Abstract
In the context of elasticity theory, rigidity theorems allow one to derive global properties of a deformation from local ones. This paper presents a new asymptotic version of rigidity, applicable to elastic bodies with sufficiently stiff components arranged into fine parallel layers. We show that strict global constraints of anisotropic nature occur in the limit of vanishing layer thickness, and give a characterization of the class of effective deformations. The optimality of the scaling relation between layer thickness and stiffness is confirmed by suitable bending constructions. Beyond its theoretical interest, this result constitutes a key ingredient for the homogenization of variational problems modeling high-contrast bilayered composite materials, where the common assumption of strict inclusion of one phase in the other is clearly not satisfied. We study a model inspired by hyperelasticity via \(\Gamma \)-convergence, for which we are able to give an explicit representation of the homogenized limit problem; it turns out to be of integral form with its density corresponding to a cell formula.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
Rigidity is a prevalent concept in different areas of mathematics. Generally speaking, it refers to powerful statements that allow one to draw far-reaching conclusions from seemingly little information, such as deducing global properties of a function from local ones. A classical result along these lines is often referred to as Liouville’s theorem on geometric rigidity, see e.g. [38]. This says that every smooth local isometry of a domain corresponds to a rigid body motion. A generalization to the Sobolev setting is due to Reshetnyak [53], and states that if \(u\in W^{1,p}(\Omega ;\mathbb {R}^n)\) with \(\Omega \subset \mathbb {R}^n\) a bounded Lipschitz domain and \(1<p<\infty \) satisfies
pointwise almost everywhere in \(\Omega \), then u is harmonic and
It is not hard to see that if connectedness of the domain fails, then global rigidity is no longer true, as different connected components can then be rotated and translated individually.
Yet, for a domain that has several rigid components arranged into very fine parallel layers (see Fig. 1), global geometric constraints of anisotropic nature occur in the limit of vanishing layer thickness. Since these restrictions become prominent only after a limit passage, we speak of asymptotic rigidity of layered structures. A first rigorous result in this direction can be found in [16] for the special case \(n=2\) and \(p=2\). There it was proven that, under the assumption of local volume preservation and up to global rotations, only shear deformations aligned with the orientation of the layers can occur as effective deformations.
In this paper, we extend the result of [16] to arbitrary dimensions \(n\geqq 2\) and general \(1<p<\infty \), and more significantly, relax the assumption of rigid layers by requiring only sufficient stiffness (see Theorem 1.1). Formally, this corresponds to replacing the exact differential inclusion (1.1) by an approximate one, very much like the quantitative rigidity estimate by Friesecke et al. [33, Theorem 3.1] generalizes Reshetnyak’s theorem. The paper [33] has initiated increased interest in rigidity and its quantification over the last few years, especially among analysts working on variational methods with applications in materials science. For instance, a quantitative version of piecewise rigidity for SBV-functions [12] was established in [32], and there is recent work on the rigidity of conformal maps [31], of non-gradient fields [50] and of the non-Euclidean setting [45].
To be more precise about our results, some notation on the geometry of bi-layered structures is needed. Throughout the manuscript, let \(\Omega \subset \mathbb {R}^n\) with \(n\geqq 2\) be a bounded Lipschitz domain, \(\lambda \in (0,1)\), and \(Y=(0,1]^n\) the periodicity cell. We set
cf. Fig. 1. Without further mention, \(Y_{\mathrm {soft}}\) and \(Y_{\mathrm {stiff}}\) are identified with their Y-periodic extensions. To describe the thickness of two neighboring layers, we introduce a parameter \(\varepsilon >0\), which is supposed to be small and captures the length scale of the heterogeneities. The disjoint sets \(\varepsilon Y_{\mathrm {stiff}}\cap \Omega \) and \(\varepsilon Y_{\mathrm {soft}}\cap \Omega \) partition the domain \(\Omega \) into two phases of alternating layers. Notice that the parameter \(\lambda \) stands for the relative thickness of the softer components.
Under certain technical assumptions on the domain, in particular, flatness and cross-section connectedness, which are specified in Definitions 3.6 and 3.7, we obtain as our first main result a characterization for the asymptotic behavior of sequences of functions on \(\Omega \) whose gradients are increasingly close to SO(n) in \(\varepsilon Y_{\mathrm {stiff}}\cap \Omega \) as \(\varepsilon \rightarrow 0\).
Theorem 1.1
Let \(\Omega \subset \mathbb {R}^n\) be a bounded, flat and cross-section connected Lipschitz domain and \(1<p<\infty \).
- (i)
Suppose that \((u_\varepsilon )_\varepsilon \subset W^{1,p}(\Omega ;\mathbb {R}^n)\) is such that
$$\begin{aligned} \int _{\varepsilon Y_{\mathrm {stiff}}\cap \Omega } {{\,\mathrm{dist}\,}}^p\big (\nabla u_\varepsilon , SO(n)\big ) \,\mathrm {d}x \leqq C\varepsilon ^\alpha \end{aligned}$$(1.3)for all \(\varepsilon >0\) with \(\alpha \geqq 0\) and a constant \(C>0\). If \(\alpha >p\) and \(u_\varepsilon \rightharpoonup u\) in \(W^{1,p}(\Omega ;\mathbb {R}^n)\) for some \(u\in W^{1,p}(\Omega ;\mathbb {R}^n)\), then
$$\begin{aligned} u(x)= R(x)x + b(x), \quad x \in \Omega , \end{aligned}$$(1.4)with \(R\in W^{1,p}(\Omega ;SO(n))\) and \(b\in W^{1,p}(\Omega ;\mathbb {R}^n)\) such that \(\partial _i R =0\) and \(\partial _i b =0\) for \(i=1, \ldots , (n-1)\).
- (ii)
If \(u\in W^{1,p}(\Omega ;\mathbb {R}^n)\) is of the form (1.4), then there exists a sequence \((u_\varepsilon )_\varepsilon \subset W^{1,p}(\Omega ;\mathbb {R}^n)\) such that \(u_\varepsilon \rightharpoonup u\) in \(W^{1,p}(\Omega ;\mathbb {R}^n)\) and \(\nabla u_\varepsilon \in SO(n)\) a.e. in \(\varepsilon Y_{\mathrm {stiff}}\cap \Omega \) for every \(\varepsilon >0\).
One observes that (1.4) resembles (1.2), just that now R will in general not be constant, but depends on the \(x_n\)-variable, and hence, varies in the direction orthogonal to the layers. This condition can be considered the result of a non-trivial interplay between the effects of rigidity and anisotropy.
The proof of Theorem 1.1(i) consists of three main steps: the layerwise approximation of each \(u_\varepsilon \) by rigid body motions, a compactness argument for the resulting one-dimensional auxiliary functions of piecewise constant rotations, and a limit representation argument. Regarding its overall structure, the reasoning is organized similarly to [16, Proposition 2.1]. Technically, however, the transition from exact to the approximate differential inclusions requires two substantial changes, which make the arguments more involved than in [16]. Instead of Reshetnyak’s theorem, we apply the quantitative rigidity estimate on each layer, and the Fréchet–Kolmogorov compactness result (see Lemma 3.4) is used as a refinement of Helly’s selection principle.
Proving the second part of Theorem 1.1 involves the explicit construction of an approximating sequence \((u_\varepsilon )_\varepsilon \) with the desired properties. To this end, we critically exploit the special structure of u as in (1.4), which features a splitting of the \(x_n\)-variable from the remaining ones, so that u has essentially the character of a one-dimensional function.
Remark 1.2
-
(a)
The gradient of u as in (1.4) takes the form
$$\begin{aligned} \nabla u = R + (\partial _n R) x \otimes e_n + \partial _n b\otimes e_n, \end{aligned}$$(1.5)which necessarily requires that \((\nabla u)e_i = Re_i\) for all \(i=1, \ldots , n-1\).
-
(b)
We point out that the scaling regime \(\alpha >p\), which quantifies the relation between thickness and stiffness of the layers, is optimal for Theorem 1.1(i). As shown in Section 2, asymptotic rigidity of layered structures fails for \(\alpha \leqq p\). We provide explicit examples inspired by bending deformations, for which the limit maps u are such that \(\partial _1 u\) depends non-trivially on \(x_1\) or \(\partial _1 u\) is not normed to one. Note that the two extreme cases \(\alpha =0\) and “\(\alpha =\infty \)” (formal for \(\varepsilon ^\alpha =0\)) in (1.3) correspond the situations of the stiff layers being actually soft or fully rigid, respectively.
-
(c)
Theorem 1.1 can be extended in different directions. One generalization concerns a (p, q)-version Theorem 1.1(i). Indeed, if the exponent p in (1.3) is replaced by \(q\in (p, \infty )\) the statement remains valid provided that \(\alpha >q\). In this more general setting, we can let \(1\leqq p<\infty \). The only modification in the case \(p=1\) is that R and b will be BV-functions. We refer to Remark 3.5(a) and Remark 3.2(b) for more details. Moreover, as mentioned in Remark 3.2(c), asymptotic rigidity in the sense of Theorem 1.1(i) still holds if the relative thickness of the stiff layers depend on \(\varepsilon \), being much larger than \(\varepsilon ^{\frac{\alpha }{p}-1}\). For a comment on reduced assumptions for the domain \(\Omega \), see Remark 4.2 as well as Theorem 3.1.
-
(d)
If one requires additionally in Theorem 1.1 that the limit function u is locally volume preserving: that is \(u \in W^{1,r}(\Omega ;\mathbb {R}^n)\) for \(r \geqq n\) with \(\det \nabla u = 1\) a.e. in \(\Omega \), then \(Re_n\) is constant, see Corollary 3.9. In the two-dimensional setting with \(n=2\), this implies that R is constant, and one can think of u as horizontal shear deformation up to global rotations, cf. also [16, Proposition 2.1].
From the viewpoint of applications in materials science, Theorem 1.1 identifies characteristics of macroscopically attainable deformations of bi-layered high-contrast composite materials. This observation constitutes an important step towards a rigorous characterization of their effective behavior via homogenization. Indeed, we will discuss in the following how asymptotic rigidity of layered structures serves as the basis for solving a relevant class of homogenization problems in the context of hyperelasticity.
In the 1970s, the Italian school around De Giorgi established the concept of \(\Gamma \)-convergence [28, 29] (see also [8, 26] for a comprehensive introduction), which has been used successfully among others in homogenization theory to bridge between microscopic and macroscopic scales. This is a natural notion for variational convergence, i.e. limit passages in parameter-dependent minimization problems. The key property is that if a sequence of energy functionals \(\Gamma \)-converges to a limit functional, this implies, under the assumption of suitable compactness, the convergence of the corresponding infima and (almost) minimizers.
By now classical homogenization results via \(\Gamma \)-convergence include the papers by Marcellini [46] in the convex setting, as well as the first work in the non-convex case with standard p-growth by Müller [47] and Braides [7]. Within multiscale analysis, which comprises homogenization and relaxation theory, variational problems with non-convex pointwise or differential constraints are known to be technically challenging, cf. [9, 17, 20, 30, 42]. Despite recent partial progress towards attacking the issue of localization, i.e. proving that limit functionals preserve integral form, with different methods, e.g. [20, 30, 40, 52], there are still general open questions that cannot be worked out with existing tools. In this article, we investigate homogenization problems subject to a special type of approximate differential inclusion constraint, which do not satisfy standard assumptions and therefore require a tailored approach.
Let \(\alpha >0\) and \(p\in (1, \infty )\). Consider for each \(\varepsilon >0\) the integral functional \(E_\varepsilon \) defined for \(u\in W^{1,p}(\Omega ;\mathbb {R}^n)\) by
with an integrand \(W_{\mathrm {soft}}:\mathbb {R}^{n\times n}\rightarrow \mathbb {R}\), which is in general not convex or quasiconvex. These functionals model the elastic energy of a layered composite. The first term with diverging elastic constants, scaling like \(\varepsilon ^{-\alpha }\), is the contribution of the stiff components and the second term is associated with the softer components.
In the regime \(\alpha >p\), we show that the \(\Gamma \)-limit of \((E_\varepsilon )_\varepsilon \) as \(\varepsilon \rightarrow 0\) with respect to strong convergence in \(L^p(\Omega ;\mathbb {R}^n)\), or equivalently weak convergence in \(W^{1,p}(\Omega ;\mathbb {R}^n)\), exists and determine a characterizing formula. The required technical assumptions on the geometry of \(\Omega \) are those of Definitions 3.6 and 3.7 and the density \(W_{\mathrm {soft}}\) is supposed to satisfy (H1)-(H3), see Section 5. In fact, the \(\Gamma \)-limit has integral form, is subject to the constraints on the admissible macroscopic deformations induced by asymptotic rigidity (cf. Theorem 1.1), and can be expressed purely in terms of the energy density \(W_{\mathrm {soft}}\) and the relative thickness \(\lambda \) of the softer layers. More precisely,
for all u of the form (1.4), and \(E_{\mathrm{hom}}(u)=\infty \) otherwise. Here, \(W_{\mathrm {soft}}^{\mathrm{qc}}\) stands for the quasiconvex envelope of \(W_{\mathrm {soft}}\); for background information on generalized notions of convexity and relaxations, see e.g. [25].
Next, we collect a few remarks to put the above mentioned homogenization result—a detailed formulation of the full version is given in Theorem 5.2—in context with related work in the literature.
Remark 1.3
-
(a)
General theorems on homogenization tend to be rather implicit in the sense that they involve (multi)cell formulas (e.g. [7, 47]), which again require to solve infinite dimensional minimization problems. In contrast, the \(\Gamma \)-limit in (1.6) is clearly explicit with regards to the macroscopic effect of the heterogeneities. If the relaxation of the softer components, or in other words, the quasiconvexification of \(W_{\mathrm {soft}}\), is known, the representation of the homogenized energy density becomes even fully explicit. To illustrate the latter, we discuss the prototypical example of the Saint-Venant Kirchhoff stored energy function in Example 5.1.
-
(b)
As we demonstrate in Remark 5.5, the density in (1.6) coincides with a single-cell formula. This indicates that microstructures ranging over multiple cells (or layers) are not energetically favorable, in contrast with the general theory. Indeed, Müller’s well-known counterexample [47], which involves a polyconvex energy density function, gives evidence that multi-cell formulas are necessary in general to describe homogenized limits of non-convex problems (see also [5] for further examples). The recent paper [51] refines this observation by showing that a single-cell formula is sufficient in a neighborhood of rotations, though.
-
(c)
Next, we highlight a selection of related references on the variational analysis of different types of elastic high-contrast composites. The case of stiff inclusions in a softer phase is covered in [10, 30], while [10, 13, 14] study the asymptotics of material models with increasingly soft inclusions. For results on the extreme regime of perforated materials, we refer to the seminal paper [1], and more recently, in the context of brittle elastic materials to [11]. The effective behavior of fiber-reinforced brittle materials is studied in [6]. A common feature of all these results is the isotropy of the derived homogenized energies. In contrast, strong anisotropy at the macroscopic level arises in Theorem 5.2 from the layered geometry of the heterogeneities, especially in the form of restrictions on the class of admissible deformations.
-
(d)
Asymptotic rigidity as a concept and technical tool is not only limited to the homogenization problem in Theorem 5.2. It can be used also in other contexts and has the potential for extensions in different directions, as recent work on the asymptotic analysis of models for layered materials in finite crystal plasticity illustrates, see [16, 27] and [15, Chapter 5, 6]. In particular, [27] contains a BV-version of Theorem 1.1 in the case of fully rigid components, which makes a first connection with applications in fracture mechanics. A step towards carrying the results to problems in stochastic homogenization is made in [15], by assuming a random distribution of the layer thickness.
We conclude the introduction with a few words about the proof of Theorem 5.2, focussing on the main ideas and technical challenges. The construction of a recovery sequence for affine limit maps (Step 1) is based on laminates made of rotations and shear components (cf. [16, Section 4]), which we augment with suitable perturbations on the softer layers. The harder part is the case of general limits (Step 3). Recall that Theorem 1.1(ii) provides an admissible approximating sequence for any possible limit map as in (1.4). However, these sequences fail to be energetically optimal in general. To remedy this problem, we localize by piecewise constant approximation of the limit functions, which can be done in a constraint preserving way due to the essentially one-dimensional character of the representation in (1.4) (see also (1.5)). Finally, we determine locally optimal microstructures as in the affine case and glue them onto the sequence from Theorem 1.1(ii) in the softer parts. This construction is sufficient to recover the energy.
In essence, our reasoning for the liminf-inequality (Steps 2 and 4) comes down to using Theorem 1.1(i) and to applying Jensen’s inequality twice, first to obtain a lower bound energy estimate on each softer layer and then, in the optimization process over the entirety of layers. Besides, we employ the properties of Null-Lagrangians. The presented arguments rely strongly on the hypothesis that \(W_{\mathrm {soft}}^{\mathrm{qc}}\) is polyconvex (referred to as (H1)), meaning that the quasiconvex envelope can be written as a convex function of the vector of minors, or in other words, that the quasiconvex envelope coincides with the polyconvex one. Notice that the same assumption can be found e.g. in [20] in the context of relaxation problems with constraints on the determinant.
