Home Quasiconvex bulk and surface energies: C1,α regularity
Article Open Access

Quasiconvex bulk and surface energies: C1,α regularity

  • Menita Carozza EMAIL logo , Luca Esposito and Lorenzo Lamberti
Published/Copyright: August 29, 2024

Abstract

We establish regularity results for equilibrium configurations of vectorial multidimensional variational problems, involving bulk and surface energies. The bulk energy densities are uniformly strictly quasiconvex functions with p -growth, p 2 , without any further structure conditions. The anisotropic surface energy is defined by means of an elliptic integrand Φ not necessarily regular. For a minimal configuration ( u , E ) , we prove partial Hölder continuity of the gradient u of the deformation.

MSC 2010: 49N15; 49N60; 49N99

1 Introduction and statements

In this article, we study multidimensional vectorial variational problems involving bulk and surface energies, mainly related to problems issuing from material science and computer vision. Namely, we deal with regularity properties of solutions to such problems. The model problem

(1.1) Ω σ E ( x ) u 2 d x + P ( E , Ω ) ,

where σ E ( x ) a 1 E + b 1 Ω \ E , 0 < a < b , with E Ω R n , and P ( E , Ω ) stands for the perimeter of the set E in Ω , dates back to the works of Ambrosio and Buttazzo and Lin. In [4,33,35], the authors proved the existence and regularity for minimal configurations ( u , E ) of (1.1) in the scalar case. Furthermore, Lin and Kohn [36] treated more general Dirichlet energies in the following:

(1.2) ( u , E ) Ω ( F ( x , u , u ) + 1 E G ( x , u , u ) ) d x + P ( E , Ω ) ,

with the constraints

u = Ψ on Ω and E = d ,

Ψ H 1 ( Ω ) , 0 < d < Ω . They considered F and G convex functions growing quadratically on the gradient, and satisfying restrictive structure assumptions.

In the cited articles, it is proved C 0 , α regularity for minimizers u in Ω and some estimates on the singular set of E are given. More precisely, define the set of regular points of E as follows:

Reg ( E ) { x E Ω : E is a C 1 , γ hypersurface in B ε ( x ) , for some ε > 0 and γ ( 0 , 1 ) } ,

where B ε ( x ) denotes the ball of center x and radius ε , and, accordingly, the set of singular points of E

Σ ( E ) ( E Ω ) \ Reg ( E ) ,

then n 1 ( Σ ( E ) ) = 0 , whereas ( u , E ) minimizes the functional (1.2).

More recently, De Philippis and Figalli [17] and Fusco and Julin [27], improved this result showing that minimal configurations of Dirichlet-type functional (1.1) satisfy

dim ( Σ ( E ) ) n 1 ε ,

for some ε > 0 depending only on a and b . The same kind of estimate for the singular set Σ ( E ) has been proved in [23,24] for the more general Dirichlet functional (1.2).

The case of general functionals with p -growth on the gradient is still not completely understood especially regarding the regularity of the free interface E and the dimension of the singular set Σ ( E ) . A first step in this direction has been done in [10], where the authors deal with constrained convex scalar problems, without structure assumptions on the bulk energies. They prove C 0 , α regularity for minimizers u , but they do not give estimates for the singular set, which is an issue still unsolved, under the assumption of p -growth (see also contributions due to [21] and [34]). Nevertheless, as originally stressed in [22], for minimizers of Dirichlet functional (1.1), the exponent α , relative to the C 0 , α regularity of minimizers u , can be affected by a closeness assumption on the coefficients a and b of σ E ( x ) appearing in (1.1). More precisely, in [22], it has been proved that the hypothesis 1 a b < γ n ensures that u C 0 , 1 2 + ε , for some ε > 0 . Exploiting this information, the regularity of the boundary E can be easily achieved by managing the bulk term as a perturbative term, since, by virtue of C 0 , 1 2 + ε regularity, the bulk term is asymptotically smaller than the perimeter term. Therefore, a well-known regularity result can be invoked for almost minimal perimeter minimizers due to Tamanini [40], thus proving the regularity of the free boundary E .

Differently from the scalar case, in the vectorial setting, only a few regularity results for minimizers of integral functionals involving both bulk and interfacial energies are available in the literature. According to our knowledge, the only articles dealing with the vectorial case are [6] and [11]. In [6], the regularity for vector-valued free interface variational problems is treated within the context of k th-order homogeneous partial differential operators A (for a detailed study of A -quasiconvexification, see [28,29]). Carozza et al. [11] studied minimal configurations of energy of the type

(1.3) ( v , A ) Ω ( F ( D v ) + 1 A G ( D v ) ) d x + P ( A , Ω ) ,

where v W loc 1 , p ( Ω ; R N ) and F , G : R n × N R are the C 2 integrands, satisfying, for p > 1 and for positive constants 1 , 2 , L 1 , L 2 > 0 , the following growth and uniformly strict p -quasiconvexity conditions:

(F1) 0 F ( ξ ) L 1 ( 1 + ξ 2 ) p 2 ,

(F2) Ω F ( ξ + D φ ) d x Ω F ( ξ ) + 1 D φ 2 ( 1 + D φ 2 ) p 2 2 d x ,

(G1) 0 G ( ξ ) L 2 ( 1 + ξ 2 ) p 2 ,

(G2) Ω G ( ξ + D φ ) d x Ω G ( ξ ) + 2 D φ 2 ( 1 + D φ 2 ) p 2 2 d x ,

for every ξ R n × N and φ C 0 1 ( Ω ; R N ) .

Under these assumptions, the authors proved the existence of local minimizers for the functional (1.3), for any p > 1 . Furthermore, they proved a partial C 1 , α regularity result for minimal configurations in the quadratic case p = 2 .

In this article, we generalize the results given in [11] under two viewpoints. First, we treat the more general case of p -growth with p 2 . Moreover, we deal with anisotropic surface energies.

In the rest of this article, we focus our attention on integral functionals defined as follows:

(1.4) ( v , A ) Ω ( F ( D v ) + 1 A G ( D v ) ) d x + Ω * A Φ ( x , ν A ( x ) ) d n 1 ( x ) ,

where A Ω is a set of finite perimeter, u W loc 1 , p ( Ω ; R N ) and 1 A is the characteristic function of the set A . Here, * A denotes the reduced boundary of A in Ω and ν A is the measure-theoretic outer unit normal to A (Section 2.1).

We assume that Φ is an elliptic integrand on Ω (see Definition 2.6), i.e., Φ : Ω ¯ × R n [ 0 , ] is lower semicontinuous, Φ ( x , ) is convex and positively one-omogeneous, Φ ( x , t ν ) = t Φ ( x , ν ) , for every t 0 . Accordingly, we define the following anisotropic surface energy of a set A of finite perimeter in Ω :

(1.5) Φ ( A ; B ) B * A Φ ( x , ν A ( x ) ) d n 1 ( x ) ,

for every Borel set B Ω . The assumption

(1.6) 1 Λ Φ ( x , ν ) Λ ,

with Λ > 1 , allows us to compare the surface energy introduced in (1.5) with the usual perimeter. Anisotropic surface energies arise in many physical areas such as the formation of crystals [7,8], liquid drops [16,26], and capillary surfaces [18,19]. Almgren was the first to study the regularity of surfaces that minimize anisotropic variational problems in his celebrated article [3].

In the early stages, the studies in this area had been carried out in the setting of varifolds and currents. These results can be applied to surfaces of arbitrary codimension, but with rather strong regularity assumptions on the integrands of the anisotropic energies [9,39].

More recently, the regularity assumptions on the integrands Φ of the anisotropic energies have been weakened (see [20,25]), assuming that Φ ( x , ) is of class C 1 and Φ ( , ξ ) is Hölder continuous.

In the vectorial setting debated in this article, where the bulk energy is of general type with p -growth, the regularity that we can expect for the gradient of the minimal deformation u : Ω R N , ( N > 1 ), even in the absence of a surface term, is a partial regularity result, which is outside a negligible set. As we observed above, the regularity of the free interface E can be achieved by means of the regularity of u . On the other hand, knowing that the singular set S of the gradient u has Lebesgue measure zero does not give information on the singular set Σ of the free boundary that could also be totally contained in S .

We say that a pair ( u , E ) is a local minimizer of in Ω , if for every open set U Ω and every pair ( v , A ) , where v u W 0 1 , p ( U ; R N ) and A is a set of finite perimeter with A Δ E U , we have

U ( F ( u ) + 1 E G ( u ) ) d x + Φ ( E ; U ) U ( F ( v ) + 1 A G ( v ) ) d x + Φ ( A ; U ) .

Existence and regularity of local minimizers of integral functionals of the type

Ω F ( D u ) d x ,

with uniformly strict p -quasiconvex integrand F , and also in the non autonomous case, have been widely investigated (we refer to [1,2,1215,31,38] and for an exhaustive treatment to [30,32]).

In order to prove the existence of local minimizers for functionals involving both bulk and surface energies of general type (1.4), we invoke a result stated in [11]. The only difference in our setting is the presence of the anisotropic term Φ ( A ; U ) , and we give a semicontinuity result for the anisotropic energy (1.5), thus ensuring the existence shown in Section 3. Therefore, we deduce the following theorem.

Theorem 1.1

Let p > 1 and assume that ( F 1 ), ( F 2 ), ( G 1 ), and ( G 2 ) hold. Then, if v W loc 1 , p ( Ω ; R N ) and A Ω is a set of finite perimeter in Ω , for every sequence { ( v k , A k ) } k N such that { v k } weakly converges to v in W loc 1 , p ( Ω ; R N ) and 1 A k strongly converges to 1 A in L loc 1 ( Ω ) , we have

( v , A ) liminf k ( v k , A k ) .

In particular, admits a minimal configuration ( u , 1 E ) W loc 1 , p ( Ω ; R N ) × B V loc ( Ω ; [ 0 , 1 ] ) .

Then, we obtain C 1 , β partial regularity for minimizers u guaranteed by Theorem 1.1, in the case of general interfacial energies given in (1.5) just assuming the comparability hypothesis (1.6). Moreover, if a closeness condition on F and G is assumed, i.e., the condition ( H ) is in order, then we can prove a sharp regularity for u , i.e., u C 1 , γ ( Ω 1 ) for every γ ( 0 , 1 p ) for a full measure set Ω 1 Ω . It is worth pointing out that we do not need any regularity assumption on the integrand Φ to prove the regularity of u .

Theorem 1.2

Let ( u , E ) be a local minimizer of . Let the bulk density energies satisfy ( F 1 ), ( F 2 ), ( G 1 ), and ( G 2 ), and let the surface energy be of general type (1.5) with Φ satisfying (1.6). Then, there exist an exponent β ( 0 , 1 ) and an open set Ω 0 Ω with full measure such that u C 1 , β ( Ω 0 ; R N ) . In addition, if we assume

(H) L 2 1 + 2 < 1 ,

then there exists an open set Ω 1 Ω with full measure such that u C 1 , γ ( Ω 1 ; R N ) for every γ ( 0 , 1 p ) .

The proof of the previous result is based on a comparison argument with solutions of a suitable linearized system. We establish decay estimates for the “hybrid” excess functions U * and U * * (see (5.2) and (5.52)). We look at the points in which the excess is small and we use, as usual for this kind of analysis, a blow-up argument reducing the problem to the study of convergence of the minimal configurations ( u h , E h ) of rescaled functionals in the unit ball. We need two Caccioppoli-type inequalities for minimizers of perturbed rescaled functionals (see (5.18) and (5.62)) involving also the perimeter of the rescaled minimal set E h .

2 Notation and preliminary results

Let Ω be a bounded open set in R n , n 2 . We deal with vectorial functions u : Ω R N , N > 1 . The open ball centered at x R n of radius r > 0 is defined as

B r ( x ) { y R n : y x < r } .

We denote by S n 1 the unit sphere of R n and by c a generic constant that may vary in the same formula and between formulae. Relevant dependencies on parameters and special constants will be suitably emphasized using parentheses or subscripts. For B r ( x 0 ) R n and u L 1 ( B r ( x 0 ) ; R N ) , we denote

( u ) x 0 , r B r ( x 0 ) u ( x ) d x .

We omit the dependence on the center when it is clear from the context:

ξ , η trace ( ξ T η ) ,

for the usual inner product of ξ and η , and accordingly, ξ ξ , ξ 1 2 .

If F : R n × N R is sufficiently differentiable, we write

D F ( ξ ) η α = 1 N i = 1 n F ξ i α ( ξ ) η i α and D 2 F ( ξ ) η η α , β = 1 N i , j = 1 n F ξ i α ξ j β ( ξ ) η i α η j β ,

for ξ , η R n × N .

