Singular perturbations models in phase transitions for anisotropic higher-order materials

Giuseppe Cosma Brusca, Davide Donati, and Chiara Trifone SISSA via Bonomea 265 34136 Trieste, Italy · 2025

Read Singular perturbations models in phase transitions for anisotropic higher-order materials by Giuseppe Cosma Brusca, Davide Donati, and Chiara Trifone SISSA via Bonomea 265 34136 Trieste, Italy (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2503-13035.txt.

Introduction

Giuseppe Brusca, Davide Donati, and Chiara Trifone study phase transitions in a twenty twenty-five paper. Their model uses a double-well potential, perturbed by terms that involve derivatives of high order. The materials are anisotropic, meaning they behave differently by direction. The question is what the model does as a small parameter shrinks toward zero.

They use Gamma-convergence to track that limit. Some coefficients in the model may be negative, which breaks the usual bounds. The authors overcome this by proving a nonlinear interpolation inequality. With it, they gain a compactness result for the functionals. Readers in analysis get a model that allows high-order and sign-changing effects in phase change problems.

Catalog record

CREATORGiuseppe Cosma Brusca, Davide Donati, and Chiara Trifone SISSA via Bonomea 265 34136 Trieste, Italy
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE147 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2503-13035

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Singular perturbations models in phase transitions for anisotropic higher-order materials Thanks: Email addresses: [email protected], [email protected], [email protected] Giuseppe Cosma Brusca, Davide Donati, and Chiara Trifone SISSA via Bonomea 265 34136 Trieste, Italy Abstract . We discuss a model for phase transitions in which a double-well potential is singularly perturbed by possibly several terms involving different, arbitrarily high orders of derivation. We study by Γ \Gamma -convergence the asymptotic behaviour as ε → 0 \varepsilon\to 0 of the functionals F ε ​ ( u ) := ∫ Ω [ 1 ε ​ W ​ ( u ) + ∑ ℓ = 1 k q ℓ ​ ε 2 ​ ℓ − 1 ​ | ∇ ( ℓ ) u | ℓ 2 ] ​ 𝑑 x , u ∈ H k ​ ( Ω ) , F_{\varepsilon}(u):=\int_{\Omega}\Bigl[\frac{1}{\varepsilon}W(u)+\sum_{\ell=1}^{k}q_{\ell}\varepsilon^{2\ell-1}|\nabla^{(\ell)}u|_{\ell}^{2}\Bigr]\,dx,\qquad u\in H^{k}(\Omega), for fixed k > 1 k>1 integer, addressing also the case in which the coefficients q 1 , … , q k − 1 q_{1},...,q_{k-1} are negative and | ⋅ | ℓ |\cdot|_{\ell} is any norm on the space of symmetric ℓ \ell -tensors for each ℓ ∈ { 1 , … , k } \ell\in\{1,...,k\} . The negativity of the coefficients leads to the lack of a priori bounds on the functionals; such issue is overcome by proving a nonlinear interpolation inequality. With this inequality at our disposal, a compactness result is achieved by resorting to the recent paper [ 10 ] . A further difficulty is the presence of general tensor norms which carry anisotropies, making standard slicing arguments not suitable. We prove that the Γ \Gamma -limit is finite only on sharp interfaces and that it equals an anisotropic perimeter, with a surface energy density described by a cell formula. MSC 2020 codes: 49J45, 26B30, 74N15, 74G65, Keywords: Phase transitions, Singular perturbations, Higher-order derivatives, Γ \Gamma -convergence, Interpolation inequalities. 1. Introduction In the recent paper [ 10 ] , it is studied a model for phase transitions in which a double-well potential is perturbed by a singular term involving derivatives of high order. In a one-dimensional setting, this model is described by the family of functionals defined for \varepsilon>0 as (1) E^{1}_{\varepsilon}(u,I):=\begin{cases}\displaystyle\int_{I}\Bigl[\frac{1}{\varepsilon}W(u)+\varepsilon^{2k-1}|u^{(k)}|^{2}\Bigr]\,dt&\text{ if }u\in H^{k}(I),\\ +\infty&\text{ if }u\in L^{2}(I)\setminus H^{k}(I),\end{cases} where I is a bounded, open interval, W is a non-negative double-well potential, and k is a fixed positive integer that can be assumed to be strictly larger than 2 , the cases k=1 and k=2 being already studied by Modica and Mortola in [ 21 ] (see also [ 20 ] ) and Fonseca and Mantegazza in [ 17 ] , respectively. As customary for phase transition models, the study of the functionals ( 1 ) is divided in two parts: first, compactness properties are investigated as \varepsilon tends to 0 , here it is proved that a family of functions having equi-bounded energy has a sharp interface as a cluster point; then, the asymptotic analysis is performed, in this case, by \Gamma -convergence. Under some mild assumptions on the potential W , and, in particular, without requiring any growth condition on W at infinity, it is proved that (see Theorem 4.1 below for the precise statement) given \{u_{\varepsilon}\}_{\varepsilon}\subset H^{k}(I) satisfying \sup_{\varepsilon}E^{1}_{\varepsilon}(u_{\varepsilon},I)<+\infty , there exists a function u\in BV(I;\{-1,1\}) such that, upon extracting a subsequence, u_{\varepsilon}\to u in measure as \varepsilon\to 0 , having set the two wells of W at -1 and 1 . Hence, the \Gamma -limit as \varepsilon\to 0 computed with respect to the convergence in measure is finite only on BV(I;\{-1,1\}) and it is proved to be equal to m_{k}\#(S(u)\cap I) , where S(u) denotes the set of the discontinuity points of u and m_{k} is a strictly positive constant determined by the optimal-profile problem (2) m_{k}:=\inf\Bigl\{\int_{-\infty}^{+\infty}\Bigl[W(v)+|v^{(k)}|^{2}\Bigr]\,dt:v\in H^{k}_{\text{loc}}(\mathbb{R}),\lim_{t\to\pm\infty}v(t)=\pm 1\Bigr\}. Note that such result generalizes the one by Fonseca and Mantegazza for k=2 to a wider class of double-well potentials. The precompactness of a family with equi-bounded energy \{u_{\varepsilon}\}_{\varepsilon} is obtained by adapting an argument originally devised by Solci [ 26 ] for perturbations giving free-discontinuity functionals, which relies on local interpolation inequalities on intervals that allow to estimate the L^{2} -norm of an intermediate derivative u_{\varepsilon}^{(\ell)},\ell\in\{1,...,k-1\} , in terms of the integral of W(u_{\varepsilon}) , which is related to the L^{2} -norm of u_{\varepsilon}-1 or u_{\varepsilon}+1 assuming that W is quadratic close to the bottom of the wells, and the L^{2} -norm of u_{\varepsilon}^{(k)} . In this way, small transition intervals on which u_{\varepsilon} ‘passes from a well to the other’ are detected, with the further condition that, at the endpoints, all the intermediate derivatives are small. These observations lead to considering the minimum problem \displaystyle m_{k}(\eta,T) \displaystyle:=\inf\Bigl\{\int_{-T}^{T}\Bigl[W(v)+|v^{(k)}|^{2}\Bigr]\,dt:v\in H^{k}((-T,T)), \displaystyle\qquad|v(-T)+1|<\eta,|v(T)-1|<\eta,|v^{(\ell)}(\pm T)|<\eta\text{ for all }\ell\in\{1,...,k-1\}\Bigr\} that, once optimized in \eta and T , yields ( 2 ). Since, in turn, ( 2 ) is proved to be strictly positive, the equi-boundedness of the energy implies that the number of the aforementioned intervals on which a transition occurs is equi-bounded as well, and this property, in broad terms, yields compactness. Having proved that for every \varepsilon each transition interval increases the energy E_{\varepsilon}^{1} by the fixed amount m_{k} , which is independent of \varepsilon , the computation of the \Gamma -limit follows by standard arguments. The general d -dimensional case deals with the functionals (3) E_{\varepsilon}(u,\Omega):=\begin{cases}\displaystyle\int_{\Omega}\Bigl[\frac{1}{\varepsilon}W(u)+\varepsilon^{2k-1}\|\nabla^{{(k)}}u\|_{k}^{2}\Bigr]\,dx&\text{ if }u\in H^{k}(\Omega),\\ +\infty&\text{ if }u\in L^{2}(\Omega)\setminus H^{k}(\Omega),\end{cases} where \Omega is a bounded, open subset of \mathbb{R}^{d} with Lipschitz boundary, W and k are as above, and \|\cdot\|_{k} is the operatorial norm of a symmetric k -tensor defined as (4) \|T\|_{k}:=|\sup\{T(\,\underbrace{\xi,...,\xi}_{k\text{ times}}\,):|\xi|=1\}|. In [ 10 ] , the study of functionals ( 3 ) is confined to the computation of the \Gamma -limit, and it is remarked that the compactness theorem holds true assuming suitable growth conditions at infinity on W . The choice of the operatorial norm makes the asymptotic analysis of such functionals a one-dimensional task since slicing arguments are well suited in this case. Indeed, ( 4 ) and the slicing properties of Sobolev functions imply that (5) \|\nabla^{(k)}v(x)\|_{k}\geq\Bigl|\frac{d^{k}v^{\xi,y}}{dt^{k}}(t)\Bigr|,\qquad\text{ for }\mathcal{L}^{d}\text{-a.e.\ }x\in\Omega, where \xi is a unit vector, y\in\{z\in\mathbb{R}^{d}:z\cdot\xi=0\} , and we set x=y+t\xi and v^{\xi,y}(t):=v(y+t\xi) . Employing the blow-up argument of Fonseca and Müller [ 18 ] together with ( 5 ) and applying the \Gamma -limit result previously obtained in dimension 1 , it is proved that the \Gamma -liminf is estimated by (6) m_{k}P(\{u=1\};\Omega), where P(\{u=1\};\Omega) is the perimeter of the set \{u=1\} in the domain \Omega , representing the measure of the interface in higher dimension. Finally, the construction of a recovery sequence, that by density can be confined to the case of a smooth interface \partial\{u=1\}\cap\Omega=\partial E\cap\Omega for some E smooth, open set, is obtained by composing the signed distance function from \partial E with an (almost) optimal-profile for the one-dimensional problem ( 2 ) suitably dilated at scale \varepsilon . In this way, the energy of such sequence can be computed by slicing in direction of the normal field to \partial E\cap\Omega and, since for every \varepsilon each slice increases the energy by the fixed amount m_{k} , ( 6 ) is proved to be the \Gamma -limit. In this paper, we study a more general family of functionals in which possibly

Read the complete text