Dropping (H1) appears to be a non-trivial task. On a technical level, if the Jensen’s inequalities mentioned above were to be replaced straight away by the related formulas defining quasiconvexity (see (5.2)), this would require careful cut-off arguments at the boundaries. In the stiff layers, though, cut-off conflicts with the rigidity constraints and difficulties may arise from non-local effects due to interaction between different layers. Hence, it remains an open question to understand whether removing (H1) from the list of assumptions makes the \(\Gamma \)-limit \(E_\mathrm{hom}\) in (1.6) (if existent) smaller. Or in more intuitive terms, can the energy be further reduced by oscillations of the rotation matrices and long range effects over multiple layers?
Structure of the article This paper is organized into five sections. In the subsequent Section 2, we discuss a range of explicit bending examples, which illustrate softer macroscopic behavior in the regimes \(0 < \alpha \leqq p\) and establish in particular the optimality of the condition \(\alpha > p\) in Theorem 1.1(i). Sections 3 and 4 contain the proofs of the asymptotic rigidity result formulated in Theorem 1.1. In Section 3, we prove a generalization of the necessity part (i) as well as Corollary 3.9, followed by a more detailed discussion on the geometric assumptions on the domain \(\Omega \). Section 4 proceeds with the proof of the sufficiency statement (ii) of Theorem 1.1. In Section 5, we state our second main result on homogenization via \(\Gamma \)-convergence, that is Theorem 5.2. For its proof, both parts of Theorem 1.1 are key. We conclude by relating the homogenization formula of (1.6) to the cell formula as it occurs in models of composites with rigid layers. The “Appendix” provides two technical auxiliary results in form of a specialized reverse Poincaré type inequality and a lemma on locally one-dimensional functions.
Notation The standard unit vectors in \(\mathbb {R}^n\) are denoted by \(e_1, \ldots , e_n\). For the Euclidean inner product between two vectors \(a, b\in \mathbb {R}^n\) we write \(a\cdot b\). Moreover, let \(a\otimes b=a b^T\in \mathbb {R}^{n\times n}\) for \(a, b\in \mathbb {R}^n\), and set \(a^\perp = (-a_n, a_2, \ldots , a_{n-1}, a_1)^T\in \mathbb {R}^n\) for \(a\in \mathbb {R}^n\), which generalizes the usual notation for perpendicular vectors in two dimensions. The Frobenius norm of \(A \in \mathbb {R}^{n\times n}\) is given by \(|A|=\sqrt{AA^T}\). Our notation for block diagonal matrices is \(A={{\,\mathrm{diag}\,}}(A_1, A_2, \ldots , A_m)\in \mathbb {R}^{n\times n}\) with \(A_i\in \mathbb {R}^{n_i\times n_i}\) and \(\sum _{i=1}^m n_i=n\). In the following, we will often split up \(a\in \mathbb {R}^n\) as \(a=(a', a_n)\), where \(a'=(a_1, \ldots , a_{n-1})\). For a matrix \(A\in \mathbb {R}^{n\times n}\) a similar splitting into its columns is used, that is \(A=(A'|Ae_n)\) with \(A'\in \mathbb {R}^{n\times (n-1)}\). For \(t\in \mathbb {R}\), the expressions \(\lfloor t \rfloor \) and \(\lceil t\rceil \) stand for the largest integer smaller and smallest integer larger than t, respectively.
By a domain \(\Omega \subset \mathbb {R}^n\) we mean an open, connected subset of \(\mathbb {R}^n\). An open cuboid is the Cartesian product \(Q=(a_1, b_1)\times \cdots \times (a_n, b_n) =: \times _i (a_i, b_i)\subset \mathbb {R}^n\) with \(a_i, b_i\in \mathbb {R}\) and \(a_i<b_i\) for \(i=1, \ldots , n\). Hence for us, cuboids will always be oriented along the coordinate axes. Furthermore, \(\mathbb {1}_E\) and \(\chi _E\) are the indicator and characteristic function corresponding to a subset \(E\subset \mathbb {R}^n\), i.e., \(\mathbb {1}_E(x) = 1\) and \(\chi _E(x)=0\) if \(x\in E\), and \(\mathbb {1}_E(x)=0\) and \(\chi _E(x)=\infty \) if \(x\notin E\). For a measurable set U and an integrable function \(f:U\rightarrow \mathbb {R}^m\), let \(\fint _{U} f \,\mathrm {d}{x} := \frac{1}{|U|} \int _U f\,\mathrm {d}x\).
We use the common notation for Lebesgue and Sobolev spaces, as well as for function spaces of continuously differentiable functions. By \(L^p_0(\Omega ;\mathbb {R}^m)\), we denote the space of functions in \(L^p(\Omega ;\mathbb {R}^m)\) with the property that their mean value vanishes. Periodic boundary condition are indicated by a lower case \(\#\), for example in \(W^{1,p}_{\#}(Y;\mathbb {R}^m)\).
The distributional derivative of a function \(f\in L^1_\mathrm{loc}(\Omega ;\mathbb {R}^m)\) is denoted by Df, for partial derivatives in the \(e_i\)-direction we write \(\partial _i u\). Moreover, \(Df=(D'f|\partial _n f)\) with \(D'f=(\partial _1 f| \ldots |\partial _{n-1} f)\). If \(f:\Omega \rightarrow \mathbb {R}^m\) is classically or weakly differentiable, we denote the (weak) gradient of f by \(\nabla f\). Here again, one has the splitting \(\nabla f=(\nabla 'f|\partial _n f)\) with \(\nabla ' f=(\partial _1 f|\ldots |\partial _{n-1}f)\). In case \(f:J\rightarrow \mathbb {R}^m\) is a one-dimensional function with \(J\subset \mathbb {R}\) an open interval, we simply write \(f'\) for the derivative of f.
Convergence of a sequence \((u_\varepsilon )_\varepsilon \) as \(\varepsilon \rightarrow 0\) means that \((u_{\varepsilon _j})_j\) converges as \(j\rightarrow \infty \) for any subsequence \(\varepsilon _j\downarrow 0\). Note finally the use of generic constants, mostly denoted by c or C, which may vary from line to line without change in notation.
2 Optimality of the Scaling Regimes
While for \(\alpha =0\) in (3.2) the class of effective deformations with finite energy comprises arbitrary Sobolev maps with vanishing mean value, the material response in the case “\(\alpha =\infty \)” is rather rigid. This raises the natural question up to which value of \(\alpha \) softer material response can be encountered. In this section, we discuss four examples of macroscopically attainable deformations. They show that Theorem 1.1 and Corollary 3.9 fail for small elastic constants in the regime \(\alpha \leqq p\), and illustrate the effect of (local) volume preservation. For simplicity, we assume throughout this section that \(\Omega \subset \mathbb {R}^n\) is the unit cube, i.e. \(\Omega =(0,1)^n\).
The idea behind the first two constructions for \(\alpha =p\) is to bend the individual stiffer layers, first uniformly in Example 2.3, and then in a locally volume-preserving way inspired by the bending of a stack of paper in Example 2.4. Example 2.5 is based on a wrinkling construction for the individual layers, and shows that compression in layer direction is possible for \(\alpha \in (0,p)\). Finally, we look into the effect of the local volume condition for \(\alpha >p\) in Example 2.6.
The calculations behind these examples share a common structure and are all based on the following auxiliary result. We deliberately keep its formulation slightly more general than actually needed in what follows. This facilitates the construction of an even larger variety of explicit deformations and yields immediate insight into their asymptotic properties.
As regards notation, we write \({\llbracket {x}\rrbracket }_\varepsilon \) for the orthogonal projection of \(x\in \varepsilon Y_{\mathrm {stiff}}\) onto the midsection of the stiff layer containing x; if \(x\in \varepsilon Y_{\mathrm {soft}}\), \({\llbracket {x}\rrbracket }_\varepsilon \) refers to the projection onto the midsection of the closest stiff layer above x. Precisely, for \(\varepsilon >0\) and \(t\in \mathbb {R}\), we let \([t]_\varepsilon = \varepsilon \big \lceil \frac{t}{\varepsilon } \big \rceil -\varepsilon + \tfrac{1+ \lambda }{2}\varepsilon \), so that
Due to
we observe that \({\llbracket {x}\rrbracket }_\varepsilon \rightarrow x\) as \(\varepsilon \rightarrow 0\).
Lemma 2.1
Let \({\overline{Q}} = [0,1]^{n-1}\times [-\frac{1}{2}, 2]\) and \(1<p<\infty \). For \(\varepsilon \in (0,1)\), let \(f_\varepsilon \in C^2({\overline{Q}}; \mathbb {R}^n)\) be such that \(|\partial _1 f_\varepsilon |=1\), \(\partial _1 f_\varepsilon \in \mathrm{span}\{e_1, e_n\}\) and \(\partial _i f_\varepsilon = e_i\) for \(i=2, \ldots , n-1\), and define a Lipschitz function \(u_\varepsilon : \Omega \rightarrow \mathbb {R}^n\) by
and by linear interpolation in the \(e_n\)-direction in \(\varepsilon Y_{\mathrm {soft}}\cap \Omega \).
Then, for any \(\varepsilon \in (0,1)\),
Moreover, if \(\lim _{\varepsilon \rightarrow 0} \varepsilon \Vert \nabla ^2 f_\varepsilon \Vert _{L^\infty ({\overline{Q}};\mathbb {R}^{n\times n\times n})}=0\) and if there is \(F\in L^p({\overline{Q}};\mathbb {R}^{n\times n})\) such that either
- (i)
\(\nabla f_\varepsilon \rightarrow F\) in \(L^p({\overline{Q}};\mathbb {R}^{n\times n})\) as \(\varepsilon \rightarrow 0\), or
- (ii)
\(\nabla f_\varepsilon \rightharpoonup F\) in \(L^p({\overline{Q}};\mathbb {R}^{n\times n})\) as \(\varepsilon \rightarrow 0\) and \(\partial _{n}(\nabla f_\varepsilon ) = 0\) for all \(\varepsilon \in (0,1)\),
then
Remark 2.2
The choice of \({{\overline{Q}}}\) as the domain of the functions \(f_\varepsilon \) ensures that \(u_\varepsilon \) as in (2.2) is well-defined. Indeed, if \(t\in [0,1]\) and \(\varepsilon \in (0,1)\), then \(-\frac{1}{2}<-\frac{\varepsilon }{2} \leqq [t]_\varepsilon \leqq 1+\varepsilon <2\).
Proof
By definition, the functions \(u_\varepsilon \) are continuously differentiable on the connected components of \(\varepsilon Y_{\mathrm {stiff}}\cap \Omega \) and \(\varepsilon Y_{\mathrm {soft}}\cap \Omega \). Then,
for \(x\in \varepsilon Y_{\mathrm {stiff}}\cap \Omega \), and a straight-forward calculation yields the gradients for \(x\in \varepsilon Y_{\mathrm {soft}}\cap \Omega \),
see [15, Lemma 3.4.3]) for more details.
In view of (2.5) and the observation that \(\partial _1f_\varepsilon ({\llbracket {x}\rrbracket }_\varepsilon )\otimes e_1 + \sum _{i=2}^{n-1} e_i\otimes e_i + \partial _1 f_\varepsilon ^\perp ({\llbracket {x}\rrbracket }_\varepsilon ) \otimes e_n \in SO(n)\) for all \(x\in \Omega \) due to \(|\partial _1 f_\varepsilon | = |\partial _1 f_\varepsilon ^\perp |=1\) and \(\partial _1 f_\varepsilon \in \mathrm{span}\{e_1, e_n\}\), the elastic energy contribution on the stiffer layers can be estimated by
For the proof of (2.4), consider the auxiliary fields \(V_\varepsilon \in L^\infty (\Omega ;\mathbb {R}^{n\times n})\) given by
Recall that the indicator function associated with a set \(E\subset \mathbb {R}^n\) is denoted by \(\mathbb {1}_E\). We will show that
Indeed, along with the mean value theorem and (2.1), one obtains for x in the interior of \(\varepsilon Y_{\mathrm {stiff}}\cap \Omega \) that
and, similarly, for \(x\in \varepsilon Y_{\mathrm {soft}}\cap \Omega \),
with some \(\xi \in ([x_n]_\varepsilon -\varepsilon , [x_n]_\varepsilon )\). Accounting for \(\lim _{\varepsilon \rightarrow 0}\varepsilon \Vert \nabla ^2 f_\varepsilon \Vert _{L^\infty ({\overline{Q}};\mathbb {R}^{n\times n\times n})}=0\) leads to (2.7).
In case (i), it follows from (2.6) along with a weak-strong convergence argument that
where we have used in particular that \(\mathbb {1}_{\varepsilon Y_{\mathrm {stiff}}\cap \Omega } \overset{*}{\rightharpoonup }(1-\lambda )\) and \(\mathbb {1}_{\varepsilon Y_{\mathrm {soft}}\cap \Omega } \overset{*}{\rightharpoonup }\lambda \) in \(L^\infty (\Omega )\), as well as \(Fe_i=e_i\) for \(i=2, \ldots , n-1\). The latter follows directly from the assumption that \(\partial _i f_\varepsilon = e_i\) for all \(\varepsilon \in (0,1)\) if \(i=2, \ldots , n-1\).
Combining (2.8) and (2.7) shows that \(\nabla u_\varepsilon \rightharpoonup F\) in \(L^1(\Omega ;\mathbb {R}^{n\times n})\). Since \((\nabla u_\varepsilon )_\varepsilon \) is uniformly bounded in \(L^p(\Omega ;\mathbb {R}^{n\times n})\) by (2.3) and the requirement that \(\varepsilon \Vert \partial _{11}^2 f_\varepsilon \Vert _{L^\infty ({\overline{Q}}; \mathbb {R}^{n})} \rightarrow 0\), we finally infer (2.4), which finishes the proof under the assumption of (i).
If assumption (ii) is satisfied, then \(\partial _1 f_\varepsilon ^\perp \) depends only on \(x_1\). Since \(\mathbb {1}_{\varepsilon Y_{\mathrm {stiff}}}\) on the other hand is constant in the \(x_1\)-variable, we observe a separation of variables in the product \((\partial _1 f_\varepsilon ^\perp ) \mathbb {1}_{\varepsilon Y_{\mathrm {stiff}}}\). In light of this observation, consider test functions \(\varphi \in C^0({\overline{\Omega }};\mathbb {R}^n)\) of the form \(\varphi (x)=(\phi \otimes \psi )(x):=\phi (x_1)\psi (x_2, \ldots , x_n)\) for \(x\in {\overline{\Omega }}\) with \(\phi \in C^0([0,1];\mathbb {R}^n)\) and \(\psi \in C^0([0,1]^{n-1})\). Then, due to Fubini’s theorem and the lemma on weak convergence of rapidly oscillating periodic functions (see e.g. [18, Section 2.3]), it follows that
We recall that as a corollary of the Stone-Weierstrass theorem (see e.g. [54, Theorem 7.32]) and the density of \(C^0({\overline{\Omega }};\mathbb {R}^n)\) in \(L^q(\Omega ;\mathbb {R}^n)\) with \(1\leqq q<\infty \), the span of functions \(\phi \otimes \psi \) is dense in \(L^q(\Omega ;\mathbb {R}^{n})\). Consequently, we infer from (2.9) that
Then the third term in (2.6) converges weakly to \((1-\lambda ) (Fe_1)^\perp \otimes e_n\) in \(L^p(\Omega ;\mathbb {R}^{n\times n})\). Arguing similarly for the other product terms in (2.6) eventually yields \(V_\varepsilon \rightharpoonup F\) in \(L^p(\Omega ;\mathbb {R}^{n\times n})\). In conjunction with (2.7) this proves (2.4), and thus the statement in case (ii). \(\quad \square \)
As announced at the beginning of the section, we will next discuss four specializations of Lemma 2.1, using the same notations. These examples illustrate the optimality of the scaling regimes in Theorem 1.1 and Corollary 3.9.
Example 2.3
(Uniform bending of the individual stiffer layers) Let \(g:[0,1]\rightarrow \mathrm{span}\{e_1, e_n\}\subset \mathbb {R}^n\) be a \(C^2\)-curve parametrized by arc length, i.e., \(|g'(t)|=1\) for all \(t\in [0,1]\). We follow Lemma 2.1 to define deformations \(u_\varepsilon \) by choosing for all \(\varepsilon \in (0,1)\),
This choice of f is motivated by uniform bending of the individual stiffer layers in the two-dimensional setting, where the curve g describes the bending of the mid-fibers, see Fig. 2.
Illustration of the deformations of Example 2.3 for \(n=2\), with uniform bending of the stiffer layers described by \(g(t) = \sin (t - \tfrac{1}{2}) e_1+ \cos (t- \tfrac{1}{2}) e_2\) for \(t\in [0,1]\)
Then, Lemma 2.1 implies that for any constant \(C> 2^p\Vert g''\Vert _{L^\infty (0,1;\mathbb {R}^n)}^p\),
which shows that the sequence \((u_\varepsilon )_\varepsilon \) has finite elastic energy on the stiffer component for \(\alpha =p\). As for the gradient of the limit deformation u, we infer from version i) of Lemma 2.1 that \(\nabla u_\varepsilon \rightharpoonup \nabla u=\nabla f\) in \(L^p(\Omega ;\mathbb {R}^{n\times n})\). In view of (2.10),
with \(R(x)=g'(x_1)\otimes e_1 + \sum _{i=1}^{n-1} e_i\otimes e_i + g'(x_1)^\perp \otimes e_n\) and \({\tilde{a}}(x)=e_n-g'(x_1)^\perp \). Clearly, for general g, \(\partial _1 R\ne 0\), so that the limit deformation u does not have the form (3.3) obtained in Theorem 1.1 for the regime \(\alpha >p\).