It is well known that for quasiconvex C 1 integrands, the assumptions ( F 1 ) and ( G 1 ) yield the upper bounds

(2.1) D ξ F ( ξ ) c 1 L 1 ( 1 + ξ 2 ) p 1 2 and D ξ G ( ξ ) c 2 L 2 ( 1 + ξ 2 ) p 1 2 ,

for all ξ R n × N , with c 1 and c 2 constants depending only on p (see [32, Lemma 5.2] or [38]).

Furthermore, if F and G are C 2 , then ( F 2 ) and ( G 2 ) imply the following strong Legendre-Hadamard conditions:

α , β = 1 N i , j = 1 n F ξ i α ξ j β ( Q ) λ i λ j μ α μ β c 3 λ 2 μ 2 and α , β = 1 N i , j = 1 n G ξ i α ξ j β ( Q ) λ i λ j μ α μ β c 4 λ 2 μ 2 ,

for all Q R n × N , λ R n , μ R N , where c 3 = c 3 ( p , 1 ) and c 4 = c 4 ( p , 2 ) are the positive constants (see [32, Proposition 5.2]). We will need the following regularity result [30,32].

Proposition 2.1

Let v W 1 , 2 ( Ω ; R N ) be such that

Ω Q α β i j D i v α D j φ β d x = 0 ,

for every φ C c ( Ω ; R N ) , where Q = { Q α β i j } is a constant matrix satisfying Q α β i j L and the strong Legendre-Hadamard condition

Q α β i j λ i λ j μ α μ β λ 2 μ 2 ,

for all λ R n , μ R N and for some positive constants , L > 0 . Then, v C and, for any B R ( x 0 ) Ω , the following estimate holds:

B R 2 ( x 0 ) D v ( D v ) x 0 , R 2 2 d x c R 2 B R ( x 0 ) D v ( D v ) x 0 , R 2 d x ,

where c = c ( n , N , , L ) > 0 .

The next iteration lemma has important applications in the regularity theory (for its proof, we refer to [32, Lemma 6.1]).

Lemma 2.2

Let 0 < ρ < R and let ψ : [ ρ , R ] R be a bounded nonnegative function. Assume that for all ρ s < t R , we have

ψ ( s ) ϑ ψ ( t ) + A + B ( s t ) α + C ( s t ) β ,

where ϑ [ 0 , 1 ) , α > β > 0 , and A , B , C 0 are the constants. Then, there exists a constant c = c ( ϑ , α ) > 0 such that

ψ ( ρ ) c A + B ( R ρ ) α + C ( R ρ ) β .

Given a C 1 function f : R n × N R , Q R n × N , and λ > 0 , we set

f Q , λ ( ξ ) f ( Q + λ ξ ) f ( Q ) D f ( Q ) λ ξ λ 2 , ξ R n × N .

We state the following lemma about the growth of f Q , λ and D f Q , λ , whose proof can be found in [2, Lemma II.3].

Lemma 2.3

Let p 2 , and let f be a C 2 ( R n × N ) function such that

f ( ξ ) C ( 1 + ξ p ) a n d D ξ f ( ξ ) C ( 1 + ξ p 1 ) ,

for any ξ R n × N . Then, for every M > 0 , there exists a constant c = c ( M ) > 0 such that, for every Q R n × N , Q M , and λ > 0 , it holds that

(2.2) f Q , λ ( ξ ) c ( ξ 2 + λ p 2 ξ p ) a n d D f Q , λ ( ξ ) c ( ξ + λ p 2 ξ p 1 ) ,

for all ξ R n × N .

2.1 Sets of finite perimeter

If E R n and t [ 0 , 1 ] , the set of points of E of density t is defined as

E ( t ) = { x R n : E B r ( x ) = t B r ( x ) + o ( r n ) , as r 0 + } .

Given a Lebesgue measurable set E R n and an open set U R n , we say that E is of locally finite perimeter in U if there exists a R n -valued Radon measure μ E (called the Gauss-Green measure of E ) on U such that

E ϕ d x = U ϕ d μ E , ϕ C c 1 ( U ) .

Moreover, we denote the perimeter of E relative to G U by P ( E , G ) = μ E ( G ) .

The support of μ E can be characterized by

(2.3) spt μ E = { x U : 0 < E B r ( x ) < ω n r n , r > 0 }

(see [37, Proposition 12.19]). It holds that spt μ E U E . The essential boundary of E is defined as e E R n \ ( E 0 E 1 ) . If E is of finite perimeter in an open set U , then the reduced boundary * E U of E is the set of those x U such that

(2.4) ν E ( x ) lim r 0 + μ E ( B r ( x ) ) μ E ( B r ( x ) )

exists and belongs to S n 1 . It is well known that

* E U e E spt μ E U E , U * E ¯ = spt μ E .

Federer’s criterion, see for instance [37, Theorem 16.2], ensures that

n 1 ( ( U e E ) \ * E ) = 0 .

Remark 2.4

(Minimal topological boundary) If E R n is a set of locally finite perimeter in U and F R n is such that ( E Δ F ) U = 0 , then F is a set of locally finite perimeter in U with μ E = μ F . In the rest of this article, the topological boundary E must be understood by considering the suitable representative of E in order to have that * E ¯ = E U . We will choose E ( 1 ) as representative of E . With such a choice, it can be easily verified that

U E = { x U : 0 < E B r ( x ) < ω n r n , r > 0 } .

Therefore, by (2.3),

* E ¯ = spt μ E = E U .

Finally, by De Giorgi’s rectifiability theorem (see [37, Theorem 15.9]), we obtain

(2.5) μ E = ν E n 1 * E a n d μ E = n 1 * E

on Borel sets compactly contained in U , where, given a Radon measure μ and a Borel set B , by μ B we refer to the measure given by μ B ( F ) = μ ( B F ) .

It is well known that if E and F are of locally finite perimeter in U , then E F , E F , and E \ F are the sets of locally finite perimeter in U . In this article, we use competitors obtained using set operations to test minimality inequalities. In fact, we just need properties involving the union of sets. A convenient way to handle these inequalities is to use the properties of Gauss-Green measures. The following result can be found in [37, Theorem 16.3].

Proposition 2.5

(Gauss-Green measure and set operations) If E and F are the sets of finite perimeter, it shows that ν E ( x ) = ± ν F ( x ) for n 1 -a.e. x * E * F . Setting

{ ν E = ν F } = { x * E * F ν E ( x ) = ν F ( x ) } ,

then

(2.6) μ E F = μ E F ( 0 ) + μ F E ( 0 ) + ν E n 1 { ν E = ν F } .

2.2 Anisotropic surface energy

Definition 2.6

(Elliptic integrands) Given an open set Ω in R n , Φ : Ω ¯ × R n [ 0 , ] is an elliptic integrand on Ω if it is lower semicontinuous, with Φ ( x , ) convex, and positively one-homogeneous for any x Ω ¯ , i.e., Φ ( x , t ν ) = t Φ ( x , ν ) for every t 0 . Accordingly, the anisotropic surface energy of a set E of finite perimeter in Ω is defined as

(2.7) Φ ( E ; B ) B * E Φ ( x , ν E ( x ) ) d n 1 ( x ) ,

for every Borel set B Ω .

Remark 2.7

(Comparability to perimeter) In order to prove the regularity of minimizers of anisotropic surface energies, it is well known that (see the seminal article [3]) a C k -dependence of the integrand Φ on the variable ν , and a continuity condition with respect to the variable x , must be assumed. In fact, one more condition is essential, i.e., a non-degeneracy-type condition for the integrand Φ . More precisely, we have to assume that there exists a constant K > 1 such that

(2.8) 1 K Φ ( x , ν ) K ,

for any x Ω and ν S n 1 . We do not need any further hypotheses on the elliptic integrands. We observe that if the elliptic integrand Φ satisfies Condition (2.8), then the anisotropic surface energy (2.7) satisfies the following comparability condition:

1 Λ n 1 ( B * E ) Φ ( E ; B ) Λ n 1 ( B * E ) ,

for any set E of finite perimeter in Ω and any Borel set B Ω .

Proposition 2.8

Let U R n be an open set and let E , F U be two sets of finite perimeter in U. It holds that

Φ ( E F ; U ) = Φ ( E ; F ( 0 ) ) + Φ ( F ; E ( 0 ) ) + Φ ( E ; { ν E = ν F } ) .

Proof

Let us observe that since * E F ( 0 ) E ( 1 2 ) F ( 0 ) and * F E ( 0 ) F ( 1 2 ) E ( 0 ) , we deduce that ( * E F ( 0 ) ) ( * F E ( 0 ) ) = . Similarly, we have ( * E F ( 0 ) ) { ν E = ν F } = and ( * F E ( 0 ) ) { ν E = ν F } = . Thus, by [37, Theorem 16.3], it holds that

Φ g ( E F ; U ) = U * ( E F ) Φ ( x , ν E F ( x ) ) d n 1 ( x ) = F ( 0 ) * E Φ ( x , ν E F ( x ) ) d n 1 ( x ) + E ( 0 ) * F Φ ( x , ν E F ( x ) ) d n 1 ( x ) + { ν E = ν F } Φ ( x , ν E F ( x ) ) d n 1 ( x ) .

Using (2.6), we deduce ν E F = ν E n 1 -a.e. on F ( 0 ) * E and ν E F = ν F n 1 -a.e. on E ( 0 ) * F ; thus, we reach the thesis.□

3 Lower semicontinuity

In this section, we prove Theorem 1.1. We base ourselves on the proof given in [11], focusing our attention on the only difference we have, the presence of the anisotropic surface energy. We obtain a semicontinuity result for the anisotropic perimeter, which is essential to handle this novelty.

Given a one-homogeneous Borel function Φ : Ω ¯ × R n [ 0 , ] as in Definition 2.6, for any R n -valued Radon measure μ on R n and any Borel set F R n , we define the Φ -anisotropic total variation of μ on F as

Φ g ( μ , F ) = F Φ x , μ μ ( x ) d μ ( x ) ,

where μ μ denotes the Radon-Nikodym derivative of μ with respect to its total variation. We also refer to the following theorem (see [5, Theorem 2.38]).

Theorem 3.1

(Reshetnyak’s lower semicontinuity theorem) Let Ω be an open subset of R n and μ and μ h be two finite R n -valued Radon measures on Ω . If μ h μ in Ω , then

Ω Φ x , μ μ ( x ) μ ( x ) liminf h Ω Φ x , μ h μ h ( x ) d μ h ( x ) ,

for every lower semicontinuous function Φ : Ω × R n [ 0 , ] , positively 1-homogeneous, and convex in the second variable.

We are now able to prove the following result.

Proposition 3.2

Let Φ be an elliptic integrand as in Definition 2.6, and let Φ be defined as in (2.7). Then,

Φ ( E ; U ) liminf h Φ ( E h ; U ) ,

whenever U R n is open, { E h } h N and E are the sets of locally finite perimeter such that 1 E h 1 E in L loc 1 ( U ) and μ E h μ E .

Proof

We recall that if E is a set of locally finite perimeter, then, by (2.5), μ E = n 1 * E and, by the definition of reduced boundary (2.4), μ E μ E = ν E on * E . Therefore, the Φ -anisotropic total variation of μ E is equal to the Φ -surface energy of E , i.e.,

Φ ( μ E ; F ) = Φ ( E ; F ) .

Thus, the claim follows by virtue of Reshetnyak’s lower semicontinuity theorem.□

Proof of Theorem 1.1

We omit the proof as it can be obtained following verbatim the arguments used in [11, Section 3]. The only difference with the proof given in [11] concerns the presence of the anisotropic perimeter. In this regard, we can use the lower semicontinuity result given earlier. We mention that the convergence 1 E h 1 E in L loc 1 ( U ) assumed in Theorem 1.1 and the condition limsup k P ( E h , K ) < for every K compact set in Ω ensure that μ E h μ E .□

4 Higher integrability result

This section is devoted to the proof of a higher integrability result for the gradient of the function u of the minimal configuration ( u , E ) .

Theorem 4.1

Assume that ( F 1 ), ( F 2 ), ( G 1 ), and ( G 2 ) hold, and let ( u , E ) be a local minimizer of . Then, there exists δ = δ ( n , p , 1 , L 1 , L 2 ) > 0 such that for every B 2 r ( x 0 ) Ω , it holds

B r ( x 0 ) D u p ( 1 + δ ) d x 1 1 + δ C B 2 r ( x 0 ) D u p d x + 1 ,

where C = C ( n , p , 1 , L 1 , L 2 ) is a positive constant.

Proof

We consider 0 < r < s < t < 2 r and let η C 0 ( B t ) be a cut-off function between B s and B t , i.e., 0 η 1 , η 1 in B s , and η c t s .

