a mapping \(y: \omega \subseteq \mathbb{R}^m \to\mathbb{R}^n\), \(m\leq n\), \(m,n\in \mathbb{Z}^+\), \(\omega\) a bounded domain in \(\mathbb{R}^m\) with sufficiently smooth boundary \(\partial\omega\), at a point \(x\in \omega\). To the general nonlinear differential expression \(F(\dot y(x),y (x),x)\) for \(x\in\omega\) belongs (under mild assumptions on \(F)\) a linearization of the form \[ Lu= \text{div} (A^*u) + {\mathbf a} \cdot u, \] where \(A\) is an \(n\times m\) matrix-valued function and \({\mathbf a}\) an \(n\)-dimensional vector field on \(\omega\). Consequently, \(u\) must also be an \(n\)-dimensional vector field on \(\omega\). Using the integral over \(\omega\) \[ \bigl(u\mid v\bigr)= \int_\omega u(x) \cdot v(x)dx, \] of the Euclidean inner product \(u\cdot v\) of two such vector fields \(u,v\) as a Hilbert space inner product for such objects, we obtain as a formal adjoint to \(L\) the operator \(L^*\) given by \(L^*p= -A \text{grad} p+p {\mathbf a}\), where \(p\) denotes a regular function in \(\omega\). Clearly, \(L\), \(L^*\) can be considered as generalizations of the vector-analytic differential operators div and -grad (the first part of this observation motivates the title reference to the operator div). The natural question arises, how the remaining vector-analytical operator curl could be implemented in this generalization. This is achieved by characterizing formal first-order differential operators \(R\) of the form \[ R\varphi= \sum^m_{k=1} C_{kj} \partial_j \varphi+ C\varphi, \] \(\varphi\) an \(\ell\)-dimensional vector field on \(\omega\), satisfying the condition \(LR\equiv 0\) on smooth \(\ell\)-dimensional vector fields, the corresponding generalization to the well-known property \(\text{div rot} \equiv 0\). The corresponding generalization for \(\text{curl grad} \equiv 0\), i.e. the relation \(R^*L^* \equiv 0\), then follows for smooth vector fields by taking formal adjoints. (In the light of this observation one might be tempted to suggest to also include grad and curl in the title of the monograph). Properties of \(L\), \(L^*\), \(R\), the generalized div, -grad, curl, are then discussed in detail in the first chapter of the monograph. In particular, the well-known, classical Helmholtz type decompositions of vector fields into gradient fields and solenoidal vector fields, which can in fact be considered as special instances of the orthogonal decomposition into null space of an operator and the (closed) range of its adjoint, are generalized to decompositions into null space of \(L\) and (closed) range of \(L^*\) and, respectively, into null space of \(L^*\) and (closed) range of \(L\), by imposing suitable boundary conditions. (These decompositions are called Weyl decompositions in the text). The construction of these decompositions is, however, not done in the way indicated in the above but rather through explicit constructions via solving boundary value problems for the operator \(L\) and more generally in \(L_q\)-Sobolev-type spaces (with \(1<q <\infty)\). Some details of the construction of the required function spaces is deferred to an appendix. The second chapter, which is in fact the last chapter of the monograph, then reconnects the linear results of chapter 1 with the original nonlinear problem by considering the original problem as a nonlinear perturbation of the operator \(L\). The discussion of solutions close to a given one is in this context naturally based on the implicit function theorem together with the Lyapunov-Schmidt splitting method. The remainder of this chapter discusses particular examples and specifically applications to the issue of giving sufficient conditions for local minima for variational problems of continuum mechanics with nonlinear partial differential equations of the above type as constraints. The multiplicative inequalities needed in the analysis are also collected in the appendix.
0 references