We remark that the limit deformation u is not locally volume preserving for g with non-trivial curvature, since \(\det \nabla u= g'\cdot e_1 \not \equiv 1\).
To recover limit deformations that satisfy the local volume constraint, a slightly more involved bending construction as in the next example is needed.
Example 2.4
(Macroscopically volume-preserving bending deformations) In the context of Lemma 2.1, we consider for \(\varepsilon \in (0,1)\) the functions
with \(g: [0,2] \rightarrow \mathrm{span}\{e_1, e_n\} \subset \mathbb {R}^n\) a \(C^2\)-curve parametrized by arclength.
Then the sequence \((u_\varepsilon )_\varepsilon \) defined by (2.2) in the stiffer component and by linear interpolation in the softer one satisfies
and we obtain that \(\nabla u_\varepsilon \rightharpoonup \nabla u=\nabla f\) in \(L^p(\Omega ;\mathbb {R}^{n\times n})\). Due to
for \(x\in \Omega \), one can rewrite the gradient of the limit deformation u with the help of a map of rotations \(R\in L^\infty (\Omega ;SO(n))\) defined for \(x\in \Omega \) by \(R(x)e_1=g'(\frac{x_1}{x_n+1})\) and \(R(x)e_i=e_i\) for \(i=2, \ldots , n-1\). Precisely,
with \({\tilde{a}}(x) = -\frac{x_1}{x_n+1}g'\big (\frac{x_1}{x_n+1}\big ) + g\big ( \frac{x_1}{x_n+1}\big ) -g' \big ( \frac{x_1}{x_n+1}\big )^\perp \) for \(x\in \Omega \). The rotations R depend non-trivially on \(x_1\), hence, the limit map u is not in compliance with Theorem 1.1. Since \(\det \nabla u=\det \nabla f = -g'\cdot g^\perp \), the deformation u is locally volume preserving if we chose g such that \(g'\cdot g^\perp \equiv 1\).
An simple deformation of this type, which is intuitively inspired by the bending of a stack of paper, is depicted in Fig. 3.
Illustration of the deformations of Example 2.4 for \(n=2\), with \(g(t) = \sin (t-\frac{1}{2}) e_1 + \cos (t -\frac{1}{2}) e_2\) for \(t\in [0,2]\). Notice also that the limit deformation u satisfies the local volume constraint, whereas the bending deformations of the individual layers in the left picture do not
Next, we discuss an example in the regime \(\alpha <p\), where macroscopic shortening in the \(e_1\)-direction occurs due to wrinkling of the stiffer layers. A similar effect occurs in the context of plate theory, cf. [33, Section 5].
Example 2.5
(Wrinkling of stiffer layers) Let \(\beta \in \mathbb {R}\), \(\gamma \in (0,1)\), and \(g:[0,1] \rightarrow \mathrm{span}\{e_1, e_n\} \subset \mathbb {R}^n\) be a 1-periodic \(C^2\)-function with \(|g'(t)|=1\) for all \(t\in \mathbb {R}\). We define \(g_\varepsilon :[0,1]\rightarrow \mathbb {R}^n\) by \(g_\varepsilon (t) = \varepsilon ^\gamma g(\varepsilon ^{-\gamma }t)\) for \(t\in [0,1]\) and \(\varepsilon \in (0,1)\), and observe that by the weak convergence of periodically oscillating sequences,
Unless \(g'\) is constant, \(|\bar{g}'| <1\). Under these assumptions, the functions
meet the requirements of Lemma 2.1 with assumption (ii) and \(F = \bar{g}'\otimes e_1 + \sum _{i=2}^{n-1} e_i\otimes e_i+ \beta e_n\otimes e_n\). Thus, for \(u_\varepsilon \) as in Lemma 2.1,
and \(\nabla u_\varepsilon \rightharpoonup \nabla u=F\) in \(L^1(\Omega ;\mathbb {R}^{n\times n})\). In particular, \(|(\nabla u)e_1| = |Fe_1|= |\bar{g}'| <1\).
Since \(\det \nabla u = \det F = \beta (\bar{g}'\cdot e_1)\), (local) volume preservation of the limit deformation u can be achieved by a suitable choice of \(\beta \) and g. Graphically speaking, \(\beta \) can be viewed as a stretching factor in the \(e_n\)-direction that compensates the loss of length in the \(e_1\)-direction due to the asymptotic shortening of the mid-fibers in the stiffer layers, so that overall volume is preserved. A specific case of this wrinkling construction is depicted in Fig. 4.
Illustration of the deformation in Example 2.5 for \(n=2\)
Our last example highlights the role of the local volume constraint of the limit deformation in the regime \(\alpha \geqq 0\). In particular, it shows that for \(\alpha >p\) local volume preservation of the limit deformation is necessary to obtain asymptotic rigidity in the sense of Corollary 3.9.
Example 2.6
(Rotation of stiffer layers) Let \(R \in C^1([-\frac{1}{2},2];SO(n))\) with \(Re_i=e_i\) for \(i=2, \ldots , n-1\). For each \(\varepsilon \in (0,1)\), we set
and take \(u_\varepsilon \) as defined in Lemma 2.1. Since \(\partial _{11} f=0\), it follows from (2.3) that
for any \(\varepsilon \in (0,1)\). Moreover, \(\nabla u_\varepsilon \rightharpoonup \nabla u= \nabla f\) in \(L^p(\Omega ;\mathbb {R}^{n\times n})\), so that for \(x\in \Omega \),
where \(d(t) = - \frac{1}{2} R'(t) e_1 - R'(t) t e_n\) for \(t\in (0,1)\). Hence, we obtain
with \(b(t) = -\frac{1}{2} Re_1 - \int _0^{t} sR'(s)e_n \,\mathrm {d}{s} + c\) for \(t\in (0,1)\) and \(c\in \mathbb {R}^n\). It is now immediate to see that u has the form stated in Theorem 1.1, but neither is \(Re_n\) constant nor is the local volume condition satisfied in general.
In 2d, this construction corresponds to a \(x_2\)-dependent rotation of the individual stiffer layers around their barycenters, see Fig. 5.
Illustration of the deformation of Example 2.6 for \(n=2\), where the increasing rotation of the stiffer layers is described by \(R\in C^2([-\frac{1}{2},2];SO(2))\) with \(R(t)e_1=\cos (t) e_1 + \sin (t)e_2\) for \(t\in [-\frac{1}{2},2]\)
We conclude this section with a reference to Fig. 6, which illustrates at one glance our findings in different scaling regimes for two space dimensions. Notice that any \((2\times 2)\)-matrix can be expressed as \(R(\beta \mathbb {I}+a\otimes e_2)\) with \(R\in SO(2)\), \(\beta \in \mathbb {R}\) and \(a\in \mathbb {R}^2\).
Overview of the results on the asymptotic behavior of weakly converging sequences \((u_\varepsilon )_\varepsilon \subset W^{1,p}(\Omega ;\mathbb {R}^2)\) satisfying (1.3) in the different scaling regimes for \(n=2\), a without and b with local volume constraint on the limit map u. It is used here that for any \(u\in W^{1,p}(\Omega ;\mathbb {R}^2)\) there are \(R\in L^\infty (\Omega ;SO(2))\), \(\beta \in L^p(\Omega )\) and \(a\in L^p(\Omega ;\mathbb {R}^2)\) with \(\nabla u= R(\beta \mathbb {I}+a\otimes e_2)\)
3 Proof of Necessity in Theorem 1.1
We will show in this section that weak limits of bounded energy sequences in the context of our model for layered materials with stiff and soft components have a strongly one-dimensional character. To make this more precise, we first introduce the following terminology. A measurable function \(f:\Omega \rightarrow \mathbb {R}^m\), where \(\Omega \subset \mathbb {R}^n\) is an open set, is said to be locally one-dimensional in the \(e_n\)-direction if for every \(x\in \Omega \) there is an open cuboid \(Q_x\subset \Omega \) with \(x\in Q_x\) such that for all \(y, z\in Q_x\),
We call f (globally) one-dimensional in the \(e_n\)-direction if (3.1) holds for all \(y, z\in \Omega \). For \(f\in W^{1,p}_\mathrm{loc}(\Omega ;\mathbb {R}^m)\) with \(p\geqq 1\) local one-dimensionality in the \(e_n\)-direction of f, which means that there exists a representative of f with the property, is equivalent to the condition \(\nabla ' f=0\), as can be seen from a standard mollification argument. Hence, if \(\nabla ' f=0\), the function f can be identified locally (i.e. for any \(x\in \Omega \) on an open cuboid \(Q_x\subset \Omega \) containing x) with a one-dimensional \(W^{1,p}\)-function. Since the latter is absolutely continuous, it follows that f is continuous.
The next result and its implications discussed subsequent to its proof generalize the necessity statement of Theorem 1.1 relaxing the assumptions on the domain.
Theorem 3.1
Let \(\Omega \subset \mathbb {R}^n\) with \(n\geqq 2\) be a bounded open set, \(1<p<\infty \) and \(\alpha >p\). Furthermore, let \((u_\varepsilon )_\varepsilon \subset W^{1,p}(\Omega ;\mathbb {R}^n)\) be such that for all \(\varepsilon >0\),
with a constant \(C>0\), and \(u_\varepsilon \rightharpoonup u\) in \(W^{1,p}(\Omega ;\mathbb {R}^n)\) for some \(u\in W^{1,p}(\Omega ;\mathbb {R}^n)\).
Then there exist \(R\in W^{1, p}_{\mathrm{loc}}(\Omega ; SO(n))\) with \(\nabla ' R=0\) and \(b\in W^{1, p}_{\mathrm{loc}}(\Omega ; \mathbb {R}^n)\) with \(\nabla ' b = 0\) such that
Remark 3.2
-
(a)
Notice that the functions R and b are both locally one-dimensional in the \(e_n\)-direction and continuous. In particular, \(u\in C^0(\Omega ;\mathbb {R}^n)\).
-
(b)
It is straightforward to generalize Theorem 3.1 to a (p, q)-version. Precisely, if p in (3.2) is replaced with any \(p<q<\infty \), the same conclusion remains true under the assumption that \(\alpha >q\), cf. [15, Section 3.3]. For a discussion of the case \(p=1\), see Remark 3.5
-
(c)
One can show that the statement of Theorem 3.1 remains true if the relative thickness of the softer layers \(\lambda \in (0,1)\) depends on \(\varepsilon \) (then denoted by \(\lambda _\varepsilon \)) in such a way that \(1-\lambda _\varepsilon \gg \varepsilon ^{\frac{\alpha }{p}-1}\). For more details, we refer to [15, Theorem 3.3.1].
Theorem 3.1 builds on two classical results, which we recall here for the readers’ convenience. The first one is the quantified rigidity result for Sobolev functions established in [33, Theorem 3.1], cf. also [19, 22, 23] for generalizations to other \(W^{1, p}\)-settings.
Theorem 3.3
(Quantitative rigidity estimate) Let \(U \subset \mathbb {R}^n\) with \(n\geqq 2\) be a bounded Lipschitz domain and \(1< p<\infty \). Then there exists a constant \(C=C(U, p)>0\) with the property that for each \(u \in W^{1,p}(U; \mathbb {R}^n)\) there is a rotation \(R \in SO(n)\) such that
A straightforward scaling argument shows that the constant C remains unaffected by uniform scaling and translation of U. Applying the above theorem to increasingly thinner domains, however, leads to degenerating constants. If \(U= P_\varepsilon =O \times \varepsilon I\subset \mathbb {R}^n\) with \(\varepsilon >0\), \(O\subset \mathbb {R}^{n-1}\) a cube and \(I\subset \mathbb {R}\) a bounded open interval one obtains that
see [34, Section 4] and [15, Section 3.5.1].
The second tool is the Fréchet–Kolmogorov theorem, a compactness result for \(L^p\)-functions, see e.g. [2, Sections 2.15, U.2] and [37]. Here, we will apply it only in the basic version formulated in the next lemma, that is, for families of functions of one real variable with uniformly bounded essential supremum.
Lemma 3.4
Let \(J, J'\subset \mathbb {R}\) be open, bounded intervals with \(J\subset \subset J'\) and \(1\leqq p<\infty \). If the sequence \((f_\varepsilon )_\varepsilon \) is uniformly bounded in \(L^\infty (J';\mathbb {R}^m)\) satisfying
then \((f_\varepsilon )_\varepsilon \) is relatively compact in \(L^p(J;\mathbb {R}^m)\).
Regarding structure, the next proof proceeds along the lines of [16, Proposition 2.1], which, as mentioned in the introduction, constitutes a special case of Theorem 1.1. Yet, the individual steps are more involved and require new, refined arguments to relax the assumption of the stiff layers being fully rigid and to overcome the restriction to two space dimensions.
Proof of Theorem 3.1
Let \(Q=O\times J\subset \Omega \) be a cuboid with \(O\subset \mathbb {R}^{n-1}\) an open cube of side length \(l>0\) and \(J\subset \mathbb {R}\) an open interval. Suppose that there exist open intervals \(J', J''\) with \(J\subset \subset J'\subset \subset J''\) and \(Q'' := O\times J''\subset \Omega \). Moreover, let \(Q':= O\times J'\). We define horizontal strips by setting
The index set \(I_\varepsilon \) contains all \(i\in \mathbb {Z}\) with \(|P_\varepsilon ^i|=\varepsilon |O|\), and we assume \(\varepsilon >0\) to be small enough, so that \(Q\subset Q' \subset \bigcup _{i \in I_\varepsilon } P^i_\varepsilon \subset Q''\).
For the proof, it suffices to show the existence of \(R\in W^{1,p}(Q;SO(n))\) and \(b\in W^{1,p}(Q;\mathbb {R}^n)\) with \(\nabla ' R=0\) and \(\nabla 'b=0\) in Q, respectively, such that the characterization (3.3) holds for \(x\in Q\). Then we can approximate \(\Omega \) from inside with overlapping cuboids to obtain the same statements for any compact \(K\subset \Omega \). Indeed, the resulting characterizations in terms of R and b coincide on the overlapping parts. Finally, exhausting \(\Omega \) with compact nested subsets proves Theorem 3.1 in the stated generality.
In what follows, the constants \(C>0\) depend at most on \(n, p, \lambda , \Omega \) and c from (3.2), in particular, they are independent of \(\varepsilon \), l and J.
Step 1: Layerwise approximation by rigid body motions In this first step, we will construct a sequence of piecewise affine functions \((w_\varepsilon )_\varepsilon \) such that the restriction of each \(w_\varepsilon \) to a strip \(P_\varepsilon ^i\) is a rigid body motion and
Applying Theorem 3.3 (under consideration of the scaling behavior of the constant according to (3.4)) to the individual stiff layers yields the existence of \(C>0\) and of rotations \(R^i_\varepsilon \in SO(n)\) for every \(i\in I_\varepsilon \) such that
Let \(w_\varepsilon \in L^\infty (Q';\mathbb {R}^n)\) be defined by \(w_\varepsilon =\sigma _\varepsilon + b_\varepsilon \), where
and
The specific choice of the values \(b_\varepsilon ^i\) implies that \(\int _{\varepsilon Y_{\mathrm {stiff}}\cap P_\varepsilon ^i} u_\varepsilon -w_\varepsilon \,\mathrm {d}{x}=0\), and therefore allows us to apply Poincaré’s inequality to \(u_\varepsilon -w_\varepsilon \) on each stiff layer. Hence, one obtains for every \(i\in I_\varepsilon \) that
see e.g. [35, Section 7.8] for details on the domain dependence of the Poincaré constant.
Next we derive a corresponding bound on the softer layers. By a shifting argument, this problem can be reduced to estimate (3.8) for the stiff layers. The error is given in terms of difference quotients in the \(e_n\)-direction of \(u_\varepsilon -w_\varepsilon \), which we control uniformly. More precisely, for fixed \(i\in I_\varepsilon \) we cover \(\varepsilon Y_{\mathrm {soft}}\cap P_\varepsilon ^i\) with finitely many shifted copies of \(\varepsilon Y_{\mathrm {stiff}}\cap P_\varepsilon ^i\), that is, we choose \(0<\delta _{\varepsilon , k} \leqq \lambda \varepsilon \) for \(k=1, \ldots , N:=\lceil \tfrac{\lambda }{1-\lambda }\rceil \) such that the \(\delta _{\varepsilon , k}\)-shifted stiff layers \(O_{\varepsilon , k}^i:= (\varepsilon Y_{\mathrm {stiff}}\cap P_\varepsilon ^i) -\delta _{\varepsilon , k} e_n\) satisfy \(\varepsilon Y_{\mathrm {soft}}\cap P_\varepsilon ^i \subset \bigcup _{k=1}^N O_{\varepsilon , k}^i\). Then,
Here, we have used a one-dimensional difference quotient estimate with respect to the \(x_n\)-variable. Summing over the N covering cuboids then leads to
notice that the last term results from the fact that \(\nabla w_\varepsilon \in SO(n)\) on \(P^i_\varepsilon \) by (3.7).