Setting

ψ 1 η ( u ( u ) x 0 , 2 r ) and ψ 2 ( 1 η ) ( u ( u ) x 0 , 2 r ) ,

by the uniformly strict quasiconvexity of F in ( F 2 ), we have

(4.1) 1 B t D ψ 1 ( x ) p d x B t F ( D ψ 1 ) d x = B t F ( D u D ψ 2 ) d x .

We write

(4.2) B t F ( D u D ψ 2 ) d x = B t F ( D u ) d x + B t F ( D u D ψ 2 ) d x B t F ( D u ) d x = B t F ( D u ) d x B t 0 1 D F ( D u θ D ψ 2 ) D ψ 2 d θ d x B t [ F ( D u ) + 1 E G ( D u ) ] d x B t 0 1 D F ( D u θ D ψ 2 ) D ψ 2 d θ d x B t [ F ( D u D ψ 1 ) + 1 E G ( D u D ψ 1 ) ] d x B t 0 1 D F ( D u θ D ψ 2 ) D ψ 2 d θ d x ,

where we used the fact that G ( ξ ) 0 and the minimality of ( u , E ) with respect to ( u ψ 1 , E ) . Combining (4.2) in (4.1) and using the upper bound on D F given by (2.1), we obtain

1 B s D u p d x = 1 B s D ψ 1 p d x B t F ( D ψ 2 ) d x + B t 1 E G ( D ψ 2 ) d x + c ( p , L 1 ) B t \ B s ( 1 + D u 2 + D ψ 2 2 ) p 1 2 D ψ 2 d x c ( p , L 1 , L 2 ) B t \ B s D ψ 2 p d y + B t \ B s D u p d x + B t c ( p , L 1 , L 2 ) B t \ B s D u p d x + c B t \ B s ( u ( u ) x 0 , 2 r ) p ( t s ) p d x + B t ,

where we used assumptions ( F 1 ) and ( G 1 ), Young’s inequality, and the properties of η . Adding c ( p , L 1 , L 2 ) B s D u p d x to both sides of the previous estimate, we obtain

( 1 + c ( p , L 1 , L 2 ) ) B s D u p d x c ( p , L 1 , L 2 ) B t D u p d x + B t \ B s u ( u ) x 0 , 2 r p ( t s ) p d x + B t c ( p , L 1 , L 2 ) B t D u p d x + c B 2 r 1 + u ( u ) x 0 , 2 r p ( t s ) p d x ,

and, by Lemma 2.2, we deduce that

B r D u p d x c ( p , 1 , L 1 , L 2 ) B 2 r 1 + u u 2 r p r p d x .

The Sobolev-Poincaré inequality (see [32, p. 102]) implies that

B r D u p d x c ( n , p , 1 , L 1 , L 2 ) B 2 r D u n p n + p d x n + p n + 1 ,

and the conclusion follows by virtue of Giaquinta-Modica theorem (see [32, p. 203]).□

5 Decay estimates

In this section, we collect some energy estimates for minimizers of the functional (1.4) that will be crucial in the proof of Theorem 1.2. In order to obtain them, we will employ a well-known blow-up technique involving a quantity called excess, which includes all the energy terms of the functional. We have to use different types of excess depending on whether the assumption ( H ) is in force or not. We consider the bulk excess function defined as

(5.1) U ( x 0 , r ) B r ( x 0 ) [ D u ( x ) ( D u ) x 0 , r 2 + D u ( x ) ( D u ) x 0 , r p ] d x ,

for B r ( x 0 ) Ω . In the case that the assumption ( H ) is in force, we will use the following hybrid" excess

(5.2) U * ( x 0 , r ) U ( x 0 , r ) + P ( E , B r ( x 0 ) ) r n 1 + r .

Proposition 5.1

Let ( u , E ) be a local minimizer of the functional introduced in (1.4), and the assumptions ( F 1 ), ( F 2 ), ( G 1 ), ( G 2 ), and ( H ) hold. For every M > 0 and every 0 < τ < 1 4 , there exist ε 0 = ε 0 ( τ , M ) > 0 and c * = c * ( n , p , 1 , 2 , L 1 , L 2 , Λ , M ) > 0 such that, whenever B r ( x 0 ) Ω verifies

( D u ) x 0 , r M and U * ( x 0 , r ) ε 0 ,

then

(5.3) U * ( x 0 , τ r ) c * τ U * ( x 0 , r ) .

Proof

In order to prove (5.3), we argue by contradiction. Let M > 0 and τ ( 0 , 1 4 ) be such that for every h N , C * > 0 , there exists a ball B r h ( x h ) Ω such that

(5.4) ( D u ) x h , r h M , U * ( x h , r h ) 0

and

(5.5) U * ( x h , τ r h ) C * τ U * ( x h , r h ) .

The constant C * will be determined later. Remark that we can confine ourselves to the case in which E B r h ( x h ) , since the case in which B r h ( x h ) Ω \ E is easier because U = U * .

Step 1. Blow-up.

Set λ h 2 U * ( x h , r h ) , A h ( D u ) x h , r h , a h ( u ) x h , r h , and define

(5.6) v h ( y ) u ( x h + r h y ) a h r h A h y λ h r h , y B 1 .

One can easily check that ( D v h ) 0 , 1 = 0 and ( v h ) 0 , 1 = 0 .

Set

E h E x h r h , E h * E x h r h B 1 .

Note that

(5.7) λ h 2 = U * ( x h , r h ) = B 1 [ D u ( x h + r h y ) A h 2 + D u ( x h + r h y ) A h p ] d y + P ( E , B r h ( x h ) ) r h n 1 + r h = B 1 [ λ h D v h 2 + λ h D v h p ] d y + P ( E h , B 1 ) + r h .

It follows that r h 0 , P ( E h , B 1 ) 0 , and

(5.8) r h λ h 2 1 , B 1 [ D v h 2 + λ h p 2 D v h p ] d y 1 , P ( E h , B 1 ) λ h 2 1 .

Therefore, by (5.4) and (5.8), there exist a (not relabeled) subsequence of { v h } h N , A R n × N and v W 1 , 2 ( B 1 ; R N ) , such that

(5.9) v h v weakly in W 1 , 2 ( B 1 ; R N ) , v h v , strongly in L 2 ( B 1 ; R N ) , A h A , λ h D v h 0 in L p ( B 1 ; R n × N ) and pointwise a.e. in B 1 ,

where we used the fact that ( v h ) 0 , 1 = 0 . Moreover, by (5.8) and (5.4), we also deduce that

(5.10) lim h ( P ( E h , B 1 ) ) n n 1 λ h 2 = lim h ( P ( E h , B 1 ) ) 1 n 1 limsup h P ( E h , B 1 ) λ h 2 = 0 .

Therefore, by the relative isoperimetric inequality,

(5.11) lim h min E h * λ h 2 , B 1 \ E h λ h 2 c ( n ) lim h ( P ( E h , B 1 ) ) n n 1 λ h 2 = 0 .

In the sequel, the proof will proceed differently depending on whether

min { E h * , B 1 \ E h } = E h * or min { E h * , B 1 \ E h } = B 1 \ E h .

The first case is easier to handle. To understand the reason, let us introduce the expansion of F and G around A h as follows:

(5.12) F h ( ξ ) F ( A h + λ h ξ ) F ( A h ) D F ( A h ) λ h ξ λ h 2 , G h ( ξ ) G ( A h + λ h ξ ) G ( A h ) D G ( A h ) λ h ξ λ h 2 ,

for any ξ R n × N . In the first case, the suitable rescaled functional to consider in the blow-up procedure is as follows:

(5.13) h ( w ) B 1 [ F h ( D w ) d y + 1 E h * G h ( D w ) ] d y .

We claim that v h satisfies the minimality inequality

(5.14) h ( v h ) h ( v h + ψ ) + 1 λ h B 1 1 E h * D G ( A h ) D ψ ( y ) d y ,

for any ψ W 0 1 , p ( B 1 ; R N ) . Indeed, using the change of variable x = x h + r h y , the minimality of ( u , E ) with respect to ( u + φ , E ) , for φ W 0 1 , p ( B r h ( x h ) ; R N ) , and setting ψ h ( y ) φ ( x h + r h y ) r h , it holds that

B 1 [ F h ( D v h ( y ) ) + 1 E h * G h ( D v h ( y ) ) ] d y B 1 [ F h ( D v h ( y ) + D ψ h ( y ) ) + 1 E h * G h ( D v h ( y ) + D ψ h ( y ) ) ] d y + 1 λ h B 1 1 E h * D G ( A h ) D ψ h ( y ) d y ,

and (5.14) follows by the definition of h in (5.13).

In the second case, the suitable rescaled functional to consider in the blow-up procedure is

h ( w ) B 1 [ F h ( D w ) + G h ( D w ) ] d y .

Then, we claim that

(5.15) h ( v h ) h ( v h + ψ ) + L 2 λ h 2 ( B 1 \ E h ) supp ψ ( μ 2 + A h + λ h D v h 2 ) p 2 d y ,

for all ψ W 0 1 , p ( B 1 ; R N ) . Indeed, the minimality of ( u , E ) with respect to ( u + φ , E ) , for φ W 0 1 , p ( B r h ( x h ) ; R N ) , implies that

(5.16) B r h ( x h ) ( F + G ) ( D u ) d x = B r h ( x h ) [ F ( D u ) + 1 E G ( D u ) ] d x + B r h ( x h ) \ E G ( D u ) d x B r h ( x h ) [ F ( D u + D φ ) + 1 E G ( D u + D φ ) ] d x + B r h ( x h ) \ E G ( D u ) d x = B r h ( x h ) ( F + G ) ( D u + D φ ) d x + B r h ( x h ) \ E [ G ( D u ) G ( D u + D φ ) ] d x B r h ( x h ) ( F + G ) ( D u + D φ ) d x + ( B r h ( x h ) \ E ) supp φ G ( D u ) d x ,

where we used that the last integral vanishes outside the support of φ and that G 0 . Using the change of variable x = x h + r h y in the previous formula, we obtain

B 1 ( F + G ) ( D u ( x h + r h y ) ) d y B 1 ( F + G ) ( D u ( x h + r h y ) + D φ ( x h + r h y ) ) d y + ( B 1 \ E h ) supp ψ G ( D u ( x h + r h y ) ) d y ,

or, equivalently, using the definitions of v h ,

B 1 ( F + G ) ( A h + λ h D v h ) d y B 1 ( F + G ) ( A h + λ h ( D v h + D ψ ) ) d y + ( B 1 \ E h ) supp ψ G ( A h + λ h D v h ) d y ,

where ψ ( y ) φ ( x h + r h y ) λ h r h , for y B 1 . Therefore, setting

H h F h + G h ,

by the definition of F h and G h in (5.12), and using the assumption ( G 1 ), we have that

(5.17) B 1 H h ( D v h ) d y B 1 H h ( D v h + D ψ ) d y + 1 λ h 2 ( B 1 \ E h ) supp ψ G ( A h + λ h D v h ) d y B 1 H h ( D v h + D ψ ) d y + L 2 λ h 2 ( B 1 \ E h ) supp ψ ( 1 + A h + λ h D v h 2 ) p 2 d y ,

i.e., (5.15).

Step 2. A Caccioppoli-type inequality.

We claim that there exists a constant c = c ( n , p , 1 , 2 , L 2 , M ) > 0 such that for every 0 < ρ < 1 , there exists h 0 = h 0 ( n , ρ ) N such that for all h > h 0 , we have

(5.18) B ρ 2 ( 1 + λ h p 2 D v h ( D v h ) ρ 2 p 2 ) D v h ( D v h ) ρ 2 2 d y c B ρ v h ( v h ) ρ ( D v h ) ρ 2 y 2 ρ 2 + λ h p 2 v h ( v h ) ρ ( D v h ) ρ 2 y p ρ p d y + P ( E h , B 1 ) n n 1 λ h 2 .

We divide the proof into two substeps.

Substep 2.a The case min { E h * , B 1 \ E h } = E h * .

Consider 0 < ρ 2 < s < t < ρ < 1 , and let η C 0 ( B t ) be a cut-off function between B s and B t , i.e., 0 η 1 , η 1 on B s and η c t s . Set b h ( v h ) B ρ , B h ( D v h ) B ρ 2 , and set

(5.19) w h ( y ) v h ( y ) b h B h y ,

for any y B 1 . Proceeding similarly as in (5.7), let us rescale F and G around A h + λ h B h ,

(5.20) F ˜ h ( ξ ) F ( A h + λ h B h + λ h ξ ) F ( A h + λ h B h ) D F ( A h + λ h B h ) λ h ξ λ h 2 , G ˜ h ( ξ ) G ( A h + λ h B h + λ h ξ ) G ( A h + λ h B h ) D G ( A h + λ h B h ) λ h ξ λ h 2 ,

for any ξ R n × N . It is easy to check that Lemma 2.3 applies to each F ˜ h and G ˜ h , for some constants that depend on M (see (5.4)) and could also depend on ρ through λ h B h . However, given ρ , we may choose h 0 = h 0 ( n , ρ ) large enough to have λ h B h < λ h ω n ρ n 2 < 1 , for any h h 0 . Indeed, by (5.8), we have

B h = B ρ 2 D v h d y B ρ 2 D v h 2 d y 1 2 1 B ρ 2 1 2 c ( n ) ρ n 2 ,

and so the constant in (2.2) can be taken independently of ρ .