Finally, we take the sum over \(i\in I_\varepsilon \) to deduce from (3.8) and (3.6) that
Therefore, by (3.2) and the uniform boundedness of \((u_\varepsilon )_\varepsilon \) in \(W^{1,p}(\Omega ;\mathbb {R}^n)\),
Since \(\alpha >p\), this implies (3.5).
Step 2: Compactness of the approximating rigid body motions Consider for \(\varepsilon >0\) the piecewise constant one-dimensional auxiliary function \(\Sigma _\varepsilon : J'\rightarrow SO(n)\) defined by
with \(R_\varepsilon ^i\) as in Step 1. In relation to (3.7), it holds that \(\sigma _\varepsilon (x) = \Sigma _\varepsilon (x_n)x\) for \(x\in Q'\).
Step 2a: Estimate for rotations on different strips Next we will show that for every \(\xi \in \mathbb {R}\) such that \(J \cup (J+\xi )\subset J'\),
To this end, we estimate the expression \(\Vert w_\varepsilon (\cdot + \xi e_n) - w_\varepsilon \Vert _{L^p(Q;\mathbb {R}^n)}\) from above and below.
The upper bound follows from
For the lower bound, we set \(d_{\varepsilon , \xi }^i = b_\varepsilon ^{i+ \lfloor \frac{\xi }{\varepsilon }\rfloor } -b_\varepsilon ^i + \xi R_\varepsilon ^{i+\lfloor \frac{\xi }{\varepsilon } \rfloor }e_n\) and use Lemma A.1 to derive that
Combining (3.12) and (3.13) gives (3.11).
Step 2b: Application of the Fréchet–Kolmogorov theorem To establish strong \(L^p\)-convergence of \((\Sigma _\varepsilon )_\varepsilon \) as \(\varepsilon \rightarrow 0\), observe that in view of (3.9) and the uniform boundedness of \((u_\varepsilon )_\varepsilon \) in \(W^{1,p}(\Omega ;\mathbb {R}^{n})\), estimate (3.11) turns into
It is standard to verify (see e.g. [33, Proof of Theorem 4.1] for an analogous argument) that then
Hence, by Theorem 3.4, there exist a subsequence (not relabeled) and a \(\Sigma _0 \in L^p(J;\mathbb {R}^{n\times n})\) such that
Note that \(\Sigma _0\) may still depend on the subsequence at this point. In Step 3, \(\Sigma _0\) will be characterized in terms of the limit function u, which makes \(\Sigma _0\) unique and the above argument independent of the choice of subsequences. Due to the strong \(L^p\)-convergence of \((\Sigma _{\varepsilon })_\varepsilon \), which preserves lengths and angles almost everywhere, we conclude that \(\Sigma _0\in SO(n)\) a.e. in \(J'\).
Step 2c: Regularity of \(\Sigma _0\) As a result of (3.14), we obtain an estimate on the difference quotients of \(\Sigma _{\varepsilon }\), precisely
Passing to the limit \(j\rightarrow \infty \) results in
which shows that \(\Sigma _0\in W^{1,p}(J;\mathbb {R}^{n\times n})\), see e.g. [44, Theorem 10.55].
Step 3: Representation of the limit function u Recall the definitions of \(\sigma _\varepsilon \) in (3.7) and \(\Sigma _\varepsilon \) in (3.10). With \(\sigma _0(x) = \Sigma _0(x_n)x\) for \(x\in Q\) one has that
Then, by (3.15),
Since \(b_\varepsilon = w_\varepsilon -\sigma _\varepsilon = (w_\varepsilon -u_\varepsilon ) + u_\varepsilon -\sigma _\varepsilon \) we find in view of (3.5), (3.17) and the convergence \(u_\varepsilon \rightarrow u\) in \(L^p(\Omega ;\mathbb {R}^n)\) by the compact embedding of \(W^{1,p}(Q;\mathbb {R}^n)\) into \(L^p(Q;\mathbb {R}^n)\) that
Due to the regularity of u and \(\sigma _0\), it follows that \(b\in W^{1,p}(Q;\mathbb {R}^n)\). Since \(b_\varepsilon \) is independent of the \(x'\)-variables, the same is true for b. Finally, defining
proves the desired representation of u. \(\quad \square \)
Remark 3.5
-
(a)
Setting \(p=1\) in Remark 3.2 b) in combination with Theorem 3.1 leads to the representation (3.3) with \(R\in BV_{\mathrm{loc}}(\Omega ;SO(n))\) and \(b\in BV_{\mathrm{loc}}(\Omega ;\mathbb {R}^n)\) satisfying \(D'R=0\) and \(D'b=0\), respectively. The reasoning is the same as for \(p>1\), but instead of getting \(\Sigma _0\in W^{1,1}(J;\mathbb {R}^{n\times n})\) from (3.16), we can only deduce that \(\Sigma _0 \in BV(J; \mathbb {R}^{n\times n})\), see e.g. [44, Corollary 2.43].
-
(b)
Notice that in view of (3.18) and (3.16) it holds that
$$\begin{aligned} \Vert R\Vert ^p_{W^{1,p}(Q;\mathbb {R}^{n\times n})} \leqq C(1+ l^{-p}). \end{aligned}$$(3.19)This estimate is not uniform for all cuboids \(Q\subset \Omega \) as used in the proof of Theorem 3.1. In fact, the bound becomes large for cuboids with small cross-section. One can therefore not expect in general that the weak derivatives of R be p-integrable on the whole of \(\Omega \).
-
(c)
If \(\Omega \) in Theorem 3.1 is of the form \(\Omega =O\times I\) with \(O\subset \mathbb {R}^{n-1}\) an open cube of side length \(l>0\) and \(I\subset \mathbb {R}\) an interval, then the proof shows that \(R\in W^{1,p}(\Omega ;SO(n))\), and hence also \(b\in W^{1,p}(\Omega ;\mathbb {R}^n)\), for any \(p>1\). Indeed, let us choose intervals \(J_k\subset \subset I\) for \(k\in \mathbb {N}\) such that \(J_k\subset J_{k+1}\) and \(I=\bigcup _{k=1}^\infty J_k\) and set \(Q_k=O\times J_k\). Then by estimate (3.19),
$$\begin{aligned} \Vert R\Vert _{W^{1,p}(Q_k;\mathbb {R}^{n\times n})}^p\leqq C, \end{aligned}$$(3.20)with \(C>0\) independent of k. Since the cuboids \(Q_k\) exhaust \(\Omega \), the uniform bound (3.20) yields that \(R\in W^{1,p}(\Omega ;SO(n))\).
The observation of Remark 3.5(c) can be extended to a larger class of Lipschitz domains. In fact, under suitable additional assumptions on \(\Omega \), namely connectedness of cross-sections and a flatness property, which are introduced in Definitions 3.6 and 3.7, we can drop the restriction to local \(W^{1,p}\)-regularity of R and b in Theorem 3.1, as Corollary 3.8 below shows.
Definition 3.6
(Connectedness of cross-sections) An open set \(\Omega \subset \mathbb {R}^n\) is called cross-section connected if for any \(t\in \mathbb {R}\) the intersection \(\Omega _t\) of \(\Omega \) with the hyperplane \(H_t=\{x\in \mathbb {R}^n: x_n =t\}\) is connected.
Clearly, every convex set is cross-section connected, but also cylinders and cones in \(\mathbb {R}^n\) (oriented in the \(e_n\)-direction) with non-convex cross section are. In Fig. 7a, b we give a two-dimensional example for illustration. An important property of domains \(\Omega \) as in Definition 3.6 is that any locally one-dimensional vector (and matrix) field in the \(e_n\)-direction defined on \(\Omega \) is already globally one-dimensional in the \(e_n\)-direction, cf. Lemma A.2.
Definition 3.7
(Flatness) We call an open set \(\Omega \subset \mathbb {R}^n\) flat, if for all \(t\in \mathbb {R}\) the intersection of \({\overline{\Omega }}\) with the hyperplane \(H_t=\{x\in \mathbb {R}^n: x_n=t\}\) is either empty or has nonempty relative interior.
The intuitive geometric interpretation of flatness of bounded domains is that it rules out sets with sharp or rounded corners and peaks pointing in the direction of \(e_n\). Simple examples include cylinders with axis parallel to \(e_n\), whereas cones with the same orientation are not flat, see also Fig. 7c, d. A bounded Lipschitz domain \(\Omega \subset \mathbb {R}^n\) does in general not satisfy the condition of Definition (3.7), but it can be turned into a flat Lipschitz domain by cutting it off on top and bottom, i.e., by taking \((\mathbb {R}^{n-1}\times (a, b))\cap \Omega \), where \(a, b\in \mathbb {R}\) with \(a<b\) are such that the cross sections \(\Omega _a\) and \(\Omega _b\) are non-empty.
Corollary 3.8
In addition to the assumptions of Theorem 3.1, let \(\Omega \subset \mathbb {R}^n\) be a flat and cross-section connected Lipschitz domain. Then the representation (3.3) holds with \(R\in W^{1,p}(\Omega ;SO(n))\) and \(b\in W^{1,p}(\Omega ;\mathbb {R}^n)\).
Proof
Let \(Q_\Omega \) be the smallest open cuboid containing \(\Omega \) and let \(a, b\in \mathbb {R}\) with \(a<b\) and \(O_\Omega \subset \mathbb {R}^{n-1}\) be such that \(Q_\Omega =O_\Omega \times J_\Omega \) with \(J_\Omega =(a, b)\). We observe first that due to the connectedness of the cross-sections of \(\Omega \), the map R from (3.3) is globally one-dimensional in the \(e_n\)-direction and can thus be identified with a one-dimensional function \(\Sigma \in W^{1,p}_\mathrm{loc}(J_\Omega ;SO(n))\), see Lemma A.2 and Remark A.3.
Moreover, since \(\Omega \) is a flat Lipschitz domain there exist \(x_a\in \Omega _a\) and \(x_b\in \Omega _b\) along with open cuboids \(Q_a=O\times (a, a+r)\) and \(Q_b=O\times (b-r,b)\) of height \(r>0\) and cross-section \(O\subset \mathbb {R}^{n-1}\) such that \(Q_a\cap Q_\Omega \subset \Omega \) and \(Q_b\cap Q_\Omega \subset \Omega \). Applying Remark 3.5(c) to the restrictions \(R_a=R|_{Q_a}\) and \(R_b=R|_{Q_b}\) gives that \(R_a\in W^{1,p}(Q_a;SO(n))\) and \(R_b\in W^{1, p}(Q_b;SO(n))\), which correspond to elements in \(\Sigma _a\in W^{1,p}(a, a+r;SO(n))\) and \(\Sigma _b\in W^{1,p}(b-r, b;SO(n))\), respectively. Hence, \(\Sigma \in W^{1,p}(J_\Omega ;SO(n))\) and \(R\in W^{1,p}(Q_\Omega ;SO(n))\), thus also \(R\in W^{1,p}(\Omega ;SO(n))\).
Since \(b=u-Rx\) with \(u\in W^{1,p}(\Omega ;\mathbb {R}^m)\), one immediately gets the desired statement for b. \(\quad \square \)
We conclude this section with the following specialization of Corollary 3.8, which involves the additional condition that the limit map is locally volume preserving.
Corollary 3.9
In addition to the assumptions on \(\Omega \), \((u_\varepsilon )_\varepsilon \) and u in Corollary 3.8, let \(u \in W^{1,r}(\Omega ;\mathbb {R}^n)\) for \(r \geqq n\) be such that \(\det \nabla u = 1\) a.e. in \(\Omega \).
Then the limit representation in (3.3) holds with \(Re_n\) constant. If \(\Omega \) is simply connected, one has in particular that
where \(Q\in SO(n)\), \(S={{\,\mathrm{diag}\,}}(S', 1)\) with \(S'\in W^{1,p}(\Omega ; SO(n-1))\) satisfying \(\nabla ' S'=0\) and \(a\in L^{\max \{r, p\}}(\Omega ;\mathbb {R}^n)\) with \(D'a=(S')^T(\partial _n S')\) and \(a_n=0\).
Proof of Corollary 3.9
By Theorem 3.1, we know that u has the representation (3.3). Hence,
with \({\tilde{a}}= (\partial _n R)x + \partial _n b\) and \(a=R^T{\tilde{a}}\). Since \(\det \nabla u=\det (R(\mathbb {I}+a\otimes e_n))= 1+a_n\), we conclude in view of the local volume preservation constraint that \(a_n=0\).
Differentiating the identity \(0= a_n= {\tilde{a}}\cdot Re_n\) with respect to the ith variable for \(i\in \{1, \ldots , n-1\}\), while taking into account that \(\nabla ' R=0\) and \(\nabla ' b=0\), implies that \(\partial _n (Re_i)\cdot Re_n =0\). Since \(Re_i\) is orthogonal on \(Re_n\) pointwise almost everywhere, it follows from the product rule that
Together with
we obtain that \(\partial _n (Re_n)=0\). Hence, \(Re_n\) is constant, and R splits multiplicatively into the product of Q and S as in the statement.
Finally, the restriction on the distributional derivatives of a with respect to the first \({n-1}\) variables follows via straightforward calculation from the gradient structure of \(\nabla u\), which requires that \({{\,\mathrm{curl}\,}}\nabla u= 0\). \(\quad \square \)
Remark 3.10
If \(n=2\), the gradient representation of u in (3.21) becomes
with \(Q\in SO(2)\) and \(\gamma \in L^p(\Omega )\) with \(\partial _1\gamma =0\), cf. also [16, Proposition 2.1]. In the two-dimensional setting, the class of limit deformations u of \(\Omega \) is highly restricted, in fact, only horizontal shearing and global rotation can occur.
4 Sufficiency Statement in Theorem 1.1
Our starting point in this section are functions \(u\in W^{1, p}(\Omega ;\mathbb {R}^n)\) with gradients of the form
where \(R\in W^{1,p}(\Omega ;SO(n))\) and \(d \in L^p(\Omega ;\mathbb {R}^n)\) with \(\nabla ' R=0\) and \(D' d =0\), respectively. If not mentioned otherwise, \(1<p<\infty \) and \(\Omega \subset \mathbb {R}^n\) is a bounded domain.
We will show how such u (under suitable technical assumptions) can be approximated in the sense of weak convergence in \(W^{1,p}(\Omega ;\mathbb {R}^{n\times n})\) by functions \(u_\varepsilon \) that are defined on a layered domain with length scale of oscillations \(\varepsilon \) and coincide with rigid body motions on the stiff components. This in particular proves Theorem 1.1(ii).
Before stating the general result, let us consider a simple example for motivation. If u is affine, then \(\nabla u=F\) for some \(F\in \mathbb {R}^{n\times n}\) and there exist a matrix \(R_F\in SO(n)\) and a vector \(d_F\in \mathbb {R}^n\) such that \(\nabla u = F= R_F+d_F\otimes e_n\). This motivates the definition
Moreover, we set
In the affine case, the construction of a suitable approximation is particularly simple. The idea is to compensate for the stiff layers by performing stronger deformations on the softer layers, which leads to the following laminate construction: for \(\varepsilon >0\), let \(v_\varepsilon ^F \in W^{1,\infty }(\Omega ;\mathbb {R}^n)\) be such that
then \(\nabla v_\varepsilon ^F \in SO(n)\) a.e. in \(\varepsilon Y_{\mathrm {stiff}}\cap \Omega \) and \(\nabla v_\varepsilon ^F = R+ \frac{1}{\lambda } \mathbb {1}_{\varepsilon Y_{\mathrm {soft}}} d\otimes e_n\rightharpoonup \nabla u\) in \(L^p(\Omega ;\mathbb {R}^{n\times n})\) as a consequence of the weak convergence of highly oscillating sequences (see e.g. [18, Section 2.3]). Finally, we set \(u_\varepsilon =v_\varepsilon ^F\) for all \(\varepsilon \) to obtain the desired approximating functions in this special case.
The construction behind the general approximation result is inspired by the case of affine limits. In view of (4.3), we have
Proposition 4.1
Let \(\Omega \subset \mathbb {R}^n\) be a bounded, flat and cross-section connected Lipschitz domain and let \(u\in W^{1,p}(\Omega ;\mathbb {R}^n)\) with \(\nabla u\) as in (4.1). Then there exists a sequence \((u_\varepsilon )_\varepsilon \subset W^{1,p}(\Omega ;\mathbb {R}^n)\) with \(\nabla u_\varepsilon \in SO(n)\) a.e. in \(\varepsilon Y_{\mathrm {stiff}}\cap \Omega \) such that \(\nabla u_\varepsilon \rightharpoonup \nabla u\) in \(L^{p}(\Omega ;\mathbb {R}^{n\times n})\).
More specifically, there is \((R_\varepsilon )_\varepsilon \subset W^{1,p}(\Omega ;SO(n))\) with \(\nabla ' R_\varepsilon =0\) on \(\Omega \) and \(\partial _n R_\varepsilon =0\) on \(\varepsilon Y_{\mathrm {stiff}}\cap \Omega \) such that
and
Remark 4.2
The same result still holds also under relaxed conditions on a bounded Lipschitz domain \(\Omega \), namely when \(\Omega \) can be partitioned into finitely many components that are flat and cross-section connected. More details can be found in [15, Section 4.2].
Proof
Let \(Q_\Omega \) denote the smallest cuboid containing \(\Omega \). By (4.1) and Lemma A.2 (see also Remark A.3(b)), we may assume after constant extension orthogonal to \(e_n\) that \(R\in W^{1, p}(Q_\Omega ;SO(n))\) is globally one-dimensional in the \(e_n\)-direction and continuous. Upon writing \(Q_\Omega =O_\Omega \times J_\Omega \) with \(O_\Omega \subset \mathbb {R}^{n-1}\) an open cuboid and \(J_\Omega \subset \mathbb {R}\) an open, bounded interval, there is a one-dimensional function \(\Sigma \in W^{1,p}(J_Q; SO(n))\) such that \(R(x)= \Sigma (x_n)\) for \(x\in Q_\Omega \).
Let \((\Sigma _\varepsilon )_\varepsilon \subset W^{1,p}(J_\Omega ;SO(n))\) be the approximating sequence for \(\Sigma \) resulting from Lemma 4.3 below, that is, \(\Sigma _\varepsilon \rightharpoonup \Sigma \) in \(W^{1,p}(J_\Omega ;\mathbb {R}^{n\times n})\) and \(\Sigma _\varepsilon ' = 0\) in \(\varepsilon I_{\mathrm{stiff}} \cap J_\Omega \). Moreover, the convergence (4.8) holds. We set \(R_\varepsilon (x)=\Sigma _\varepsilon (x_n)\) for \(x\in Q_\Omega \), so that \(R_\varepsilon \in W^{1,p}(Q_\Omega ;SO(n))\) with \(\nabla ' R_\varepsilon =0\), and define
where
We claim that for each \(\varepsilon >0\) the function \(U_\varepsilon \) has gradient structure, meaning that there exists a potential \(u_\varepsilon \in W^{1,p}(Q_\Omega ;\mathbb {R}^n)\) with \(\nabla u_\varepsilon = U_\varepsilon \). To see this, it suffices to show that the distributional curl of \(U_\varepsilon \) vanishes on \(Q_\Omega \). We remark that \(Q_\Omega \) as a cuboid is simply connected. Indeed, let \(\varphi \in C^\infty _c(Q_\Omega ;\mathbb {R}^n)\) and \(k, l\in \{1, \ldots , n\}\) with \(k<l\). Due to \(\nabla 'R_\varepsilon =0\) and \(\nabla ' d=0\), one obtains in the case \(l<n\) that
and for \(l=n\), along with \(\partial _n R_\varepsilon =0\) on \(\varepsilon Y_{\mathrm {stiff}}\cap Q_\Omega \), that
Thus, \({{\,\mathrm{curl}\,}}U_\varepsilon =0\) as desired.
After restricting \(u_\varepsilon \) and \(R_\varepsilon \) to \(\Omega \), the statements (4.6) and (4.7) follow now directly from the properties of the sequence \((\Sigma _\varepsilon )_\varepsilon \). \(\quad \square \)
The proof of the previous proposition builds builds on a structure preserving approximation result for one-dimensional functions with values in the set of rotations. Let us denote by \(I_{\mathrm {soft}}\) the 1-periodic extensions of the interval \((0, \lambda )\) to the real line, which corresponds to a one-dimensional section of \(Y_{\mathrm {stiff}}\) in the \(e_n\)-direction, that is \(Y_{\mathrm {stiff}}= \mathbb {R}^{n-1}\times I_{\mathrm {stiff}}\). In addition, we set \(I_{\mathrm {stiff}}= \mathbb {R}^n{\setminus } I_{\mathrm {soft}}\).
Lemma 4.3
Let \(J\subset \mathbb {R}\) be an open and bounded interval, \(1\leqq p<\infty \) and \(\Sigma \in W^{1,p}(J;SO(n))\). Then there exists a sequence \((\Sigma _\varepsilon )_\varepsilon \subset W^{1,p}(J;SO(n))\) with
such that \(\Sigma _\varepsilon \rightharpoonup \Sigma \) in \(W^{1,p}(J; \mathbb {R}^{n\times n})\). Furthermore,
Proof
Instead of trying to approximate \(\Sigma \) directly with SO(n)-valued functions, it seems easier to parametrize \(\Sigma \) in a suitable way. Intuitively speaking, the idea is to stop the parametrization on the stiff layers and accelerate it on the softer ones.
More precisely, for every \(\varepsilon >0\), take \(\varphi _\varepsilon : \mathbb {R}\rightarrow \mathbb {R}\) as the piecewise affine function defined by
and by linear interpolation on \(\varepsilon I_{\mathrm {soft}}\), see Fig. 8. By construction, one has that
and \((\varphi _\varepsilon )_\varepsilon \) converges locally uniformly to the identity function on \(\mathbb {R}\) for \(\varepsilon \rightarrow 0\).
First, we extend the function \(\Sigma \) from J to an open real interval \(J'\) that contains J compactly. In fact, via reflection one obtains \(\Sigma \in W^{1,p}(J';SO(n))\) (not renamed) with
where \(c>0\) depends only on \(J'\).
Next, we define \(\Sigma _\varepsilon :J\rightarrow SO(n)\) by \(\Sigma _\varepsilon = \Sigma \circ \varphi _\varepsilon \) for sufficiently small \(\varepsilon \). Notice that \(\Sigma _\varepsilon \) is well-defined, since \(\varphi _\varepsilon (J)=\varphi _\varepsilon (\varepsilon I_{\mathrm {soft}}\cap J)\subset J'\) if \(\varepsilon \) is small enough. As the composition of an absolutely continuous function with a monotone Lipschitz function, \(\Sigma _\varepsilon \) is absolutely continuous. In particular, the chain rule holds (see e.g. [44, Theorem 3.44]), i.e.
and thus, \(\Sigma _\varepsilon \in W^{1,p}(J; SO(n))\). Since \(\Sigma _\varepsilon \rightarrow \Sigma \) pointwise and the functions \(|\Sigma _\varepsilon |^2\leqq n\) a.e. in J, it follows from Lebesgue’s dominated convergence theorem that \(\Sigma _\varepsilon \rightarrow \Sigma \) in \(L^p(J;\mathbb {R}^{n\times n})\).
For the asserted weak convergence of \((\Sigma _\varepsilon )_\varepsilon \) in \(W^{1,p}(J;\mathbb {R}^{n\times n})\), it suffices according to Urysohn’s lemma to show that the sequence \((\Sigma _\varepsilon ')_\varepsilon \) is uniformly bounded in \(L^p(J;\mathbb {R}^{n\times n})\). Indeed,
Here we have exploited (4.10) and (4.9), the fact that \(\Sigma _{\varepsilon }\) is constant on \(\varepsilon I_{\mathrm {stiff}}\), as well as the chain rule and transformation formula on the (finitely many) connected components of \(\varepsilon I_{\mathrm{soft}}\), where the restriction of \(\varphi _\varepsilon \) is invertible.
To show (4.8), we approximate \(\Sigma '\) in \(L^p(J';\mathbb {R}^{n\times n})\) by a sequence \((g_j)_j\subset C^\infty _c(J';\mathbb {R}^{n\times n})\). By change of variables on the connected components of \(\varepsilon I_{\mathrm{soft}}\) it follows that
and therefore
Since \(g_j\circ \varphi _\varepsilon \rightarrow g_j\) in \(L^p(J;\mathbb {R}^{n\times n})\) for every \(j\in \mathbb {N}\) by dominated convergence, passing to the limits \(\varepsilon \rightarrow 0\) and \(j\rightarrow \infty \) (in this order) in (4.11) proves (4.8). \(\quad \square \)
5 Homogenization of Layered High-Contrast Materials
Before proving Theorem 5.2, formulated below, we introduce the setting and precise assumptions. Throughout this section, \(\Omega \subset \mathbb {R}^n\) is a bounded Lipschitz domain that satisfies the flatness condition and connectedness property of Definitions 3.7 and 3.6, respectively, and \(p > n\). For \(\varepsilon > 0\) and \(\alpha >0\) we consider the heterogeneous energy density \(W_\varepsilon ^\alpha : \Omega \times \mathbb {R}^{n\times n}\rightarrow [0, \infty )\) given by
where \(W_{\mathrm {stiff}}, W_{\mathrm {soft}}:\mathbb {R}^{n\times n}\rightarrow [0, \infty )\) are continuous functions that satisfy the following conditions regarding convexity, growth and coercivity, and local Lipschitz continuity:
(H1) \(W_{\mathrm {soft}}^{\mathrm{qc}}\) is polyconvex;
(H2) \(c|F|^p - \frac{1}{C} \leqq W_{\mathrm {soft}}(F) \leqq C(1+|F|^p)\) for all \(F\in \mathbb {R}^{n\times n}\) with constants \(C, c >0\);
(H3) \(|W_{\mathrm {soft}}(F) -W_{\mathrm {soft}}(G)| \leqq L(1+|F|^{p-1}+|G|^{p-1})|F-G|\) for all \(F, G\in \mathbb {R}^{n\times n}\) with \(L>0\);
(H4) \(W_{\mathrm {stiff}}(F)\geqq k {{\,\mathrm{dist}\,}}^{p}(F, {SO(n)})\) for all \(F\in \mathbb {R}^{n\times n}\) with a constant \(k>0\).
An equivalent way of expressing (H1) is by
where \(W_{\mathrm {soft}}^{\mathrm{qc}}\) and \(W_{\mathrm {soft}}^{\mathrm{pc}}\) are the quasiconvex and polyconvex envelopes of \(W_{\mathrm {soft}}\), that is, the largest quasiconvex and polyconvex functions below \(W_{\mathrm {soft}}\). For a detailed introduction to generalized notions of convexity and the corresponding generalized convexifications we refer to [25]. Let us just recall briefly that a continuous function \(W:\mathbb {R}^{n\times n}\rightarrow \mathbb {R}\) with standard p-growth (i.e., with an the upper bound as in (H2)) is quasiconvex if for any \(F\in \mathbb {R}^{n\times n}\),
Moreover, a continuous \(W:\mathbb {R}^{n\times n}\rightarrow \mathbb {R}\) is polyconvex if there exists a convex function \(g:\mathbb {R}^{\tau (n)}\rightarrow \mathbb {R}\) such that
where \(\mathcal {M}(F)\in \mathbb {R}^{\tau (n)}\) with \(\tau (n) = \sum _{i=1}^n \begin{pmatrix}n \\ k\end{pmatrix}\) is the vector of minors of F.
We remark that explicit formulas for quasiconvex envelopes are in general hard to obtain. This is why quasiconvexifications are rather rare in the literature, see e.g. [21, 24, 43] for a few examples (including extended-valued densities). A common strategy is to determine upper and lower bounds in terms of rank-one and polyconvex envelopes and to show that the latter two match. Hence, in those cases where relaxations are explicitly known, (H1) is usually satisfied.
Example 5.1
Let \(n=2\) or \(n=3\). The Saint Venant-Kirchhoff stored energy function,
with the Lamé constants \(\lambda , \mu >0\), is one of the simplest energy densities of relevance in hyperelasticity (see e.g. [36, Section 28]), and meets requirements for \(W_\mathrm{soft}\). It is straightforward to see that \(W_{SK}\) has standard growth (H2) with \(p=4\) and is locally Lipschitz continuous in the sense of (H3). In [43], Le Dret and Raoult give an explicit expression of the quasiconvexification \(W^{\mathrm{qc}}_{SK}\), which coincides with the convex, polyconvex and rank-one convex envelopes. Thus, in particular, (H1) is satisfied, too.
Let \(E_\varepsilon : L^p(\Omega ;\mathbb {R}^n) \rightarrow \mathbb {R}\cup \{\infty \}\) be the integral functional with density \(W_\varepsilon ^\alpha \), i.e.
if \(u\in W^{1, p}(\Omega ;\mathbb {R}^n)\) and \(E_\varepsilon (u)=\infty \) otherwise in \(L^p(\Omega ;\mathbb {R}^n)\).
Recalling that \(\mathcal {A}=\{F\in \mathbb {R}^{n\times n}: F = R_F + d_F \otimes e_n \text { with }R_F \in SO(n)\hbox { and }d_F \in \mathbb {R}^n\}\) (cf. (4.2)), we define for \(F\in \mathcal {A}\),
where \(F_\lambda =R_F + \tfrac{1}{\lambda }d_F\otimes e_n =\tfrac{1}{\lambda }(F-(1-\lambda )R_F) \in \mathcal {A}\).
Now we are ready to formulate the main theorem of this section. Theorem 5.2 provides a characterization of the effective behavior of the bilayered materials modeled by (5.3) by homogenization via \(\Gamma \)-convergence for vanishing layer thickness. The limit problem shows a splitting of the effects of the heterogeneities and relaxation of microstructures on the softer components. With regards to homogenization, the resulting formulas are explicit and can be expressed in terms of the relative layer thickness. Provided the relaxation of \(W_{\mathrm{soft}}\) is known, \(W_{\mathrm{hom}}\) is even fully explicit.
Theorem 5.2
If \(\alpha >p\), the family \((E_\varepsilon )_\varepsilon \) as in (5.3) converges in the sense of \(\Gamma \)-convergence regarding the strong \(L^p\)-topology to the limit functional \(E_{\mathrm{hom}}: L^p(\Omega ;\mathbb {R}^n) \rightarrow \mathbb {R}_\infty \) given by
Precisely, this means that the following two conditions are satisfied:
- (i)
(Lower bound) For each \(u \in L^p(\Omega ;\mathbb {R}^n)\) and any sequence \((u_\varepsilon )_\varepsilon \subset L^p(\Omega ;\mathbb {R}^n)\) with \(u_\varepsilon \rightarrow u\) in \(L^p(\Omega ;\mathbb {R}^n)\) as \(\varepsilon \rightarrow 0\) it holds that
$$\begin{aligned} \liminf _{\varepsilon \rightarrow 0} E_\varepsilon (u_\varepsilon ) \geqq E_\mathrm {hom}(u); \end{aligned}$$ - (ii)
(Existence of a recovery sequence) For each \(u \in L^p(\Omega ;\mathbb {R}^n)\) there exists a sequence \((u_\varepsilon )_\varepsilon \subset L^p(\Omega ;\mathbb {R}^n)\) with \(u_\varepsilon \rightarrow u\) in \(L^p(\Omega ;\mathbb {R}^n)\) as \(\varepsilon \rightarrow 0\) such that
$$\begin{aligned} \lim _{\varepsilon \rightarrow 0} E_\varepsilon (u_\varepsilon ) = E_\mathrm {hom}(u). \end{aligned}$$
Moreover, any sequence \((u_\varepsilon )_\varepsilon \subset L^p_0(\Omega ;\mathbb {R}^n)\) of uniformly bounded energy for \((E_\varepsilon )_\varepsilon \), that is \(E_\varepsilon (u_\varepsilon )<C\) for all \(\varepsilon >0\), is relatively compact in \(L^{p}(\Omega ;\mathbb {R}^n)\).
Remark 5.3
-
(a)
If \(W_{\mathrm {soft}}\) is convex, then \(W_{\mathrm {soft}}^{\mathrm{qc}}=W_{\mathrm {soft}}^{\mathrm{c}}= W_{\mathrm {soft}}\), so that \(W_{\mathrm{hom}}(F) = \lambda W_{\mathrm {soft}}(F_\lambda )\) for \(F\in \mathcal {A}\). In this case, the proof of Theorem 5.2 can be simplified as indicated below.