Set

ψ 1 , h η w h and ψ 2 , h ( 1 η ) w h .

By the uniformly strict quasiconvexity of F ˜ h , we have

(5.21) 1 B s ( 1 + λ h D w h 2 ) p 2 2 D w h 2 d y 1 B t ( 1 + λ h D ψ 1 , h 2 ) p 2 2 D ψ 1 , h 2 d y B t F ˜ h ( D ψ 1 , h ) d y = B t F ˜ h ( D w h D ψ 2 , h ) d y = B t F ˜ h ( D w h ) d y + B t F ˜ h ( D w h D ψ 2 , h ) d y B t F ˜ h ( D w h ) d y = B t F ˜ h ( D w h ) d y B t 0 1 D F ˜ h ( D w h θ D ψ 2 , h ) D ψ 2 , h d θ d y .

We estimate separately the two addends in the right-hand side of the previous chain of inequalities. We deal with the first addend by means of a rescaling of the minimality condition of ( u , E ) . Using the change of variable x = x h + r h y , the fact that G 0 , and the minimality of ( u , E ) with respect to ( u + φ , E ) for φ W 0 1 , p ( B r h ( x h ) ; R N ) , we have

B 1 F ( D u ( x h + r h y ) ) d y B 1 [ F ( D u ( x h + r h y ) ) + 1 E h * G ( D u ( x h + r h y ) ) ] d y B 1 [ F ( D u ( x h + r h y ) + D φ ( x h + r h y ) ) + 1 E h * G ( D u ( x h + r h y ) + D φ ( x h + r h y ) ) ] d y ,

i.e., by the definitions of v h (see (5.6)) and w h (see (5.19)),

B 1 F ( A h + λ h B h + λ h D w h ) d y B 1 [ F ( A h + λ h B h + λ h ( D w h + D ψ ) ) + 1 E h * G ( A h + λ h B h + λ h ( D w h + D ψ ) ) ] d y ,

for ψ φ ( x h + r h y ) λ h r h W 0 1 , p ( B 1 ; R N ) . Therefore, recalling the definitions of F ˜ h and G ˜ h in (5.20), we have that

B 1 F ˜ h ( D w h ) d y B 1 [ F ˜ h ( D w h + D ψ ) + 1 E h * G ˜ h ( D w h + D ψ ) ] d y + 1 λ h 2 B 1 1 E h * [ G ( A h + λ h B h ) + D G ( A h + λ h B h ) λ h ( D w h + D ψ ) ] d y .

Choosing ψ = ψ 1 , h as a test function in the previous inequality, we obtain

(5.22) B t F ˜ h ( D w h ) d y B t [ F ˜ h ( D w h D ψ 1 , h ) + 1 E h * G ˜ h ( D w h D ψ 1 , h ) ] d y + 1 λ h 2 B 1 1 E h * [ G ( A h + λ h B h ) + D G ( A h + λ h B h ) λ h ( D w h D ψ 1 , h ) ] d y = B t \ B s [ F ˜ h ( D ψ 2 , h ) + 1 E h * G ˜ h ( D ψ 2 , h ) ] d y + 1 λ h 2 B 1 1 E h * [ G ( A h + λ h B h ) + D G ( A h + λ h B h ) λ h D ψ 2 , h ] d y c ( M ) B t \ B s [ D ψ 2 , h 2 + λ h p 2 D ψ 2 , h p ] d y + c ( n , p , L 2 , M ) E h * λ h 2 + 1 λ h E h * D ψ 2 , h d y ,

where we used Lemma 2.3 and the second estimate in (2.1), and the fact that A h + λ h B h M + 1 . Now, we estimate the second addend in the right-hand side of (5.21). Using the upper bound on D F ˜ h in Lemma 2.3 and Hölder’s inequality, we obtain

(5.23) B t 0 1 D F ˜ h ( D w h θ D ψ 2 , h ) D ψ 2 , h d θ d y c ( M ) B t \ B s 0 1 ( D w h θ D ψ 2 , h + λ h p 2 D w h θ D ψ 2 , h p 1 ) D ψ 2 , h d θ d y c ( p , M ) B t \ B s ( D w h 2 + λ h p 2 D w h p + D ψ 2 , h 2 + λ h p 2 D ψ 2 , h p ) d y .

Hence, combining (5.21) with (5.22) and (5.23), using the properties of η , we obtain

1 B s ( 1 + λ h D w h 2 ) p 2 2 D w h 2 d y c ( p , M ) B t \ B s ( D w h 2 + λ h p 2 D w h p + D ψ 2 , h 2 + λ h p 2 D ψ 2 , h p ) d y + c ( n , p , L 2 , M ) 1 λ h E h * D ψ 2 , h d y + E h * λ h 2 c ( p , M ) B t \ B s ( D w h 2 + λ h p 2 D w h p ) d y + c ( p , M ) B t \ B s w h 2 ( t s ) 2 + λ h p 2 w h p ( t s ) p d y + c ( n , p , L 2 , M ) 1 λ h E h * D ψ 2 , h ( y ) 2 d y 1 2 E h * 1 2 + E h * λ h 2 c ( p , M ) B t \ B s ( 1 + λ h 2 D w h 2 ) p 2 2 D w h 2 d y + c ( p , M ) B t \ B s w h 2 ( t s ) 2 + λ h p 2 w h p ( t s ) p d y + c ( n , p , L 2 , M ) 1 λ h E h * D ψ 2 , h ( y ) 2 d y 1 2 E h * 1 2 + E h * λ h 2 c ( n , p , L 2 , M ) B t \ B s ( 1 + λ h 2 D w h 2 ) p 2 2 D w h 2 d y + B ρ w h 2 ( t s ) 2 + λ h p 2 w h p ( t s ) p d y + E h * λ h 2 ,

where we used Young’s inequality. By adding c ( n , p , L 2 , M ) B s ( 1 + λ h D w h 2 ) p 2 2 D w h 2 d y to both sides of the previous estimate, dividing by 1 + c ( n , p , L 2 , M ) , and thanks to the iteration Lemma 2.2, we deduce that

B ρ 2 ( 1 + λ h D w h 2 ) p 2 2 D w h 2 d y c ( n , p , 1 , L 2 , M ) B ρ w h 2 ρ 2 + λ h p 2 w h p ρ p d y + E h * λ h 2 .

Therefore, by the definition of w h , we conclude that

(5.24) B ρ 2 ( 1 + λ h p 2 D v h ( D v h ) ρ 2 p 2 ) D v h ( D v h ) ρ 2 2 d y c ( n , p , 1 , L 2 , M ) B ρ v h ( v h ) ρ ( D v h ) ρ 2 y 2 ρ 2 + λ h p 2 v h ( v h ) ρ ( D v h ) ρ 2 y p ρ p d y + E h * λ h 2 ,

which, by the relative isoperimetric inequality and the hypothesis of this substep, i.e., min { E h * , B 1 \ E h } = E h * , yields the estimate (5.18).

Substep 2.b The case min { E h * , B 1 \ E h } = B 1 \ E h .

As in the previous substep, we fix 0 < ρ 2 < s < t < ρ < 1 and let η C 0 ( B t ) be a cutoff function between B s and B t , i.e., 0 η 1 , η 1 on B s and η c t s . Also, we set b h ( v h ) B ρ , B h ( D v h ) B ρ 2 , and define

w h ( y ) v h ( y ) b h B h y ,

for any y B 1 , and

H ˜ h F ˜ h + G ˜ h .

We remark that Lemma 2.3 applies to H ˜ h , i.e.,

H ˜ h ( ξ ) c ( M ) ( ξ 2 + λ h p 2 ξ p ) , ξ R n × N ,

and, by the uniformly strict quasiconvexity conditions ( F 2 ) and ( G 2 ),

(5.25) B 1 H ˜ h ( ξ + D ψ ) d x B t [ H ˜ h ( ξ ) + ˜ ( 1 + λ h D ψ 1 , h 2 ) p 2 2 D ψ 1 , h 2 ] d y ,

for all ψ W 0 1 , p ( B 1 ; R N ) , where ˜ is such that

˜ = 1 + 2 .

Set

ψ 1 , h η w h and ψ 2 , h ( 1 η ) w h .

By (5.25) and since H ˜ h ( 0 ) = 0 , we have

(5.26) ˜ B s ( 1 + λ h D w h 2 ) p 2 2 D w h 2 d y ˜ B t ( 1 + λ h D ψ 1 , h 2 ) p 2 2 D ψ 1 , h 2 d y B t H ˜ h ( D ψ 1 , h ) d y = B t H ˜ h ( D w h D ψ 2 , h ) d y = B t H ˜ h ( D w h ) d y + B t H ˜ h ( D w h D ψ 2 , h ) d y B t H ˜ h ( D w h ) d y = B t H ˜ h ( D w h ) d y B t 0 1 D H ˜ h ( D w h θ D ψ 2 , h ) D ψ 2 , h d θ d y .

As in the previous step, we estimate separately the two addends in the right-hand side of the previous chain of inequalities. We deal with the first addend by means of a rescaling of the minimality condition of ( u , E ) . By virtue of the minimality inequality in (5.17) and since D v h = D w h + B h , we obtain

B 1 H h ( D w h + B h ) d y B 1 H h ( D w h + B h + D ψ ) d y + L 2 λ h 2 ( B 1 \ E h ) supp ψ ( 1 + A h + λ h B h + λ h D w h 2 ) p 2 d y ,

or, equivalently, by the definition of H ˜ h ,

(5.27) B 1 H ˜ h ( D w h ) d y B 1 H ˜ h ( D w h + D ψ ) d y + L 2 λ h 2 ( B 1 \ E h ) supp ψ ( 1 + A h + λ h B h + λ h D w h 2 ) p 2 d y .

Choosing ψ = ψ 1 , h as a test function in (5.27) and using the fact that H ˜ h ( 0 ) = 0 , we obtain

B t H ˜ h ( D w h ) d y B t H ˜ h ( D w h ( y ) D ψ 1 , h ) d y + L 2 λ h 2 B t \ E h ( 1 + A h + λ h B h + λ h D w h 2 ) p 2 d y = B t \ B s H ˜ h ( D ψ 2 , h ) d y + L 2 λ h 2 B t \ E h ( 1 + A h + λ h B h + λ h D w h 2 ) p 2 d y c ( M ) B t \ B s ( D ψ 2 , h 2 + λ h p 2 D ψ 2 , h p ) d y + L 2 λ h 2 B t \ E h ( 1 + A h + λ h B h + λ h D w h 2 ) p 2 d y .

By remarking that

( 1 + A h + λ h B h + λ h D w h 2 ) p 2 ( 1 + A h + λ h B h 2 + 2 A h + λ h B h λ h D w h + λ h 2 D w h 2 ) p 2 1 + 1 ε c ( M ) + ( 1 + ε ) λ h 2 D w h 2 p 2 1 + 1 ε p 2 + 1 c ( M ) p 2 + ( 1 + ε ) p 2 + 1 λ h p D w h p ,

for every ε > 0 , we obtain

(5.28) B t H ˜ h ( D w h ) d y c ( M ) B t \ B s ( D ψ 2 , h 2 + λ h p 2 D ψ 2 , h p ) d y + ( 1 + ε ) p 2 + 1 L 2 λ h p 2 B t D w h p d y + c ( p , L 2 , M , ε ) B 1 \ E h λ h 2 .

Now, we estimate the second addend in the right-hand side of (5.26). Using the upper bound on D H ˜ h in Lemma 2.3, we obtain

(5.29) B t 0 1 D H ˜ h ( D w h θ D ψ 2 , h ) D ψ 2 , h d θ d y c ( M ) B t \ B s 0 1 ( D w h θ D ψ 2 , h + λ h p 2 D w h θ D ψ 2 , h p 1 ) D ψ 2 , h d θ d y c ( p , M ) B t \ B s ( D w h 2 + λ h p 2 D w h p + D ψ 2 , h 2 + λ h p 2 D ψ 2 , h p ) d y .