-
(b)
It is well-known that the definition of quasiconvexity in (5.2), as well as the representation formula for the quasiconvex envelope \(W^{\mathrm{qc}}\), is independent of the choice of the domain, see e.g. [25, Proposition 5.11]. Therefore, we have for any open set \(O\subset \mathbb {R}^n\) that
$$\begin{aligned} W_{\mathrm {soft}}^{\mathrm{qc}}(F) = \inf _{\varphi \in W_0^{1,p}(O;\mathbb {R}^n)}\fint _{O} W_{\mathrm {soft}}(F+ \nabla \varphi )\,\mathrm {d}{y}, \quad F\in \mathbb {R}^{n\times n}. \end{aligned}$$Alternatively, \(W_{\mathrm {soft}}^{\mathrm{qc}}\) can be expressed with the help periodic perturbations on a cube \(Q\subset \mathbb {R}^n\) as
$$\begin{aligned} W_{\mathrm {soft}}^{\mathrm{qc}}(F) = \inf _{\varphi \in W_{\#}^{1,p}(Q;\mathbb {R}^n)} \fint _{Q} W_{\mathrm {soft}}(F+\nabla \varphi )\,\mathrm {d}{y}, \quad F\in \mathbb {R}^{n\times n}, \end{aligned}$$ -
(c)
The homogenized energy density \(W_{\mathrm {hom}}\) is non-negative and inherits the property (H2) from \(W_{\mathrm {soft}}\). This follows from the fact that \(W_{\mathrm {soft}}^{\mathrm{qc}}\) has standard p-growth, because \(W_{\mathrm {soft}}\) has, along with the estimate
$$\begin{aligned} \tfrac{1}{2 \lambda } |F| - \tfrac{n}{\lambda } \leqq |F_\lambda |\leqq \tfrac{1}{\lambda }(|F| + 1)\quad \text {for }F\in \mathcal {A}. \end{aligned}$$(5.5)Moreover, \(W_{\mathrm{hom}}\) is locally Lipschitz continuous in the sense that, just as \(W_{\mathrm {hom}}\), it satisfies hypothesis (H3). Precisely, one can find \(L_{\mathrm{hom}}>0\) such that
$$\begin{aligned}&|W_{\mathrm {hom}}(F) -W_{\mathrm {hom}}(G)| \leqq L_{\mathrm{hom}}(1+|F|^{p-1}+|G|^{p-1})|F-G| \quad \text {for all }F, G\in \mathcal {A}. \end{aligned}$$(5.6)To see this, we exploit that the property (H3) carries over from \(W_{\mathrm {soft}}\) to \(W_{\mathrm {soft}}^{\mathrm{qc}}\) (cf. e.g. [48, Lemma 2.1 c)]). Hence,
$$\begin{aligned}&|W_{\mathrm {hom}}(F) -W_{\mathrm {hom}}(G)|\leqq \lambda |W_{\mathrm {soft}}^{\mathrm{qc}}(F_\lambda ) -W_{\mathrm {soft}}^{\mathrm{qc}}(G_\lambda )| \\&\quad \leqq \lambda {\tilde{L}}(1+|F_\lambda |^{p-1}+|G_\lambda |^{p-1})|F_\lambda -G_\lambda | \end{aligned}$$for \(F, G\in \mathcal {A}\) with a constant \({\tilde{L}}>0\). In view of (5.5), it only remains to estimate \(|F_\lambda -G_\lambda |\) suitably from above by \(|F-G|\). We observe that
$$\begin{aligned} |F_\lambda -G_\lambda |&\leqq \tfrac{1}{\lambda } |F-G| + \tfrac{1-\lambda }{\lambda } (|\widehat{F}-\widehat{G}| + |R_Fe_n - R_Ge_n|) \\&\leqq \tfrac{2-\lambda }{\lambda }|F-G| + \tfrac{1-\lambda }{\lambda } |R_Fe_n-R_Ge_n|, \end{aligned}$$where \(\widehat{A}\) stands for the \(n\times (n-1)\)-matrix that results from removing the last column of \(A\in \mathbb {R}^{n\times n}\). We denote the n-dimensional cross product of vectors \(v_1, \ldots , v_n\in \mathbb {R}^n\) by \(v_1\times \cdots \times v_{n-1} = \times _{i=1}^{n-1} v_i\in \mathbb {R}^n\). The latter is by definition the uniquely determined vector that is orthogonal on the hyperplane spanned by \(v_1, \ldots , v_{n-1}\) such that the orientation of \(v_1, \ldots , v_{n-1}, \times _{i=1}^{n-1}v_i\) is positive and its norm is the volume of the parallelotope associated with \(v_1, \ldots , v_{n-1}\). For every rotation \(R\in SO(n)\), one has that \(Re_n=\times _{i=1}^{n-1} Re_i\). The multilinearity of the cross product in \(\mathbb {R}^n\) and the fact that \(|R_Fe_i| = |R_Ge_i|=1\) for \(i=1, \ldots , n\) allows us to obtain iteratively that
$$\begin{aligned} |R_Fe_n-R_Ge_n|&= |\times _{i=1}^{n-1} R_F e_{i} - \times _{i=1}^{n-1} R_Ge_i | \\&\leqq |R_Fe_1-R_Ge_1| + |R_Ge_1\times R_G e_2\times \cdots \times R_Ge_{n-1}- R_Ge_1\\&\qquad \times R_Fe_2\times \cdots \times R_Fe_{n-1}| \\&\leqq \cdots \leqq \sum _{i=1}^{n-1}|R_F e_i-R_G e_i|\leqq (n-1) |\widehat{R_F}-\widehat{R_G}|\leqq (n-1) |F-G|. \end{aligned}$$Finally, we combine the above estimates to deduce the desired local Lipschitz property (5.6).
-
(d)
As mentioned in the introduction, proving a \(\Gamma \)-limit homogenization result as above without the hypothesis (H1) is an open problem. In any case, Theorem 5.2 provides an upper bound on the \(\Gamma \)-limit (if existent) in that situation.
We subdivide the proof of Theorem 5.2 into three main parts. After showing compactness, we first determine the homogenization \(\Gamma \)-limit for all affine functions, and then prove the general statement via a localization argument. Note that the specific structure of the admissible limit deformations as characterized in Theorem 3.1, in particular the resulting multiplicative separation of \(x'\) and \(x_n\)-variables in (4.1), is key. This observation allows us to construct an approximation that fulfills the (asymptotic) constraints on the stiff layers, cf. Proposition 4.1.
The first part of the proof is standard, yet, we sketch it here for the readers’ convenience.
Proof of Theorem 5.2 (Part I): Compactness
Let \((u_\varepsilon )_\varepsilon \subset L^p_0(\Omega ;\mathbb {R}^n)\) be such that \(E_\varepsilon (u_\varepsilon )<C\) for all \(\varepsilon >0\). Then, since \({{\,\mathrm{dist}\,}}(F, SO(n))\geqq |F| - \sqrt{n}\) for all \(F\in \mathbb {R}^{n\times n}\), the lower bounds on \(W_{\mathrm {soft}}\) and \(W_{\mathrm {stiff}}\) in (H2) and (H4), imply that \((\nabla u_\varepsilon )_\varepsilon \) is uniformly bounded in \(L^p(\Omega ;\mathbb {R}^n)\). The stated relative compactness of \((u_\varepsilon )_\varepsilon \) in \(L^p(\Omega ;\mathbb {R}^n)\) follows now from Poincaré’s inequality, which shows that
along with the compact embedding \(W^{1,p}(\Omega ;\mathbb {R}^n) \hookrightarrow \hookrightarrow L^p(\Omega ;\mathbb {R}^n)\). \(\quad \square \)
Proof of Theorem 5.2 (Part II): Affine case
Suppose that \(u\in W^{1,p}(\Omega ;\mathbb {R}^n)\) with \(E_\mathrm{hom}(u)<\infty \) is affine. Hence, there is \(F\in \mathbb {R}^{n\times n}\) with \(F\in \mathcal {A}\), cf. (4.2).
Step 1: Existence of a recovery sequence The construction of a recovery sequence for u as above, that is, finding \((u_\varepsilon )_\varepsilon \subset L^p(\Omega ;\mathbb {R}^n)\) with
requires a careful adaptation of by now classical techniques, see e.g. [47]. Indeed, instead of glueing small-scale oscillations on top of an affine function, the former are glued onto an appropriate laminate, namely the one constructed in (4.4).
Let \(\delta >0\). In view of Remark 5.3 b), one can find \(\varphi _\delta \in W^{1,p}_0(Y_{\mathrm {soft}};\mathbb {R}^n)\) such that
We set \(\varphi _\delta \) equal to zero in the remainder of the unit cube and extend it Y-periodically to \(\mathbb {R}^n\). For \(\varepsilon >0\) let \(v_\varepsilon ^F\) be a Lipschitz function with gradients as in (4.4) and vanishing mean value on \(\Omega \). Then, \(v_\varepsilon ^F\rightarrow u\) in \(L^p(\Omega ;\mathbb {R}^n)\) as \(\varepsilon \rightarrow 0\). With
it follows that \(u_{\delta , \varepsilon }\rightarrow u\) in \(L^p(\Omega ;\mathbb {R}^n)\) as \(\varepsilon \rightarrow 0\). Regarding energies, we obtain that
Hence, as \(\varepsilon \) tends to zero,
and we infer, along with (5.8), that
By Attouch’s diagonalization lemma (see e.g. [4, Lemma 1.15, Corollary 1.16]) there exist \(\delta (\varepsilon )\) such that \(u_{\delta (\varepsilon ),\varepsilon } \rightarrow u\) in \(L^p(\Omega ;\mathbb {R}^n)\) and \(E_\varepsilon (u_{\delta (\varepsilon ), \varepsilon }) \rightarrow E_{\mathrm{hom}}(u)\) as \(\varepsilon \rightarrow 0\). Finally, defining \(u_\varepsilon :=u_{\delta (\varepsilon ), \varepsilon }\) yields the desired recovery sequence for u.
Step 2: Lower bound Let \((u_\varepsilon )_\varepsilon \subset L^p(\Omega ;\mathbb {R}^n)\) be such that \(u_\varepsilon \rightarrow u\) in \(L^p(\Omega ;\mathbb {R}^n)\). Since the energies \(E_\varepsilon \) and \(E_{\mathrm{hom}}\) depend only on gradients, one may assume that the functions \(u_\varepsilon \) and u have vanishing mean value, i.e., \(u_\varepsilon , u\in L_0^p(\Omega ;\mathbb {R}^n)\). We will show that
Without loss of generality, let \(\liminf _{\varepsilon \rightarrow 0}E_{\varepsilon }(u_\varepsilon ) = \lim _{\varepsilon \rightarrow 0}E_\varepsilon (u_\varepsilon )<\infty \). In view of Part I, one may further assume that
We remark that (5.9) follows immediately, if one can prove that
for any open cuboid \(Q = O\times J \subset \subset \Omega \), where \(O\subset \mathbb {R}^{n-1}\) and \(J\subset \mathbb {R}\) and open interval. To deduce (5.9), we can then exhaust \(\Omega \) with disjoint cuboids \(Q_i\subset \Omega \) for \(i\in \mathbb {N}\) such that \(|\Omega {\setminus } \bigcup _{i=1}^\infty Q_i| = 0\) and apply (5.9) on each \(Q_i\). More precisely, for any \(N\in \mathbb {N}\),
so that taking the supremum over \(N\in \mathbb {N}\) implies (5.9).
It remains to prove (5.11), which relies substantially on hypothesis (H1), or equivalently on (5.1). Since \(W_{\mathrm {soft}}^{\mathrm{pc}}\) is polyconvex, we can find a convex function \(g:\mathbb {R}^{\tau (n)}\rightarrow \mathbb {R}\) such that \(W^\mathrm {pc}(F) = g(\mathcal {M}(F))\) for all \(F \in \mathbb {R}^{n\times n}\). Moreover, let \(P_\varepsilon ^i = (\mathbb {R}^{n-1}\times \varepsilon [i, i+1))\cap Q\) for \(i\in \mathbb {N}\) and \(I_\varepsilon \subset \mathbb {N}\) an index set such that \(i\in I_\varepsilon \) if and only if \(|P_\varepsilon ^i| = \varepsilon |O|\). As a consequence, \(|\varepsilon Y_{\mathrm {soft}}\cap P_\varepsilon ^i|= \lambda \varepsilon |O|\) for all \(i\in I_\varepsilon \), and one finds with \(Q_\varepsilon =\bigcup _{i\in I_\varepsilon } P_\varepsilon ^i\subset Q\) that
Due to the convexity of g we can invoke Jensen’s inequality, applied twice, first in the version for Lebesgue-measurable functions and second in the discrete version, to obtain
With the aim of eventually passing to the limit \(\varepsilon \rightarrow 0\) in (5.13), we will show first that
For this the properties of \(u_\varepsilon \) due to the presence of the stiff layers need to be taken into account. Owing to (H4) and (5.10), the sequence \((u_\varepsilon )_\varepsilon \) satisfies the requirements of Theorem 3.1, and also Corollary 3.8. Following the proofs, we find the one-dimensional auxiliary sequence \((\Sigma _\varepsilon )_\varepsilon \subset L^p(J;SO(n))\) defined in (3.10) with the properties that \(\Sigma _\varepsilon \rightarrow \Sigma _0\) in \(L^p(J;\mathbb {R}^{n\times n})\) and \(\Sigma _0(x_n)=R_F\) for \(x\in Q\), cf. (3.15) and (3.18).
For each \(\varepsilon \), we extend \(\Sigma _\varepsilon \) constantly in \(x'\) and call the resulting function \(S_\varepsilon \in L^\infty (Q;SO(n))\). As a consequence of (3.6) (cf. also (3.10)) it holds that
Summing up, we have hence found a sequence \((S_\varepsilon )_\varepsilon \subset L^\infty (Q;SO(n))\) such that
as \(\varepsilon \rightarrow 0\).
To see (5.14), let us rewrite the expression \(\mathcal {M}(\nabla u_\varepsilon )\mathbb {1}_{\varepsilon Y_{\mathrm {soft}}\cap Q}\) as follows:
It is well-known that for \(p>n\) weak continuity of minors holds, that is, \(\mathcal {M}(\nabla u_\varepsilon ) \rightharpoonup \mathcal {M}(\nabla u)=\mathcal {M}(F)\) in \(L^1(\Omega ;\mathbb {R}^{\tau (n)})\), see e.g. [25, Theorem 8.20, Part 4]. Furthermore, the first convergence in (5.15) yields
while the second allows us to conclude that
Both convergences follow from the following estimate based on the Leibniz formula for determinants, together with Hölder’s inequality and the uniform bounds on \((S_\varepsilon )_\varepsilon \) and \((\nabla u_\varepsilon )_\varepsilon \) in \(L^\infty (\Omega ;\mathbb {R}^{n\times n})\) and \(L^p(\Omega ;\mathbb {R}^{n\times n})\), respectively. For any \(A,B\in \mathbb {R}^{m\times m}\),
where \(\mathcal {S}_m\) denotes the symmetric group over a set of m elements. To give more detail regarding the argument behind (5.17), we observe that with \(\mathcal {M}^m(F)\) denoting any subdeterminant of \(F\in \mathbb {R}^{n\times n}\) of order \(m\leqq n\),
with a constant \(C>0\) depending on m, p and Q and the uniform bound on \(\Vert \nabla u_\varepsilon \Vert _{L^p(\Omega ;\mathbb {R}^{n\times n})}\).
From the lemma on weak convergence of highly oscillating periodic functions [18, Section 2.3] we infer that \(\mathbb {1}_{\varepsilon Y_{\mathrm {stiff}}\cap Q}\overset{*}{\rightharpoonup }(1-\lambda )\) in \(L^\infty (Q)\). Finally, applying these results to the individual terms in (5.16) along with a weak-strong convergence argument implies (5.14).
Next, we observe that, as a Null-Lagrangian or polyaffine function, \(G\mapsto \mathcal {M}(G)\) for \(G\in \mathbb {R}^{n\times n}\) is also rank-one affine, cf. [25, Theorem 5.20]. Since \(F = \lambda F_\lambda +(1-\lambda )R_F\) and \(F_\lambda -R_F=\frac{1}{\lambda }(F-R_F) = \frac{1}{\lambda }d_F\otimes e_n\), it follows that
Then, together with (5.14), we obtain
which, in view of (5.12) and the uniform boundedness of \((\nabla u_\varepsilon )_\varepsilon \) in \(L^p(Q;\mathbb {R}^n)\), results in
Finally, we combine (5.13) with (5.12) and (5.19) and exploit the continuity of g as a convex function to arrive at (5.11). This concludes the proof of the lower bound. \(\quad \square \)
Remark 5.4
-
(a)
Step 1 can be performed as above for any open and bounded set \(\Omega \), meaning that the restriction to a flat, cross-section Lipschitz domain is not necessary for the construction of a sequence satisfying (5.7).
-
(b)
Note that the recovery sequence constructed in Step 1 can be assumed to have the same boundary values as \(v^F_\varepsilon \), i.e. \(u_{\varepsilon } - v^F_\varepsilon \in W^{1,p}_0(\Omega ;\mathbb {R}^n)\). Indeed, the small-scale oscillations glued onto the laminate \(v^F_\varepsilon \) for sufficiently small \(\varepsilon \) can be adapted outside of \(\{x\in \Omega : {{\,\mathrm{dist}\,}}(x, \partial \Omega )>2\varepsilon \}\) to vanish on \(\{x\in \Omega : {{\,\mathrm{dist}\,}}(x, \partial \Omega )<\varepsilon \}\). This modification affects neither the convergence of \((u_\varepsilon )_\varepsilon \) nor of \((E_\varepsilon (u_\varepsilon ))_\varepsilon \).
Based on the findings of Part II for the affine case, we will now prove the homogenization \(\Gamma \)-convergence result for general limit functions.
Proof of Theorem 5.2 (Part III): General case
Let \(u\in W^{1,p}(\Omega ;\mathbb {R}^n)\) be such that \(u(x) = R(x)x + b(x)\) for \(x\in \Omega \), where \(R \in W^{1,p}(\Omega ;SO(n))\) and \(b \in W^{1,p}(\Omega ;\mathbb {R}^{n})\) satisfy \(\nabla ' R =0\) and \(\nabla ' b =0\). As in the previous parts, we have arranged the arguments in several steps, numbered consecutively.
Step 3: Existence of a recovery sequence We aim to find a sequence \((u_\varepsilon )_\varepsilon \subset W^{1,p}(\Omega ;\mathbb {R}^n)\) such that \(u_\varepsilon \rightharpoonup u\) in \(W^{1,p}(\Omega ;\mathbb {R}^n)\) and \(\limsup _{\varepsilon \rightarrow 0} E_\varepsilon (u_\varepsilon ) \leqq E_{\mathrm{hom}}(u)\). The idea behind the construction of a recovery sequence for u is to use the approximating sequence from Proposition 4.1 and to perturb it in the softer layers by suitably relaxing microstructures that guarantee the optimal energy. To obtain these perturbations, the results from Step 1 (Part II) are applied to piecewise affine approximations of u.
Step 3a: Piecewise constant approximation of\(\nabla u\). Recall that the gradient of u is
First we approximate the functions in (5.20), that is d, \(\partial _n R\), R, and the identity map \(\mathrm{id}_{\mathbb {R}^n}:x\mapsto x\), by simple functions. Indeed, by following standard constructions (e.g. [3, Theorem 1.2]), it is not hard to see that uniform approximation of the continuous function R is possible while preserving the values in SO(n). Without loss of generality, we may assume that all four approximations above have a common partition of \(\Omega \). Due to the globally one-dimensional character of \(d, \partial _n R\) and R, the elements of the partition that do not intersect with \(\partial \Omega \) can be assumed to be cubes aligned with the coordinate axes. To be precise, for every \(\delta >0\) there are finitely many cubes \(Q_\delta ^i\subset \mathbb {R}^n\), which we index by \(I_\delta \), with maximal side length \(\delta \) such that \(|\Omega {\setminus }\bigcup _{i\in I_\delta }Q_\delta ^i|= 0\) and \(Q_\delta ^i\cap \Omega \ne \emptyset \) for \(i\in I_\delta \), and \(d^i_\delta , \xi ^i_\delta \in \mathbb {R}^n\), \(S_\delta ^i\in \mathbb {R}^{n\times n}\), and \(R^i_\delta \in SO(n)\) such that the simple functions defined by
satisfy
Consider the piecewise constant function \(U_\delta \in L^\infty (\Omega ;\mathbb {R}^{n\times n})\) defined by
where \(U_\delta ^i = R_\delta ^i + S_\delta ^i \xi _\delta ^i \otimes e_n+ d_\delta ^i \otimes e_n\in \mathcal {A}\) for \(i \in I_\delta \). Then,
with a constant \(C>0\) independent of \(\delta \). Indeed, in view of (5.21) and (5.20), this is an immediate consequence of the estimate
Step 3b: Locally optimal microstructure By Step 1 (Part II), where recovery sequences in the affine case were established, we can find under consideration of Remark 5.4 a) on each \(Q^i_\delta \cap \Omega \) with \(\delta >0\) and \(i\in I_\delta \) a sequence \((u^i_{\delta , \varepsilon })_\varepsilon \subset W^{1,p}(Q^i_\delta \cap \Omega ;\mathbb {R}^n)\) such that \(\nabla u^i_{\delta ,\varepsilon } \rightharpoonup U_\delta \) in \(L^p(Q^i_\delta \cap \Omega ;\mathbb {R}^{n\times n})\) as \(\varepsilon \rightarrow 0\) and
Now, with \(w_{\delta , \varepsilon }^i:=v^{U^i_{\delta }}_\varepsilon \in W^{1,\infty }(Q^i_\delta \cap \Omega ;\mathbb {R}^n)\) a laminate as introduced in (4.4), let
According to Remark 5.4 b), we may assume that the boundary values of \(u^i_{\delta ,\varepsilon }\) and \(w_{\delta , \varepsilon }^i\) coincide, which entails that \(\varphi _{\delta ,\varepsilon }^i \in W^{1,p}_0(Q^i_\delta \cap \Omega ;\mathbb {R}^n)\). Let us join these local components together in one function \(\varphi _{\delta , \varepsilon }\in W^{1,p}_0(\Omega ;\mathbb {R}^n)\) given by
Note that by construction \(\varphi _{\delta , \varepsilon }=0\) in \(\varepsilon Y_{\mathrm {stiff}}\cap \Omega \). Moreover,
and \(\Vert \nabla \varphi _{\delta , \varepsilon }\Vert _{L^p(\Omega ;\mathbb {R}^{n\times n})}\) is uniformly bounded with respect to \(\varepsilon \) and \(\delta \). In analogy to (5.25) we define for later reference the map of local laminates
Since the homogenized energy density \(W_{\mathrm {hom}}\) satisfies the local Lipschitz condition (5.6) according to Remark 5.3(c), we infer along with (5.23) and Hölder’s inequality that
Summing over all \(i \in I_\delta \) in (5.24) and taking the limit \(\varepsilon \rightarrow 0\) gives that
where \(U_{\varepsilon , \delta } = \sum _{i\in I_\delta } \nabla u^i_{\delta , \varepsilon } \mathbb {1}_{Q^i_{\delta }\cap \Omega }\).
Step 3c: Optimal construction with admissible gradient structure After diagonalization, the functions \(U_{\varepsilon , \delta (\varepsilon )}\) would define a recovery sequence as desired, provided they have gradient structure, i.e., there is a potential \(u_\varepsilon \in W^{1, p}(\Omega ;\mathbb {R}^n)\) with \(\nabla u_\varepsilon = U_{\varepsilon , \delta (\varepsilon )}\). Due to incompatibilities at the interfaces between neighboring cubes, however, this can in general not be expected. To overcome this issue and to obtain an admissible recovery sequence, we discard the local laminates \(w_{\varepsilon , \delta }\) from (5.27), and instead add the locally optimal microstructures \(\varphi _{\delta ,\varepsilon }\) onto the functions \(v_\varepsilon \), which result from Proposition 4.1 applied to u.
More precisely, applying Proposition 4.1 to the given u provides us with an approximating sequence in \(W^{1,p}(\Omega ;\mathbb {R}^n)\) with useful properties, which we call \((v_\varepsilon )_\varepsilon \). In particular,
\(\nabla v_\varepsilon \in SO(n)\) a.e. in \(\varepsilon Y_{\mathrm {stiff}}\cap \Omega \),
with \((\nabla u)_\lambda \) as in (4.5).
Let \(u_{\delta , \varepsilon }\in W^{1,p}(\Omega ;\mathbb {R}^n)\) be given by
Next, we estimate the energetic error brought about by replacing \(w_{\varepsilon , \delta }\) in Step 3b with \(v_\varepsilon \). By (H3), Hölder’s inequality and the above definitions,
with \(C>0\) independent of \(\varepsilon \) and \(\delta \). The first factor in the last line of (5.31) is uniformly bounded (with respect to \(\delta \) and \(\varepsilon \)) as a consequence of (5.29), (5.23) and the remark below (5.26). The second factor can be controlled with the help of (5.30) and the following estimate, which exploits (5.21) and (5.23):
Thus,
Step 3d: Diagonalization As both \(U_{\varepsilon , \delta }\) and \(\nabla u_{\delta , \varepsilon }\) lie in SO(n) almost everywhere on the stiff layers, (5.28) in combination with (5.33), (5.30) and (H2) yields that
Moreover, we derive from (5.29) and (5.26) that \(\nabla u_{\varepsilon , \delta }\rightharpoonup \nabla u\) in \(L^{p}(\Omega ;\mathbb {R}^n)\) as \(\varepsilon \rightarrow 0\) for every \(\delta \). After exploiting Poincaré’s inequality, the compact embedding of \(W^{1,p}\) into \(L^p\), and the Urysohn subsequence principle it follows then that \(u_{\varepsilon , \delta } \rightarrow u\) in \(L^p(\Omega ;\mathbb {R}^n)\) as \(\varepsilon \rightarrow 0\).
Finally, the diagonalization lemma by Attouch (see e.g. [4, Lemma 1.15, Corollary 1.16]) guarantees the existence of a sequence \(\delta (\varepsilon )\) such that \(u_\varepsilon :=u_{\varepsilon , \delta (\varepsilon )}\in W^{1,p}(\Omega ;\mathbb {R}^n)\) satisfies that
and that \(u_\varepsilon \rightarrow u\) in \(L^p(\Omega ;\mathbb {R}^n)\). This shows that \((u_\varepsilon )_\varepsilon \) is a recovery sequence for u as stated.
Step 4: Lower bound. Let \((u_\varepsilon )_\varepsilon \subset W^{1,p}(\Omega ;\mathbb {R}^n)\) be a sequence of uniformly bounded energy, i.e., \(E_\varepsilon (u_\varepsilon ) < C\) for all \(\varepsilon >0\), such that \(u_\varepsilon \rightharpoonup u\) in \(W^{1,p}(\Omega ; \mathbb {R}^n)\) for some \(u\in W^{1,p}(\Omega ; \mathbb {R}^n)\). By Theorem 3.1, \(\nabla u\) has the form (4.1). We will show that
which implies the desired liminf-inequality \(\liminf _{\varepsilon \rightarrow 0}E_\varepsilon (u_\varepsilon )\geqq E_{\mathrm{hom}}(u)\).
To tie this general case to the affine one in Step 2, we adjust to our specific situation a common approximation strategy (see e.g. [47, Theorem 1.3]) based on comparison sequences that involve elements of the constructed recovery sequences. Note that there is no need for the comparison sequence to have full gradient structure, which allows us to argue separately on each piece of the piecewise constant approximation of \(\nabla u\).
Step 4a: Construction of a comparison sequence. First, we approximate \(\nabla u\) by piecewise constant functions \(U_\delta \) as in Step 3a, see (5.22) and (5.23). For \(\varepsilon , \delta >0\) let \(w_{\varepsilon , \delta }\) and \(v_\varepsilon \) be as in Step 3c. Recall that for any \(\delta >0\) and \(i\in I_\delta \),
and that the sequence \((v_\varepsilon )_\varepsilon \subset W^{1,p}(\Omega ;\mathbb {R}^n)\) satisfies (5.29) and (5.30). Moreover,
in view of Proposition 4.1.
Now let us introduce
These functions have vanishing mean value on \(\Omega \) and satisfy \(z_{\delta , \varepsilon }^i=z_{\delta , \varepsilon }|_{Q_\delta ^i} \in W^{1,p}(Q_\delta ^i;\mathbb {R}^n)\) for any \(i\in I_\delta \). Due to (5.35), (5.29) and the assumption on the weak convergence of \((u_\varepsilon )_\varepsilon \), it follows for every \(\delta >0\) that
Hence, as a consequence of the result in the affine case (see Step 2, Part II), applied to the restriction of \(z_{\delta , \varepsilon }\) to any cuboid \(Q^i_\delta \) with \(i\in {\tilde{I}}_\delta :=\{i\in I_\delta : Q_\delta ^i\subset \subset Q\}\), we deduce that
In fact, if
as \(\varepsilon \rightarrow 0\), one can follow the reasoning of Step 2 in Part II to see that even
To verify (5.37) for \(i\in {\tilde{I}}_\delta \), we mimic the arguments leading to (5.15) on the cuboid \(Q_\delta ^i\subset \subset \Omega \). This implies in particular that
Then,
which, in light of (5.39) and (5.36), gives (5.37).
Step 4b: Energy estimates For the homogenized energy, we derive from the local Lipschitz continuity of \(W_{\mathrm{hom}}\) (cf. Remark 5.3(c)), along with (5.23) and Hölder’s inequality, that
Furthermore, with (H2) and the uniform \(L^p\)-bounds on \(\nabla u_\varepsilon \) and \(\nabla w_{\delta , \varepsilon }^i\), we have for any \(i\in I_\delta \),
Due to (5.30), the first expression on the right hand side converges to zero as \(\varepsilon \rightarrow 0\), while the second can be estimated from above by \(\delta \) by (5.32). Considering (5.38), we conclude after summing over \(i\in {\tilde{I}}_\delta \) that
where \(\Omega _\delta = \bigcup _{i\in {\tilde{I}}_\delta } Q_\delta ^i\). Since \(|\Omega {\setminus } \Omega _\delta |\rightarrow 0\) by construction, passing to the limit \(\delta \rightarrow 0\) establishes (5.34), which concludes the proof. \(\quad \square \)
As the next remark shows, the homogenized energy density \(W_\mathrm{hom}\) from (5.4) coincides with the single-cell formula arising from a related model without elasticity (“\(\alpha =\infty \)”) on the stiff layers. This observation indicates that microstructures developing over multiple cells, as they are to be expected in general homogenization problems with non-convex energy densities (cf. [47] and more recently [5]), do not occur. They are indeed inhibited by the presence of the stiff horizontal layers.
Remark 5.5
With \(W_{\mathrm {soft}}\) satisfying (H1)-(H3) and \(W_\mathrm{rig}(F)=\chi _{SO(n)}(F)\) for \(F\in \mathbb {R}^{n\times n}\), let \({\overline{W}}: \Omega \times \mathbb {R}^{n\times n}\rightarrow [0,\infty ]\) be given by
and denote by \({\overline{W}}_{\mathrm{cell}}\) the cell formula associated with \({\overline{W}}\), i.e.,
We will show that for \(F\in \mathbb {R}^{n\times n}\),
Indeed, if \({\overline{W}}_{\mathrm{cell}}(F)<\infty \), there exists \(\psi \in W^{1,p}_{\#}(Y;\mathbb {R}^n)\) such that the expression \(\fint _{Y} W_{\mathrm{rig}}(y, F+\nabla \psi )\,\mathrm {d}y\) is finite. This implies \(F+ \nabla \psi \in SO(n) \) a.e. in \(Y_{\mathrm {stiff}}\), and we infer from Reshetnyak’s theorem [53] (cf. also Theorem 3.3) that for some \(R \in SO(n)\),
Therefore, since \(\psi \) is periodic, one obtains for \(i=1, \ldots , n-1\) that
and hence, \(F\in \mathcal {A}\) and in particular, \(F=R+d\otimes e_n\) with \(d\in \mathbb {R}^n\). By (5.41), \(\nabla \psi = - d\otimes e_n\) on \(Y_{\mathrm {stiff}}\).
Considering the piecewise affine function \(v \in W^{1, \infty }_{\#}(Y;\mathbb {R}^n)\) with zero mean value and gradient
we can find \(\varphi \in W^{1,p}_{\#}(Y;\mathbb {R}^n)\) such that \(\nabla \varphi =0\) in \(Y_{\mathrm {stiff}}\) and \(\psi \) is represented as \(\psi = v+\varphi \). Thus,
where the last equality makes use of Remark 5.3 b). This verifies (5.40).