Inserting (5.28) and (5.29) into (5.26), we infer that

˜ B s ( 1 + λ h D w h 2 ) p 2 2 D w h 2 d y c ( p , M ) B t \ B s ( D w h 2 + λ h p 2 D w h p + D ψ 2 , h 2 + λ h p 2 D ψ 2 , h p ) d y + ( 1 + ε ) p 2 + 1 L 2 λ h p 2 B t D w h p d y + c ( p , L 2 , M , ε ) B 1 \ E h λ h 2 c ( p , M ) B t \ B s ( 1 + λ h D w h 2 ) p 2 2 D w h 2 d y + c ( p , M ) B t \ B s w h 2 ( t s ) 2 + λ h p 2 w h p ( t s ) p d y + ( 1 + ε ) p 2 + 1 L 2 B t ( 1 + λ h D w h 2 ) p 2 2 D w h 2 d y + c ( p , L 2 , M , ε ) B 1 \ E h λ h 2 .

Using the hole-filling technique as in the previous case, we obtain

B s ( 1 + λ h D w h 2 ) p 2 2 D w h 2 d y ( c ( p , M ) + ( 1 + ε ) p 2 + 1 L 2 ) ( c ( p , M ) + ˜ ) B t ( 1 + λ h D w h 2 ) p 2 2 D w h 2 d y + B t \ B s w h 2 ( t s ) 2 + λ h p 2 w h p ( t s ) p d y + c ( p , 1 , 2 , L 2 , M , ε ) B 1 \ E h λ h 2 .

The assumption ( H ) implies that there exists ε = ε ( p , 1 , 2 , L 2 ) > 0 such that ( 1 + ε ) p 2 + 1 L 2 1 + 2 < 1 . Therefore, we have

c + ( 1 + ε ) p 2 + 1 L 2 c + ˜ = c + ( 1 + ε ) p 2 + 1 L 2 c + 1 + 2 < 1 .

So, by virtue of Lemma 2.2, from the previous estimate, we deduce that

B ρ 2 ( 1 + λ h D w h 2 ) p 2 2 D w h 2 d y c ( p , 1 , 2 , L 2 , M ) B ρ w h 2 ρ 2 + λ h p 2 w h p ρ p d y + B 1 \ E h λ h 2 .

Therefore, by the definition of w h , we conclude that

B ρ 2 ( 1 + λ h p 2 D v h ( D v h ) ρ 2 p 2 ) D v h ( D v h ) ρ 2 2 d y c ( p , 1 , 2 , L 2 , M ) B ρ v h ( v h ) ρ ( D v h ) ρ 2 y 2 ρ 2 + λ h p 2 v h ( v h ) ρ ( D v h ) ρ 2 y p ρ p d y + B 1 \ E h λ h 2 ,

which, by the relative isoperimetric inequality and since we have B 1 \ E h = min { E h * , B 1 \ E h } , gives the estimate (5.18).

Step 3. We prove that there exists a constant c ˜ = c ( n , N , 1 , 2 , L 1 , L 2 ) > 0 such that

(5.30) B τ 2 D v ( D v ) τ 2 2 c ˜ τ 2 B τ D v ( D v ) τ 2 d x ,

for any τ < 1 .

It will follow that

(5.31) B τ 2 D v ( D v ) τ 2 2 c ˜ τ 2 B τ D v ( D v ) τ 2 c ˜ τ 2 ,

since

D v L 2 ( B 1 ) limsup h D v h L 2 ( B 1 ) c ( n ) .

As mentioned earlier, we will divide the proof in two substeps.

Substep 3.a The case min { E h * , B 1 \ E h } = E h * .

We claim that v solves the linear system

B 1 D 2 F ( A ) D v D ψ d y = 0 ,

for all ψ C 0 1 ( B 1 ; R N ) . Since v h satisfies (5.14), we have that

0 h ( v h + s ψ ) h ( v h ) + 1 λ h B 1 1 E h * D G ( A h ) s D ψ d y ,

for every ψ C 0 1 ( B 1 ; R N ) and s ( 0 , 1 ) . By the definition of h , we obtain

0 h ( v h + s ψ ) h ( v h ) + 1 λ h B 1 1 E h * D G ( A h ) s D ψ d y = 1 λ h B 1 0 1 [ D F ( A h + λ h ( D v h + t s D ψ ) ) ] s D ψ d t D F ( A h ) s D ψ d y + 1 λ h B 1 1 E h * 0 1 D G ( A h + λ h ( D v h + t s D ψ ) ) s D ψ d t D G ( A h ) s D ψ d y + 1 λ h B 1 1 E h * D G ( A h ) s D ψ ( y ) d y = 1 λ h B 1 0 1 [ D F ( A h + λ h ( D v h + t s D ψ ) ) ] s D ψ d t D F ( A h ) s D ψ d y + 1 λ h B 1 0 1 1 E h * D G ( A h + λ h ( D v h + t s D ψ ) ) s D ψ d t d y .

We divide by s and pass to the limit as s 0 ; therefore, we deduce that

(5.32) 0 1 λ h B 1 ( D F ( A h + λ h D v h ) D F ( A h ) ) D ψ d y + 1 λ h B 1 1 E h * D G ( A h + λ h D v h ) D ψ d y .

We partition the unit ball as

B 1 = B h + B h = { y B 1 : λ h D v h > 1 } { y B 1 : λ h D v h 1 } .

By (5.8), we obtain

(5.33) B h + B h + λ h 2 D v h 2 d y λ h 2 B 1 D v h 2 d y c ( n ) λ h 2 .

We rewrite (5.32) as follows:

(5.34) 0 1 λ h B 1 ( D F ( A h + λ h D v h ) D F ( A h ) ) D ψ d y + 1 λ h B 1 1 E h * D G ( A h + λ h D v h ) D ψ d y = 1 λ h B h + ( D F ( A h + λ h D v h ) D F ( A h ) ) D ψ d y + 1 λ h B h ( D F ( A h + λ h D v h ) D F ( A h ) ) D ψ d y + 1 λ h B 1 1 E h * D G ( A h + λ h D v h ) D ψ d y = 1 λ h B h + ( D F ( A h + λ h D v h ) D F ( A h ) ) D ψ d y + B h 0 1 ( D 2 F ( A h + t λ h D v h ) D 2 F ( A ) ) d t D v h D ψ d y + B h D 2 F ( A ) D v h D ψ d y + 1 λ h B 1 1 E h * D G ( A h + λ h D v h ) D ψ d y .

By virtue of the first estimate in (2.1) and Hölder’s inequality, we obtain

1 λ h B h + ( D F ( A h + λ h D v h ) D F ( A h ) ) D ψ d y c ( p , L 1 , M , D ψ ) B h + λ h + λ h p 2 B h + D v h p 1 d y c ( n , p , L 1 , M , D ψ ) λ h + λ h B 1 λ h p 2 D v h p d y p 1 p B h + λ h 2 1 p c ( n , p , L 1 , M , D ψ ) λ h ,

thanks to (5.4) (to bound A h M ), (5.8), and (5.33). Thus,

(5.35) lim h 1 λ h B h + ( D F ( A h + λ h D v h ) D F ( A h ) ) D ψ d y = 0 .

By (5.4) and the definition of B h we have that A h + λ h D v h M + 1 on B h . Hence, we estimate

(5.36) B h 0 1 ( D 2 F ( A h + t λ h D v h ) D 2 F ( A ) ) d t D v h D ψ d y B h 0 1 ( D 2 F ( A h + t λ h D v h ) D 2 F ( A ) ) d t D v h D ψ d y B h 0 1 ( D 2 F ( A h + t λ h D v h ) D 2 F ( A ) ) d t 2 d y 1 2 D v h L 2 ( B 1 ) D ψ L ( B 1 ) c ( n , D ψ ) B h 0 1 ( D 2 F ( A h + t λ h D v h ) D 2 F ( A ) ) d t 2 d y 1 2 ,

where we used (5.8). Since, by (5.9), λ h D v h 0 a.e. in B 1 , the uniform continuity of D 2 F on bounded sets implies that

(5.37) lim h B h 0 1 ( D 2 F ( A h + t λ h D v h ) D 2 F ( A ) ) d t D v h D ψ d y = 0 .

Note that (5.33) yields that 1 B h 1 B 1 in L r ( B 1 ) , for every r < . Therefore, by (5.8),

B h D 2 F ( A ) D v h D ψ d y B 1 D 2 F ( A ) D v D ψ d y B h D 2 F ( A ) D v h D ψ d y B 1 D 2 F ( A ) D v h D ψ d y + B 1 D 2 F ( A ) D v h D ψ d y B 1 D 2 F ( A ) D v D ψ d y c ( D ψ ) 1 B h 1 B 1 L 2 ( B 1 ) D v h L 2 ( B 1 ) 0 1 D 2 F ( A ) d t + B 1 D 2 F ( A ) ( D v h D v ) D ψ d y .

Thus, by the weak convergence of D v h to D v in L 2 ( B 1 ) , it follows that

(5.38) lim h B h D 2 F ( A ) D v h D ψ d y = B 1 D 2 F ( A ) D v D ψ d y .

By the second estimate in (2.1), we deduce that

1 λ h B 1 1 E h * D ξ G ( A h + λ h D v h ) D ψ d y c ( p , L 2 ) λ h B 1 1 E h * ( 1 + A h + λ h D v h 2 ) p 1 2 D ψ d y c ( p , L 2 , M , D ψ ) 1 λ h E h * + λ h p 2 E h * D v h p 1 d y c ( p , L 2 , M , D ψ ) 1 λ h E h * + λ h B 1 λ h p 2 D v h p d y p 1 p E h * λ h 2 1 p c ( n , p , L 2 , M , D ψ ) 1 λ h E h * + c λ h E h * λ h 2 1 p ,

where we used (5.4) in the second inequality to bound A h by M and (5.8). Since min { E h * , B 1 \ E h } = E h * , by (5.11), we have

lim h E h * λ h 2 = 0 ,

and so,

(5.39) lim h 1 λ h B 1 1 E h * D G ( A h + λ h D v h ) D ψ d y = 0 .

By (5.35), (5.37), (5.38), and (5.39), passing to the limit as h in (5.34), we obtain

B 1 D F ( A ) D v D ψ d y 0 ,

and with ψ in place of ψ , we obtain

B 1 D F ( A ) D v D ψ d y = 0 ,

i.e., v solves a linear system with constant coefficients. By Proposition 2.1, we deduce that v C and, for every 0 < τ < 1 , we have

B τ 2 D v ( D v ) τ 2 2 c ( n , N , 1 , L 1 ) τ 2 B τ D v ( D v ) τ 2 d x c ( n , N , 1 , L 1 ) τ 2 ,

since

D v L 2 ( B 1 ) limsup h D v h L 2 ( B 1 ) c ( n ) .

Substep 3.b The case min { E h * , B 1 \ E h } = B 1 \ E h .

We claim that v solves the linear system

B 1 D 2 ( F + G ) ( A ) D v D ψ d y = 0 ,

for all ψ C 0 1 ( B 1 ; R N ) . Arguing as (5.16) and rescaling, we have that

B 1 H h ( D v h ) d y B 1 H h ( D v h + s D ψ ) + 1 λ h 2 B 1 \ E h [ G ( A h + λ h D v h ) G ( A h + λ h D v h + s λ h D ψ ) ] d y = B 1 H h ( D v h + s D ψ ) d y + 1 λ h B 1 \ E h 0 1 D G ( A h + λ h D v h + t s λ h D ψ ) s D ψ d t d y B 1 H h ( D v h + s D ψ ) d y + c ( p , L 2 ) λ h B 1 \ E h 0 1 ( 1 + A h + λ h D v h + t s λ h D ψ 2 ) p 1 2 s D ψ d t d y B 1 H h ( D v h + s D ψ ) d y + c ( p , L 2 , M ) 1 λ h B 1 \ E h s D ψ d y + B 1 \ E h 0 1 λ h p 2 D v h + t s D ψ p 1 s D ψ d t d y ,

for every ψ C 0 1 ( B 1 ; R N ) and for every s ( 0 , 1 ) . Therefore,

0 B 1 0 1 D H h ( D v h + s θ D ψ ) d θ s D ψ d y + c ( p , L 2 , M ) × 1 λ h B 1 \ E h s D ψ d y + B 1 \ E h 0 1 λ h p 2 D v h + t s D ψ p 1 s D ψ d t d y .

Dividing the previous inequality by s and taking the limit as s 0 , we obtain that

0 B 1 D H h ( D v h ) D ψ d y + c ( p , L 2 , M ) 1 λ h B 1 \ E h D ψ d y + B 1 \ E h λ h p 2 D v h p 1 D ψ d y .

By the definition of H h , we conclude that

0 1 λ h B 1 [ D ( F + G ) ( A h + λ h D v h ) D ψ D ( F + G ) ( A h ) D ψ ] d y + c ( p , L 2 , M ) 1 λ h B 1 \ E h D ψ d y + B 1 \ E h λ h p 2 D v h p 1 D ψ d y .