References
Acerbi, E., Chiadò Piat, V., Dal Maso, G., Percivale, D.: An extension theorem from connected sets, and homogenization in general periodic domains. Nonlinear Anal. 18(5), 481–496, 1992
Alt, H.W.: Linear Functional Analysis. Universitext. Springer, London. An application-oriented introduction, Translated from the German edition by Robert Nürnberg 2016
Amann, H., Escher, J.: Analysis. II. Birkhäuser Verlag, Basel. Translated from the 1999 German original by Silvio Levy and Matthew Cargo 2008
Attouch, H.: Variational Convergence for Functions and Operators. Applicable Mathematics Series. Pitman (Advanced Publishing Program), Boston 1984
Barchiesi, M., Gloria, A.: New counterexamples to the cell formula in nonconvex homogenization. Arch. Ration. Mech. Anal. 195(3), 991–1024, 2010
Barchiesi, M., Lazzaroni, G., Zeppieri, C.I.: A bridging mechanism in the homogenization of brittle composites with soft inclusions. SIAM J. Math. Anal. 48(2), 1178–1209, 2016
Braides, A.: Homogenization of some almost periodic coercive functional. Rend. Accad. Naz. Sci. XL Mem. Mat. (5)9(1), 313–321, 1985
Braides, A.: \(\Gamma \)-Convergence for Beginners. Number 22 in Oxford Lecture Series in Mathematics and Its Applications, 1st edn. Oxford University Press, Oxford 2005
Braides, A., Fonseca, I., Leoni, G.: \({\cal{A}}\)-quasiconvexity: relaxation and homogenization. ESAIM Control Optim. Calc. Var. 5, 539–577, 2000
Braides, A., Garroni, A.: Homogenization of periodic nonlinear media with stiff and soft inclusions. Math. Models Methods Appl. Sci. 5(4), 543–564, 1995
Cagnetti, F., Scardia, L.: An extension theorem in SBV and an application to the homogenization of the Mumford–Shah functional in perforated domains. J. Math. Pures Appl. (9)95(4), 349–381, 2011
Chambolle, A., Giacomini, A., Ponsiglione, M.: Piecewise rigidity. J. Funct. Anal. 244(1), 134–153, 2007
Cherdantsev, M., Cherednichenko, K., Neukamm, S.: High contrast homogenisation in nonlinear elasticity under small loads. Asymptot. Anal. 104(1–2), 67–102, 2017
Cherdantsev, M., Cherednichenko, K.D.: Two-scale \(\Gamma \)-convergence of integral functionals and its application to homogenisation of nonlinear high-contrast periodic composites. Arch. Ration. Mech. Anal. 204(2), 445–478, 2012
Christowiak, F.: Homogenization of Layered Materials with Stiff Components. Ph.D. thesis, Universität Regensburg, 2018
Christowiak, F., Kreisbeck, C.: Homogenization of layered materials with rigid components in single-slip finite crystal plasticity. Calc. Var. Partial Differ. Equ. 56(3), 75, 2017
Cioranescu, D., Damlamian, A., De Arcangelis, R.: Homogenization of integrals with pointwise gradient constraints via the periodic unfolding method. Ric. Mat. 55(1), 31–53, 2006
Cioranescu, D., Donato, P.: An Introduction to Homogenization, Volume 17 of Oxford Lecture Series in Mathematics and Its Applications. The Clarendon Press, Oxford University Press, New York 1999
Conti, S.: Low-Energy Deformations of Thin Elastic Plates: Isometric Embeddings and Branching Patterns. Habilitation thesis, University of Leipzig, 2003
Conti, S., Dolzmann, G.: On the theory of relaxation in nonlinear elasticity with constraints on the determinant. Arch. Ration. Mech. Anal. 217(2), 413–437, 2015
Conti, S., Dolzmann, G., Kreisbeck, C.: Relaxation of a model in finite plasticity with two slip systems. Math. Models Methods Appl. Sci. 23(11), 2111–2128, 2013
Conti, S., Dolzmann, G., Müller, S.: Korn’s second inequality and geometric rigidity with mixed growth conditions. Calc. Var. Partial Differ. Equ. 50(1–2), 437–454, 2014
Conti, S., Schweizer, B.: Rigidity and \(Gamma\)-convergence for solid-solid phase transitions with SO(2) invariance. Commun. Pure Appl. Math. 59(6), 830–868, 2006
Conti, S., Theil, F.: Single-slip elastoplastic microstructures. Arch. Ration. Mech. Anal. 178(1), 125–148, 2005
Dacorogna, B.: Direct Methods in the Calculus of Variations, Volume 78 of Applied Mathematical Sciences, 2nd edn. Springer, New York 2008
Dal Maso, G.: An Introduction to \(\Gamma \)-Convergence. Number 8 in Progress in Nonlinear Differential Equations and Their Applications. Birkhäuser, Boston 1993
Davoli, E., Ferreira, R., Kreisbeck, C.: Homogenization in BV of a model for layered composites in finite crystal plasticity. Preprint arXiv:1901.11517, 2019
De Giorgi, E.: Sulla convergenza di alcune successioni d’integrali del tipo dell’ area. Rendiconti di Matematica e delle sue applicazioni8(6), 277–294, 1975. Collection of articles dedicated to Mauro Picone on the occasion of his ninetieth birthday
De Giorgi, E., Franzoni, T.: Su un tipo di convergenza variazionale. Atti Accad. Naz. Lincei Rend. Cl. Sci. Fis. Mat. Natur. (8)58(6), 842–850, 1975
Duerinckx, M., Gloria, A.: Stochastic homogenization of nonconvex unbounded integral functionals with convex growth. Arch. Ration. Mech. Anal. 221(3), 1511–1584, 2016
Faraco, D., Zhong, X.: Geometric rigidity of conformal matrices. Ann. Sc. Norm. Super. Pisa Cl. Sci. (5)4(4), 557–585, 2005
Friedrich, M., Schmidt, B.: A quantitative geometric rigidity result in SBD. Preprint arXiv:1503.06821, 2015
Friesecke, G., James, R.D., Müller, S.: A theorem on geometric rigidity and the derivation of nonlinear plate theory from three-dimensional elasticity. Commun. Pure Appl. Math. 55(11), 1461–1506, 2002
Friesecke, G., James, R.D., Müller, S.: A hierarchy of plate models derived from nonlinear elasticity by \(\Gamma \)-convergence. Arch. Ration. Mech. Anal. 180(2), 183–236, 2006
Gilbarg, D., Trudinger, N.S.: Elliptic Partial Differential Equations of Second Order. Classics in Mathematics, 2nd edn. Springer, Berlin 2001
Gurtin, M.E.: An Introduction to Continuum Mechanics. Number 158 in Mathematics in Science and Engineering. Academic Press, New York 1981
Hanche-Olsen, H., Holden, H.: The Kolmogorov–Riesz compactness theorem. Expo. Math. 28(4), 385–394, 2010
Iwaniec, T., Martin, G.: The Liouville theorem. In: Cazacu, C.A., Lehto, E.O. (eds.) Analysis and Topology, pp. 339–361. World Scientific Publishing, River Edge 1998
Iwaniec, T., Sbordone, C.: Caccioppoli estimates and very weak solutions of elliptic equations. Atti Accad. Naz. Lincei Cl. Sci. Fis. Mat. Natur. Rend. Lincei (9) Mat. Appl.14(3), 189–205, 2004, 2003. Renato Caccioppoli and modern analysis
Koumatos, K., Rindler, F., Wiedemann, E.: Differential inclusions and Young measures involving prescribed Jacobians. SIAM J. Math. Anal. 47(2), 1169–1195, 2015
Kowalsky, H.-J., Michler, G.: Lineare Algebra. De Gruyter Lehrbuch. De Gruyter, Berlin, 2003. 12te Auflage.
Kreisbeck, C., Krömer, S.: Heterogeneous thin films: combining homogenization and dimension reduction with directors. SIAM J. Math. Anal. 48(2), 785–820, 2016
Le Dret, H., Raoult, A.: The quasiconvex envelope of the Saint Venant–Kirchhoff stored energy function. Proc. R. Soc. Edinb. Sect. A125(6), 1179–1192, 1995
Leoni, G.: A First Course in Sobolev Spaces, Volume105 of Graduate Studies in Mathematics. American Mathematical Society, Providence 2009
Lewicka, M., Pakzad, M.R.: Scaling laws for non-Euclidean plates and the \(W^{2,2}\) isometric immersions of Riemannian metrics. ESAIM Control Optim. Calc. Var. 17(4), 1158–1173, 2011
Marcellini, P.: Periodic solutions and homogenization of nonlinear variational problems. Ann. Mat. Pura Appl. 4(117), 139–152, 1978
Müller, S.: Homogenization of nonconvex integral functionals and cellular elastic materials. Arch. Ration. Mech. Anal. 99(3), 189–212, 1987
Müller, S.: Homogenization of nonconvex integral functionals and cellular elastic materials. Arch. Ration. Mech. Anal. 99(3), 189–212, 1987
Müller, S.: Variational models for microstructure and phase transitions. In: Hildebrandt, S., Struwe, M. (eds.) Calculus of Variations and Geometric Evolution Problems. Lecture Notes in Mathematics, vol. 1713, pp. 85–210. Springer, Berlin 1999
Müller, S., Scardia, L., Zeppieri, C.I.: Geometric rigidity for incompatible fields, and an application to strain-gradient plasticity. Indiana Univ. Math. J. 63(5), 1365–1396, 2014
Neukamm, S., Schäffner, M.: Quantitative homogenization in nonlinear elasticity for small loads. Arch. Ration. Mech. Anal. 230(1), 343–396, 2018
Pratelli, A.: On the bi-Sobolev planar homeomorphisms and their approximation. Nonlinear Anal. 154, 258–268, 2017
Rešetnjak, J.G.: Liouville’s conformal mapping theorem under minimal regularity hypotheses. Sibirsk. Mat. Ž. 8, 835–840, 1967
Rudin, W.: Principles of Mathematical Analysis. International Series in Pure and Applied Mathematics, 3rd edn. McGraw-Hill, New York 1976
Acknowledgements
The authors would like to thank Georg Dolzmann for his valuable comments on FC’s Ph.D. thesis, which helped to improve also the presentation of this manuscript. FC gratefully acknowledges a traveling grant by the DFG Graduiertenkolleg 1692 “Curvature, Cyles, and Cohomology”. CK was partially supported by a Westerdijk Fellowship from Utrecht University.
Author information
Authors and Affiliations
Corresponding author
Additional information
Communicated by I. Fonseca
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendix A. Collected Auxiliary Results
Appendix A. Collected Auxiliary Results
In the next lemma, we provide a type of reverse Poincaré inequality for special affine maps given as the difference of two rotations on a domain that is thin in one dimension. The special feature of this result (e.g. in comparison with classical Caccioppoli estimates for harmonic maps [39]) is that the constant can be chosen independently of the thickness of the domain in the \(e_n\)-direction.
Lemma A.1
For an integer \(n\geqq 2\) let \(P=O \times I\) with \(O\subset \mathbb {R}^{n-1}\) an open cube of side length \(l>0\) and \(I\subset \mathbb {R}\) an interval of length \(h>0\), and let \(1\leqq p<\infty \). Then there exists a constant \(C>0\) depending only on n and p such that for all rotations \(R_1, R_2\in SO(n)\) and translation vectors \(d\in \mathbb {R}^n\),
Proof
We will prove the result for \(p=1\), for general p the statement then follows immediately from Hölder’s inequality.
Moreover, without loss of generality let \(R_2\) be the identity matrix \(\mathbb {I}=\mathbb {I}_n\in \mathbb {R}^{n\times n}\). We set \(R=R_1\in SO(n)\) and write \(A:= \mathbb {I}-R\in \mathbb {R}^{n\times n}\). Let \({\overline{P}}\) denote the translation of the open cuboid P centered in the origin. The arguments below make use of the nested sets \(\widehat{P}\subset Z\subset {\overline{P}}\), where Z is the cylinder with circular cross section inscribed in \({\overline{P}}\) and \(\widehat{P}\) is the largest centered, open cuboid contained in Z. Precisely,
where \(B^{n-1}_r\) the \((n-1)\)-dimensional ball around the origin with radius r.
With this notation in place, we observe that
To derive the desired estimate, we determine the singular values of A. It follows from the specific structure of A that
Considering that every \(R\in SO(n)\) can be transformed into block diagonal form with the help of another rotation \(U\in SO(n)\), i.e. there is an integer \(k\leqq \frac{n}{2}\) and two-dimensional rotations \(\Theta _1, \ldots , \Theta _{k} \in SO(2)\) such that
see e.g. [41, Satz 8.3.10], we conclude from the fact that the symmetric part of a two-dimensional rotation matrix is diagonal that \(A^TA = U^T D U\), where
with \(\theta _i=(\Theta _i)_{11}\in [-1,1)\). One may assume without loss of generality that \(\theta _1\leqq \theta _2\leqq \cdots \leqq \theta _k\), which implies that \(2(1-\theta _1)\) is the largest eigenvalue of \(A^TA\), and hence corresponds to the squared spectral norm of A. Since all norms on \(\mathbb {R}^{n\times n}\) are equivalent, there is a constant \(C=C(n)>0\) such that \(\sqrt{2(1-\theta _1)}\geqq C |A|\), where \(|\cdot |\) denotes the Frobenius norm. Hence,
In view of (A.1) and (A.2) it remains to show that
with \(C>0\) depending only on n. If \(U=\mathbb {I}\), we simply neglect one of the two additive terms in the integrand, say \(x_2^2\), and estimate that
For general U, our argument requires to select a suitable rotation of the plane spanned by the unit vectors \(e_1\) and \(e_2\) to guarantee that the axes of the rotated cylinder UZ is orthogonal to \(e_1\). More precisely, one observes that any planar rotation \(S=\mathrm{diag}(\Sigma , \mathbb {I}_{n-2})\) with \(\Sigma \in SO(2)\) leaves the integral in (A.3) unchanged, and therefore
Since the intersection of \(\mathrm{span}\{e_1, e_2\}\) with the \((n-1)\)-dimensional orthogonal complement of \(\mathrm{span}\{Ue_n\}\) is at least a one-dimensional subspace, we can choose a planar rotation S such that \(Ue_n\cdot S^Te_1=0\), and thus \((SU)^Te_1\cdot e_n=0\). Then there exists \(Q = \mathrm{diag}(\Xi , \mathbb {I}_1)\in SO(n)\) with \(\Xi \in {{\,\mathrm{SO}\,}}(n-1)\) such that \(Q^Te_1=(SU)^Te_1\), and
where we have used the invariance of the cylinder Z under rotations that leave the \(x_n\)-component unaffected. In view of (A.5) and (A.4) this shows (A.3), and hence, finishes the proof. \(\quad \square \)
Next we give details on the extension result for locally one-dimensional functions in the \(e_n\)-direction used in Sections 3 and 4. Recall that for a bounded domain \(\Omega \subset \mathbb {R}^n\), the smallest open cuboid containing \(\Omega \) is denoted by \(Q_\Omega \) and \(Q_\Omega = O_\Omega \times J_\Omega \) with \(O_\Omega \subset \mathbb {R}^{n-1}\) an open cuboid and an open interval \(J_\Omega \subset \mathbb {R}\).
Lemma A.2
Let \(\Omega \subset \mathbb {R}^n\) be a bounded, flat and cross-section connected Lipschitz domain. If \(v\in W^{1, p}(\Omega ;\mathbb {R}^m)\) satisfies \(\nabla ' v=0\), then v can be extended to \(Q_\Omega \) by a globally one-dimensional function in the \(e_n\)-direction \(\tilde{v}\in W^{1,p}(Q_\Omega ;\mathbb {R}^m)\cap C^0(Q_\Omega ;\mathbb {R}^m)\).
In particular, one can identify v with the one-dimensional function \(\nu \in W^{1,p}(J_\Omega ;\mathbb {R}^m)\) defined by the identity \({\tilde{v}}(x)=\nu (x_n)\) for \(x\in Q_\Omega \).
Proof
As pointed out at the beginning of Section 3, v is locally one-dimensional in the \(e_n\)-direction, and hence, locally constant on any non-empty cross section \(\Omega _t=H_t\cap \Omega = \{x\in \mathbb {R}^n: x_n=t\}\cap \Omega \). Since the latter are connected by assumption, it follows that v is also globally one-dimensional in the \(e_n\)-direction.
We can now define an extension \(\tilde{v}\) of v to \(Q_\Omega \) by setting
for \(x\in Q_\Omega \). Observe that with \(Q_\Omega \) the smallest open cuboid such that \(\Omega \subset Q_{\Omega }\), the intersection \(H_{x_n}\cap \Omega =\Omega _{x_n}\) is non-empty for all \(x\in \Omega \). Clearly, \(\tilde{v}\) is globally one-dimensional in the \(e_n\)-direction by definition. It therefore remains to prove that \(\tilde{v}\in W^{1,p}(Q_\Omega ;\mathbb {R}^m)\) (for continuity one can then argue as in the first paragraph of Section 3).
To see this we will construct a sequence \(w_j\in C^\infty (\overline{Q_\Omega };\mathbb {R}^m)\) that approximates \(\tilde{v}\) in \(W^{1,p}(Q_{\Omega };\mathbb {R}^m)\). Let \(J_\Omega =(a, b)\) with \(a, b\in \mathbb {R}\), \(a<b\). Since \(\Omega \) is a flat Lipschitz domain there exist \(x_a\in \Omega _a\) and \(x_b\in \Omega _b\) and balls \(B_r(x_a)\) and \(B_r(x_b)\) with radius \(r>0\) such that \(B_r(x_a)\cap Q_\Omega \subset \Omega \) and \(B_r(x_b)\cap Q_{\Omega }\subset \Omega \). Exploiting further that \(\Omega \) is open and connected, hence also path-connected, we can connect the edge points \(x_a\) with \(x_b\) by a \(C^1\)-curve \(\gamma \) (after smoothing of a continuous curve). Moreover, one can be chosen \(\gamma \) to be monotone in \(x_n\) due to the cross-section connectedness of \(\Omega \) and even strictly monotone, which implies that \(\gamma \) is a regular curve, considering that \(\Omega \) is open. After reparametrization we obtain
For the composition \(w=v\circ \gamma \in W^{1,p}(J_\Omega ;\mathbb {R}^m)\) there exist approximating functions \(w_j\in C^\infty (\overline{J_\Omega };\mathbb {R}^m)\) such that \(w_j\rightarrow w\) in \(W^{1,p}(a,b)\). Without changing notation, let us identify \(w_j\) and w with their constant expansion in \(x'\), that is with elements in \(W^{1,p}(Q_\Omega ;\mathbb {R}^m)\) and \(C^\infty (\overline{Q_\Omega };\mathbb {R}^m)\), respectively. Finally, in view of (A.6) and (A.7),
which shows that \(\tilde{v}\in W^{1,p}(Q_\Omega ;\mathbb {R}^m)\) and concludes the proof. \(\quad \square \)
Remark A.3
-
(a)
Since only local arguments have been used in the proof above, Lemma A.2 still holds if \(W^{1,p}(\Omega ;\mathbb {R}^m)\) is replaced with \(W^{1,p}_{\mathrm{loc}}(\Omega ;\mathbb {R}^m)\). In this case, it is even enough to require that \(\Omega \subset \mathbb {R}^n\) is a bounded, cross-section connected domain.
-
(b)
As Lemma A.2 relies on constant extensions only, changing the codomain of v from \(\mathbb {R}^m\) to SO(n) does not change the statement.
Rights and permissions
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.
About this article
Cite this article
Christowiak, F., Kreisbeck, C. Asymptotic Rigidity of Layered Structures and Its Application in Homogenization Theory. Arch Rational Mech Anal 235, 51–98 (2020). https://doi.org/10.1007/s00205-019-01418-0
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00205-019-01418-0