Just as earlier, we partition B 1 as

B 1 = B h + B h = { y B 1 : λ h D v h > 1 } { y B 1 : λ h D v h 1 } ,

and we write

(5.40) 0 1 λ h B 1 ( D ( F + G ) ( A h + λ h D v h ) D ( F + G ) ( A h ) ) D ψ d y + c ( p , L 2 , M ) 1 λ h B 1 \ E h D ψ d y + B 1 \ E h λ h p 2 D v h p 1 D ψ d y = 1 λ h B h + ( D ( F + G ) ( A h + λ h D v h ) D ( F + G ) ( A h ) ) D ψ d y + 1 λ h B h ( D ( F + G ) ( A h + λ h D v h ) D ( F + G ) ( A h ) ) D ψ d y + c ( p , L 2 , M ) 1 λ h B 1 \ E h D ψ d y + B 1 \ E h λ h p 2 D v h p 1 D ψ d y .

Arguing as in (5.35) , we obtain that

(5.41) lim h 1 λ h B h + ( D ( F + G ) ( A h + λ h D v h ) D ( F + G ) ( A h ) ) D ψ d y = 0 ,

and, as in (5.37) and (5.38),

(5.42) lim h 1 λ h B h [ D ( F + G ) ( A h + λ h D v h ) D ( F + G ) ( A h ) ] D ψ d y = B 1 D ( F + G ) ( A ) D v D ψ d y .

Moreover, we have that

1 λ h B 1 \ E h D ψ d y + B 1 \ E h λ h p 2 D v h p 1 D ψ d y c ( p , D ψ ) B 1 \ E h λ h + λ h B 1 λ h p 2 D v h p d y p 1 p B 1 \ E h λ h 2 1 p c ( n , p , D ψ ) B 1 \ E h λ h + λ h B 1 \ E h λ h 2 1 p ,

where we used (5.8). Since min { E h * , B 1 \ E h } = B 1 \ E h , by (5.11), we have

lim h B 1 \ E h λ h 2 = 0 ,

and we obtain

(5.43) lim h 1 λ h B 1 \ E h D ψ d y + B 1 \ E h λ h p 2 D v h p 1 D ψ d y = 0 .

By (5.41)–(5.43), passing to the limit as h in (5.40), we conclude that

B 1 D 2 ( F + G ) ( A ) D v D ψ d y 0 ,

and with ψ in place of ψ , we finally obtain

B 1 D 2 ( F + G ) ( A ) D v D ψ d y = 0 ,

asserting the claim. By Proposition 2.1, we deduce also in this case that v C and for every 0 < τ < 1 satisfies estimate (5.30).

Step 4. An estimate for the perimeters

Our aim is to show that there exists a constant c = c ( n , p , L 2 , Λ , M ) > 0 such that

(5.44) P ( E h , B τ ) c 1 τ P ( E h , B 1 ) n n 1 + r h τ n + r h λ h 2 .

By the minimality of ( u , E ) with respect to ( u , E ˜ ) , where E ˜ is a set of finite perimeter such that E ˜ Δ E B r h ( x h ) and B r h ( x h ) are the balls of the contradiction argument, we obtain

B r h ( x h ) 1 E G ( D u ) + Φ ( E ; B r h ( x h ) ) B r h ( x h ) 1 E ˜ G ( D u ) + Φ ( E ˜ ; B r h ( x h ) ) .

Using the change of variable x = x h + r h y and dividing by r h n 1 , we have

(5.45) r h B 1 1 E h G ( A h + λ h D v h ) d y + Φ h ( E h ; B 1 ) r h B 1 1 E ˜ h G ( A h + λ h D v h ) d y + Φ h ( E ˜ h ; B 1 ) ,

where

Φ h ( E h ; V ) V * E h Φ ( x h + r h y , ν E h ( y ) ) d n 1 ( y ) ,

for every Borel set V Ω . Assume first that min { B 1 \ E h , E h * } = B 1 \ E h . Choosing E ˜ h E h B ρ , we obtain

(5.46) Φ h ( E h ; B 1 ) r h B 1 1 B ρ G ( A h + λ h D v h ) d y + Φ h ( E ˜ h ; B 1 ) .

By coarea formula, the relative isoperimetric inequality, the choice of the representative E h ( 1 ) of E h , which is a Borel set, we obtain

τ 2 τ n 1 ( B ρ \ E h ) d ρ B 1 \ E h c ( n ) P ( E h , B 1 ) n n 1 .

Therefore, we may choose ρ ( τ , 2 τ ) , independent of n , such that, up to subsequences, it holds

(5.47) n 1 ( * E h B ρ ) = 0 and n 1 ( B ρ \ E h ) c ( n ) τ P ( E h , B 1 ) n n 1 .

We remark that Proposition 2.8 holds also for Φ h . Thus, thanks to the choice of ρ , being n 1 ( * E h B ρ ) = 0 , we have that

Φ h ( E ˜ h ; B 1 ) = Φ h ( E h ; B ρ ( 0 ) ) + Φ h ( B ρ ; E h ( 0 ) ) + Φ h ( E h ; { ν E h = ν B ρ } ) = Φ h ( E h ; B 1 \ B ρ ¯ ) + Φ h ( B ρ ; E h ( 0 ) ) .

By the choice of the representative of E h (see Remark 2.4), taking into account (2.8) and the inequality in (5.47), it follows that

(5.48) Φ h ( E ˜ h ; B 1 ) Φ h ( E h ; B 1 \ B ρ ¯ ) + Λ n 1 ( B ρ E h ( 0 ) ) Φ h ( E h ; B 1 \ B ρ ¯ ) + Λ n 1 ( B ρ \ E h ) . Φ h ( E h ; B 1 \ B ρ ¯ ) + Λ c ( n ) τ P ( E h , B 1 ) n n 1 .

On the other hand, by (2.8) and the additivity of the measure Φ h ( E h , ) it holds that

(5.49) 1 Λ P ( E h , B τ ) Φ h ( E h ; B τ ) Φ h ( E h ; B 1 ) Φ h ( E h ; B 1 \ B ¯ ρ ) ,

since ρ > τ . Combining (5.46), (5.48), and (5.49), we obtain

(5.50) 1 Λ P ( E h , B τ ) Φ h ( E h ; B 1 ) Φ h ( E h ; B 1 \ B ¯ ρ ) Φ h ( E ˜ h ; B 1 ) + r h B 1 1 B ρ G ( A h + λ h D v h ) d y Φ h ( E h ; B 1 \ B ¯ ρ ) Λ c ( n ) τ P ( E h , B 1 ) n n 1 + r h B 1 1 B ρ G ( A h + λ h D v h ) d y Λ c ( n ) τ P ( E h , B 1 ) n n 1 + c ( p , L 2 ) r h B 2 τ ( 1 + A h + λ h D v h 2 ) p 2 d y Λ c ( n ) τ P ( E h , B 1 ) n n 1 + c ( n , p , L 2 , M ) r h τ n + c ( p , L 2 ) r h λ h 2 B 2 τ λ h p 2 D v h p d y Λ c ( n ) τ P ( E h , B 1 ) n n 1 + c ( n , p , L 2 , M ) r h τ n + c ( n , p , L 2 ) r h λ h 2 ,

where we used (5.8). The previous estimate leads to (5.44). We reach the same conclusion if min { B 1 \ E h , E h * } = E h * , choosing E ˜ h = E h \ B ρ as a competing set.

Step 5. Conclusion.

By the change of variable x = x h + r h y and the Caccioppoli inequality in (5.18), for every 0 < τ < 1 4 we have

limsup h U * ( x h , τ r h ) λ h 2 limsup h 1 λ h 2 B τ r h ( x h ) [ D u ( x ) ( D u ) x h , τ r h 2 + D u ( x ) ( D u ) x h , τ r h p ] d x + limsup h P ( E , B τ r h ( x h ) ) λ h 2 τ n 1 r h n 1 + limsup h τ r h λ h 2 limsup h B τ [ D v h ( D v h ) τ 2 + λ h p 2 D v h ( D v h ) τ p ] d y + limsup h P ( E h , B τ ) λ h 2 τ n 1 + τ c ( n , p , 1 , 2 , L 2 , , Λ , M ) limsup h B 2 τ v h ( v h ) 2 τ ( D v h ) τ y 2 τ 2 + λ h p 2 v h ( v h ) 2 τ ( D v h ) τ y p τ p d y + 1 τ n limsup h P ( E h , B 1 ) n n 1 λ h 2 + 1 τ n 1 limsup h r h τ n λ h 2 + r h + τ ,

where we used (5.8) and estimate (5.50). We remark that

(5.51) lim h B 2 τ λ h p 2 v h ( v h ) 2 τ ( D v h ) τ y p d y = 0 ,

being p > 2 . Indeed, fixed r > p , we consider

p n p n p , if 2 < p < n , r , if p n .

There exists α ( 0 , 1 ) such that

1 p = 1 α 2 + α p .

Thus, we interpolate by Hölder’s inequality, and we obtain

B 2 τ λ h p 2 v h ( v h ) 2 τ ( D v h ) τ y p d y = λ h p 2 B 2 τ v h ( v h ) 2 τ ( D v h ) τ y p d y α p p B 2 τ v h ( v h ) 2 τ ( D v h ) τ y 2 d y ( 1 α ) p 2 .

On the one hand, by the Poincaré-Wirtinger inequality and (5.31), we obtain

lim h B 2 τ v h ( v h ) 2 τ ( D v h ) τ y 2 d y = B 2 τ v v 2 τ ( D v ) τ y 2 d y c ( n ) τ 2 B 2 τ D v ( D v ) τ 2 d y c ( n , N , 1 , 2 , L 1 , L 2 ) τ 2 .

On the other hand, by the Sobolev-Poincaré inequality, we infer

λ h p 2 B 2 τ v h ( v h ) 2 τ ( D v h ) τ y p d y α p p c ( n , p ) λ h p 2 B 2 τ D v h ( D v h ) τ p d y α c ( n , p ) λ h p 2 B 2 τ D v h p d y α = c ( n , p ) λ h ( p 2 ) ( 1 α ) B 2 τ λ h p 2 D v h p d y α c ( n , p ) λ h ( p 2 ) ( 1 α ) ,

where we used (5.30) and (5.8). Therefore, (5.51) follows at once.

By virtue of the strong convergence of v h v in L 2 ( B 1 ) , since ( D v h ) τ ( D v ) τ in R n × N , by (5.9)–(5.11), (5.31), (5.51), and by the Poincaré-Wirtinger inequality, we obtain

limsup h U * ( x h , τ r h ) λ h 2 c ( n , p , 1 , 2 , L 2 , Λ , M ) B 2 τ v ( v ) 2 τ ( D v ) τ y 2 τ 2 d y + τ c ( n , p , 1 , 2 , L 2 , Λ , M ) B 2 τ D v ( D v ) τ 2 d y + τ c ( n , N , p , 1 , 2 , L 1 , L 2 , Λ , M ) [ τ 2 + τ ] C ( n , N , p , 1 , 2 , L 1 , L 2 , Λ , M ) τ .

The contradiction follows, by choosing C * such that C * > C , since, by (5.5),

liminf h U * ( x h , τ r h ) λ h 2 C * τ .

Next, we prove a suitable decay estimate that allows us to prove Theorem 1.2 without the assumption ( H ) . To this aim, we introduce a new “hybrid" excess as

(5.52) U * * ( x 0 , r ) U ( x 0 , r ) + P ( E , B r ( x 0 ) ) r n 1 δ 1 + δ + r β ,

where U ( x 0 , r ) is defined in (5.1), δ has been determined in Theorem 4.1, and 0 < β < δ 1 + δ .

In the proof of Proposition 5.2, we will only elaborate the steps substantially different from the corresponding ones in the proof of Proposition 5.1.

Proposition 5.2

Let ( u , E ) be a local minimizer of under the assumptions ( F 1 ), ( F 2 ), ( G 1 ), and ( G 2 ). For every M > 0 and 0 < τ < 1 4 , there exist two positive constants ε 0 = ε 0 ( τ , M ) and c * * = c * * ( n , p , 1 , 2 , L 1 , L 2 , Λ , δ , M ) for which, whenever B r ( x 0 ) Ω verifies

( D u ) x 0 , r M and U * * ( x 0 , r ) ε 0 ,

then

(5.53) U * * ( x 0 , τ r ) c * * τ β U * * ( x 0 , r ) .

Proof

In order to prove (5.53), we argue by contradiction. Let M > 0 and τ ( 0 , 1 4 ) be such that for every h N , C * * > 0 , there exists a ball B r h ( x h ) Ω such that

(5.54) ( D u ) x h , r h M , U * * ( x h , r h ) 0

and

(5.55) U * * ( x h , τ r h ) C * * τ β U * * ( x h , r h ) .

The constant C * * will be determined later. We remark that we can confine ourselves to the case E B r h ( x h ) , the case B r h ( x h ) Ω \ E being easier because U = U * * r β .

Step 1. Blow-up.

We set λ h 2 U * * ( x h , r h ) , A h ( D u ) x h , r h , a h ( u ) x h , r h , and we define as earlier

v h ( y ) u ( x h + r h y ) a h r h A h y λ h r h , y B 1 .

One can easily check that ( D v h ) 0 , 1 = 0 and ( v h ) 0 , 1 = 0 . Again, as before, we set

E h E x h r h , E h * E x h r h B 1 .

Let us note that

(5.56) λ h 2 = U * * ( x h , r h ) = B 1 [ D u ( x h + r h y ) A h 2 + D u ( x h + r h y ) A h p ] d y + P ( E , B r h ( x h ) ) r h n 1 δ 1 + δ + r h β = B 1 [ λ h D v h 2 + λ h D v h p ] d y + P ( E h , B 1 ) δ 1 + δ + r h β .

It follows that

(5.57) r h 0 , P ( E h , B 1 ) 0 , r h β λ h 2 1 , B 1 [ D v h 2 + λ h p 2 D v h p ] 1 , P ( E h , B 1 ) δ 1 + δ λ h 2 1 .

Therefore, by virtue of (5.54), (5.56), and (5.57), there exist a (not relabeled) subsequence { v h } h N , A R n × N , and v W 1 , 2 ( B 1 ; R N ) , such that

(5.58) v h v weakly in W 1 , 2 ( B 1 ; R N ) , v h v , strongly in L 2 ( B 1 ; R N ) , A h A , λ h D v h 0 in L 2 ( B 1 ; R n N ) and pointwise a.e. ,

where we used the fact that ( v h ) 0 , 1 = 0 . We also note that

(5.59) r h δ 1 + δ λ h 2 = r h δ 1 + δ β r h β λ h 2 0 ,

since 0 < β < δ 1 + δ . Moreover, by (5.57), we deduce that

(5.60) lim h P ( E h , B 1 ) n n 1 δ 1 + δ λ h 2 = lim h P ( E h , B 1 ) δ ( n 1 ) ( 1 + δ ) limsup h P ( E h , B 1 ) δ 1 + δ λ h 2 = 0 .

Therefore, by the relative isoperimetric inequality,

(5.61) lim h min E h * δ 1 + δ λ h 2 , B 1 \ E h δ 1 + δ λ h 2 c ( n , δ ) lim h P ( E h , B 1 ) n δ ( 1 + δ ) ( n 1 ) λ h 2 = 0 .

Step 2. A Caccioppoli-type inequality.

We claim that there exists a constant c = c ( n , p , 1 , L 1 , L 2 , M ) > 0 such that, for every 0 < ρ < 1 , there exists h 0 N such that for all h > h 0 , we have

(5.62) B ρ 2 ( 1 + λ h p 2 D v h ( D v h ) ρ 2 p 2 ) D v h ( D v h ) ρ 2 2 d y c B ρ v h ( v h ) ρ ( D v h ) ρ 2 y 2 ρ 2 + λ h p 2 v h ( v h ) ρ ( D v h ) ρ 2 y p ρ p d y + P ( E h , B 1 ) n δ ( n 1 ) ( 1 + δ ) ρ n δ 1 + δ λ h 2 .

We divide the proof into two substeps.

Substep 2.a The case min { E h * , B 1 \ E h } = E h *

The proof of this substep goes exactly as that of Substep 2.a of Proposition 5.1 up to estimate (5.24). Next, we observe that

B ρ 2 ( 1 + λ h p 2 D v h ( D v h ) ρ 2 p 2 ) D v h ( D v h ) ρ 2 2 d y c B ρ v h ( v h ) ρ ( D v h ) ρ 2 y 2 ρ 2 + λ h p 2 v h ( v h ) ρ ( D v h ) ρ 2 y p ρ p d y + E h * λ h 2 c B ρ v h ( v h ) ρ ( D v h ) ρ 2 y 2 ρ 2 + λ h p 2 v h ( v h ) ρ ( D v h ) ρ 2 y p ρ p d y + E h * δ 1 + δ λ h 2 ,

and this, by the relative isoperimetric inequality, yields the estimate (5.62).

Substep 2.b The case min { E h * , B 1 \ E h } = B 1 \ E h

We fix 0 < ρ 2 < s < t < ρ < 1 , and let η C 0 ( B t ) be a cutoff function between B s and B t , i.e., 0 η 1 , η 1 on B s and η c t s . Furthermore, we set b h ( v h ) B ρ , B h ( D v h ) B ρ 2 and define

w h ( y ) v h ( y ) b h B h y , ψ 1 , h η w h and ψ 2 , h ( 1 η ) w h ,

for any y B 1 . By (5.26) and (5.29), we obtain

(5.63) ˜ B s ( 1 + λ h D w h 2 ) p 2 2 D w h 2 d y B t H ˜ h ( D w h ) d y B t 0 1 D H ˜ h ( D w h θ D ψ 2 , h ) D ψ 2 , h d θ d y B t H ˜ h ( D w h ) d y + c ( p , M ) B t \ B s ( D w h 2 + λ h p 2 D w h p + D ψ 2 , h 2 + λ h p 2 D ψ 2 , h p ) d y .

In order to estimate the first addend of the right-hand side of the previous inequality, we recall that

(5.64) B t H ˜ h ( D w h ) d y B t H ˜ h ( D w h ( y ) D ψ 1 , h ) d y + L 2 λ h 2 B t \ E h ( 1 + A h + λ h D v h 2 ) p 2 d y = B t \ B s H ˜ h ( D ψ 2 , h ) d y + L 2 λ h 2 B t \ E h ( 1 + A h + λ h D v h 2 ) p 2 d y .

We remark that the reverse Hölder inequality stated in Theorem 4.1, through the change of variable x = x h + r h y , can be rescaled in the following way:

(5.65) B t A h + λ h D v h p ( 1 + δ ) d y 1 1 + δ c ( n , p , 1 , L 1 , L 2 ) B 2 t A h + λ h D v h p d y + 1 .

By Hölder’s inequality and inserting the previous inequality into estimate (5.64), we obtain

B t H ˜ h ( D w h ) d y B t \ B s H ˜ h ( D ψ 2 , h ) d y + c ( p ) L 2 λ h 2 B t \ E h ( 1 + A h + λ h D v h p ( 1 + δ ) ) d y 1 1 + δ B t \ E h δ 1 + δ B t \ B s H ˜ h ( D ψ 2 , h ) d y + c ( p ) L 2 λ h 2 t n 1 + δ B t ( 1 + A h + λ h D v h p ( 1 + δ ) ) d y 1 1 + δ B 1 \ E h δ 1 + δ B t \ B s H ˜ h ( D ψ 2 , h ) d y + c ( n , p , 1 , L 1 , L 2 ) t n 1 + δ λ h 2 1 + B 2 t A h + λ h D v h p d y B 1 \ E h δ 1 + δ B t \ B s H ˜ h ( D ψ 2 , h ) d y + c ( n , p , 1 , L 1 , L 2 , M ) ρ n δ 1 + δ B 1 \ E h δ 1 + δ λ h 2 ,

where we used the fact that t > ρ 2 . Hence, inserting the previous estimate into (5.63), we obtain

˜ B s ( 1 + λ h D w h 2 ) p 2 2 D w h 2 d y B t \ B s H ˜ h ( D ψ 2 , h ) d y + c ( n , p , 1 , L 1 , L 2 , M ) ρ n δ 1 + δ B 1 \ E h δ 1 + δ λ h 2 + c ( p , M ) B t \ B s ( D w h + D ψ 2 , h + λ h p 2 D w h p 1 + λ h p 2 D ψ 2 , h p 1 ) D ψ 2 , h d y .

Thanks to the Lemma 2.3, Young’s inequality, and the properties of η , we obtain

˜ B s ( 1 + λ h D w h 2 ) p 2 2 D w h 2 d y c ( n , p , 1 , L 1 , L 2 , M ) B t \ B s [ D ψ 2 , h 2 + λ h p 2 D ψ 2 , h p ] d y + 1 ρ n δ 1 + δ B 1 \ E h δ 1 + δ λ h 2 + B t \ B s [ D w h 2 + λ h p 2 D w h p ] d y c ( n , p , 1 , L 1 , L 2 , M ) B t \ B s ( 1 + λ h 2 D w h 2 ) p 2 2 D w h 2 d y + B ρ w h 2 ( t s ) 2 + λ h p 2 w h p ( t s ) p d y + 1 ρ n δ 1 + δ B 1 \ E h δ 1 + δ λ h 2 .

Using the hole-filling technique as in the proof of the previous theorem, we obtain

B s ( 1 + λ h D w h 2 ) p 2 2 D w h 2 d y c c + ˜ B t ( 1 + λ h D w h 2 ) p 2 2 D w h 2 d y + B ρ w h 2 ( t s ) 2 + λ h p 2 w h p ( t s ) p d y + 1 ρ n δ 1 + δ B 1 \ E h δ 1 + δ λ h 2

By virtue of the iteration Lemma 2.2, the previous estimate gives

B ρ 2 ( 1 + λ h D w h 2 ) p 2 2 D w h 2 d y c B ρ w h 2 ρ 2 + λ h p 2 w h p ρ p d y + 1 ρ n δ 1 + δ B 1 \ E h δ 1 + δ λ h 2 ,

where c = c ( n , p , 1 , 2 , L 1 , L 2 , M ) . Therefore, by the definition of w h , we have

B ρ 2 ( 1 + λ h p 2 D v h ( D v h ) ρ 2 p 2 ) D v h ( D v h ) ρ 2 2 d y c B ρ v h ( v h ) ρ ( D v h ) ρ 2 y 2 ρ 2 + λ h p 2 v h ( v h ) ρ ( D v h ) ρ 2 y p ρ p d y + 1 ρ n δ 1 + δ B 1 \ E h δ 1 + δ λ h 2 ,

which, by the relative isoperimetric inequality and since B 1 \ E h = min { E h * , B 1 \ E h } , gives the estimate (5.62).

The proofs of Step 3 and Step 4 of Proposition 5.1 hold true also in this case.

Step 5. Conclusion

The change of variable x = x h + r h y , the Caccioppoli inequality in (5.62) and (5.50), for every 0 < τ < 1 4 , give

limsup h U * * ( x h , τ r h ) λ h 2 limsup h 1 λ h 2 B τ r h ( x h ) [ D u ( D u ) x h , τ r h 2 + D u ( D u ) x h , τ r h p ] d x + limsup h 1 λ h 2 P ( E , B τ r h ) ( x h ) τ n 1 r h n 1 δ 1 + δ + limsup h τ β r h β λ h 2 limsup h B τ [ D v h ( D v h ) τ 2 + λ h p 2 D v h ( D v h ) τ p ] d y + limsup h 1 λ h 2 P ( E h , B τ ) τ n 1 δ 1 + δ + τ β c limsup h B 2 τ v h ( v h ) 2 τ ( D v h ) τ y 2 τ 2 + λ h p 2 v h ( v h ) 2 τ ( D v h ) τ y p τ p d y + limsup h P ( E h , B 1 ) n δ ( n 1 ) ( 1 + δ ) λ h 2 1 τ n δ 1 + δ + r h δ 1 + δ λ h 2 τ δ 1 + δ + r h δ 1 + δ λ h 2 λ h 2 δ 1 + δ τ ( n 1 ) δ 1 + δ + τ β c limsup h B 2 τ v h ( v h ) 2 τ ( D v h ) τ y 2 τ 2 + λ h p 2 v h ( v h ) 2 τ ( D v h ) τ y p τ p d y + limsup h P ( E h , B 1 ) n δ ( n 1 ) ( 1 + δ ) λ h 2 1 τ n δ 1 + δ + r h δ 1 + δ λ h 2 τ δ 1 + δ + r h δ 1 + δ λ h 2 λ h 2 δ 1 + δ τ ( n 1 ) δ 1 + δ + τ β ,

where c = c ( n , p , 1 , 2 , L 1 , L 2 , Λ , δ , M ) . Proceeding exactly as in Step 5 of Proposition 5.1, by virtue of the strong convergence of v h v in L 2 ( B 1 ) , since ( D v h ) τ ( D v ) τ in R n N , by (5.31), (5.51), (5.57), (5.58), (5.59) (5.60), (5.61), and by the use of Poincaré-Wirtinger inequality, we obtain

limsup h U * * ( x h , τ r h ) λ h 2 C τ β ,

where C = C ( n , p , 1 , 2 , L 1 , L 2 , Λ , δ , M ) . The contradiction follows by choosing C * * such that C * * > C , since by (5.55)

liminf h U * * ( x h , τ r h ) λ h 2 C * * τ β .

6 Proof of the main theorem

Here, we give the proof of Theorem 1.2 through a suitable iteration procedure. It is easy to obtain the following Lemmata, arguing exactly in the same way as in [11, Lemma 6.1].

Lemma 6.1

Let ( u , E ) be a minimizer of the functional . For every M > 0 , α ( 0 , 1 ) , and ϑ ( 0 , ϑ 0 ) , with ϑ 0 min c * 1 1 α , 1 4 , there exist 0 < ε 1 ( 1 ϑ 1 2 ) 2 ϑ n 1 and R > 0 such that if r < R and x 0 Ω satisfy

B r ( x 0 ) Ω , D u x 0 , r < M , and U * ( x 0 , r ) < ε 1 ,

where c * is the constant introduced in Proposition 5.1, then

U * ( x 0 , ϑ k r ) ϑ k α U * ( x 0 , r ) , k N .

Lemma 6.2

Let ( u , E ) be a minimizer of the functional , and let β be the exponent of Lemma 5.2. For every M > 0 and ϑ ( 0 , ϑ 0 ) , with ϑ 0 < min c * * , 1 4 , there exist ε 1 > 0 and R > 0 such that if r < R and x 0 Ω satisfy

B r ( x 0 ) Ω , D u x 0 , r < M a n d U * * ( x 0 , r ) < ε 1 ,

where c * * is the constant introduced in Proposition 5.2, then

U * * ( x 0 , ϑ k r ) ϑ k β U * * ( x 0 , r ) , k N .

6.1 Proof of Theorem 1.2

Proof

We consider the set

Ω 0 { x Ω : limsup ρ 0 ( D u ) x , ρ < and limsup ρ 0 U * ( x , ρ ) = 0 } ,

and let x 0 Ω 0 . For every M > 0 and for ε 1 determined in Lemma 6.1, there exists a radius R M , ε 1 > 0 such that

D u x 0 , r < M and U * ( x 0 , r ) < ε 1 ,

for every 0 < r < R M , ε 1 . If 0 < ρ < ϑ r < R , let h N be such that ϑ h + 1 r < ρ < ϑ h r , where ϑ = ϑ 0 2 and ϑ 0 is the same constant appearing in Lemma 6.1. By Lemma 6.1, we obtain

U * ( x 0 , ρ ) c ( p ) B ρ ( x 0 ) [ D u ( D u ) x 0 , ϑ h r 2 + D u ( D u ) x 0 , ϑ h r p ] d x + c ( p ) [ ( D u ) x 0 , ϑ h r ( D u ) x 0 , ρ 2 + ( D u ) x 0 , ϑ h r ( D u ) x 0 , ρ p ] + P ( E , B ρ ( x 0 ) ) ρ n 1 + ρ c ( p ) B ρ ( x 0 ) [ D u ( D u ) x 0 , ϑ h r 2 + D u ( D u ) x 0 , ϑ h r p ] d x + P ( E , B ρ ( x 0 ) ) ρ n 1 + ρ c ( p ) ϑ h + 1 r ϑ ρ n B ϑ h r ( x 0 ) [ D u ( D u ) x 0 , ϑ h r 2 + D u ( D u ) x 0 , ϑ h r p ] d x + P ( E , B ϑ h r ( x 0 ) ) ( ϑ h + 1 r ) n 1 + ϑ h r c ( p ) ϑ n B ϑ h r ( x 0 ) [ D u ( D u ) x 0 , ϑ h r 2 + D u ( D u ) x 0 , ϑ h r p ] d x + 1 ϑ n 1 P ( E , B ϑ h r ( x 0 ) ) ( ϑ h r ) n 1 + ϑ h r = c ( n , p ) ϑ 0 n B ϑ h r ( x 0 ) [ D u ( D u ) x 0 , ϑ h r 2 + D u ( D u ) x 0 , ϑ h r p ] d x + 2 n 1 ϑ 0 n 1 P ( E , B ϑ h r ( x 0 ) ) ( ϑ h r ) n 1 + ϑ h r c ( n , p , ϑ 0 ) U * ( x 0 , ϑ h r ) c ( n , p , ϑ 0 ) c * ϑ h α U * ( x 0 , r ) c ( n , p , ϑ 0 ) c * ρ r α U * ( x 0 , r ) ,

where we used Jensen’s inequality to estimate

( D u ) x 0 , ϑ h r ( D u ) x 0 , ρ 2 + ( D u ) x 0 , ϑ h r ( D u ) x 0 , ρ p = B ρ ( x 0 ) [ D u ( D u ) x 0 , ϑ h r ] d x 2 + B ρ ( x 0 ) [ D u ( D u ) x 0 , ϑ h r ] d x p B ρ ( x 0 ) D u ( D u ) x 0 , ϑ h r 2 d x + B ρ ( x 0 ) D u ( D u ) x 0 , ϑ h r p d x .

The previous estimate implies that

U ( x 0 , ρ ) C * ρ r α U * ( x 0 , r ) ,

where C * = C * ( n , p , θ 0 , c * ) . Since U * ( y , r ) is continuous in y , we have that U * ( y , r ) < ε 1 for all y in a suitable neighborhood I of x 0 . Therefore, for every y I , we have that

U ( y , ρ ) C * ρ r α U * ( y , r ) .

The last inequality implies, by the Campanato characterization of Hölder continuous functions (see [32, Theorem 2.9]), that u is C 1 , α in I for every 0 < α < 1 2 , and we can conclude that the set Ω 0 is open and the function u has Hölder continuous derivatives in Ω 0 .

When the assumption ( H ) is not enforced, the proof goes exactly in the same way provided we use Lemma 6.2 in place of Lemma 6.1, with

Ω 1 { x Ω : limsup ρ 0 ( D u ) x 0 , ρ < and limsup ρ 0 U * * ( x 0 , ρ ) = 0 } .

Acknowledgements

The authors are members of the Gruppo Nazionale per l’Analisi Matematica, la Probabilità e le loro Applicazioni (GNAMPA) of the Istituto Nazionale di Alta Matematica (INdAM). M. Carozza acknowledged support by GNAMPA.

  1. Author contributions: All authors have accepted their responsibility for the entire content of this manuscript and consented to its submission to the journal, reviewed all the results and approved the final version of the manuscript. The authors MC, LE and LL contributed equally to the entire manuscript.

  2. Conflict of interest: The authors state no conflict of interest.

  3. Data availability statement: Data sharing not applicable to this article as no data sets were generated or analyzed during the current study.

References

[1] E. Acerbi and N. Fusco, An approximation lemma for W1;p functions, Material instabilities in continuum mechanics (Edinburgh, 19851986), Oxford Sci. Publ., Oxford Univ. Press, New York, 1988, pp. 1–5. Search in Google Scholar

[2] E. Acerbi and N. Fusco, A regularity theorem for minimizers of quasiconvex integrals, Arch. Ration. Mech. Anal. 99 (1987), 261–281. 10.1007/BF00284509Search in Google Scholar

[3] F. J. Almgren, Existence and regularity almost everywhere of solutions to elliptic variational problems among surfaces of varying topological type and singularity structure, Ann. Math. 87 (1968), 321–391. 10.2307/1970587Search in Google Scholar

[4] L. Ambrosio and G. Buttazzo, An optimal design problem with perimeter penalization. Calc. Var. Partial Differential Equations 1 (1993), 55–69. 10.1007/BF02163264Search in Google Scholar

[5] L. Ambrosio, N. Fusco, and D. Pallara, Functions of Bounded Variation and Free Discontinuity Problems, Oxford University Press, Oxford, United Kingdom, 2000. 10.1093/oso/9780198502456.001.0001Search in Google Scholar

[6] A. Arroyo-Rabasa, Regularity for free interface variational problems in a general class of gradients, Calc. Var. Partial Differential Equations 55 (2016), 1–44. 10.1007/s00526-016-1085-5Search in Google Scholar

[7] G. Bellettini, M. Novaga, and M. Paolini, On a crystalline variational problem, Part I: First variation and global L∞ regularity, Arch. Rational Mech. Anal. 157 (2001), 165–191. 10.1007/s002050010127Search in Google Scholar

[8] G. Bellettini, M. Novaga, and M. Paolini, On a crystalline variational problem, Part II: BV regularity and structure of minimizers on facets, Arch. Rational Mech. Anal. 157 (2001), 193–217. 10.1007/s002050100126Search in Google Scholar

[9] E. Bombieri, Regularity theory for almost minimal currents, Arch. Ration. Mech. Anal. 78 (1982), 99–130. 10.1007/BF00250836Search in Google Scholar

[10] M. Carozza, I. Fonseca, and A. Passarelli Di Napoli, Regularity results for an optimal design problem with a volume constraint, ESAIM Control Optim. Calc. Var. 20 (2014), no. 2, 460–487. 10.1051/cocv/2013071Search in Google Scholar

[11] M. Carozza, I. Fonseca, and A. Passarelli Di Napoli, Regularity results for an optimal design problem with quasiconvex bulk energies, Calc. Var. Partial Differential Equations 57 (2018), 68. 10.1007/s00526-018-1343-9Search in Google Scholar

[12] M. Carozza, N. Fusco, and G. Mingione, Partial regularity of minimizers of quasiconvex integrals with subquadratic growth, Ann. Mat. Pura Appl. 175 (1998), 141–164. 10.1007/BF01783679Search in Google Scholar

[13] M. Carozza and G. Mingione, Partial regularity of minimizers of quasiconvex integrals of subquadratic growth: the general case, Ann. Pol. Math. 77 (2001), 219–243. 10.4064/ap77-3-3Search in Google Scholar

[14] M. Carozza and A. Passarelli Di Napoli, A regularity theorem for minimisers of quasiconvex integrals: The case 1<p<2, Proc. Roy. Soc. Edinburgh Sect. A. 126 (1996), 1181–1200. 10.1017/S0308210500023350Search in Google Scholar

[15] M. Carozza and A. Passarelli Di Napoli, Partial regularity of local minimizers of quasiconvex integrals with sub-quadratic growth. Proc. Roy. Soc. Edinburgh Sect. A 133 (2003), 1249–1262. 10.1017/S0308210500002900Search in Google Scholar

[16] R. Choksi, R. Neumayer, and I. Topaloglu, Anisotropic liquid drop models, Adv. Calc. Var. 15, no. 1, (2022), 109–131. 10.1515/acv-2019-0088Search in Google Scholar

[17] G. De Philippis and A. Figalli, A note on the dimension of the singular set in free interface problems, Differ. Integral Equ. 28 (2015), 523–536. 10.57262/die/1427744099Search in Google Scholar

[18] G. De Philippis, N. Fusco and M. Morini, Regularity of capillarity droplets with obstacle, (2022), http://arxiv.org/abs/2205.15272. Search in Google Scholar

[19] G. De Philippis and F. Maggi, Dimensional estimates for singular sets in geometric variational problems with free boundaries, J. Reine Angew. Math. 725 (2017), 217–234. 10.1515/crelle-2014-0100Search in Google Scholar

[20] F. Duzaar and K. Steffen, Optimal interior and boundary regularity for almost minimizers to elliptic variational integrals, J. Reine Angew. Math., 546 (2002), 73–138. 10.1515/crll.2002.046Search in Google Scholar

[21] L. Esposito, Density lower bound estimate for local minimizer of free interface problem with volume constraint, Ric. di Mat. 68 (2019), no. 2, 359–373. 10.1007/s11587-018-0407-7Search in Google Scholar

[22] L. Esposito and N. Fusco, A remark on a free interface problem with volume constraint, J. Convex Anal. 18 (2011), no. 2, 417–426. Search in Google Scholar

[23] L. Esposito and L. Lamberti, Regularity results for an optimal design problem with lower order terms, Adv. Calc. Var. 16 (2023), no. 4, 1093–1122.10.1515/acv-2021-0080Search in Google Scholar

[24] L. Esposito and L. Lamberti, Regularity results for a free interface problem with Hölder coefficients, Calc. Var. Partial Differential Equations 62 (2023), 156. 10.1007/s00526-023-02490-xSearch in Google Scholar

[25] A. Figalli, Regularity of codimension-1 minimizing currents under minimal assumptions on the integrand, J. Differential Geom. 106 (2017), 371–391. 10.4310/jdg/1500084021Search in Google Scholar

[26] A. Figalli and F. Maggi, On the shape of liquid drops and crystals in the small mass regime, Arch. Rational Mech. Anal. 201 (2011), 143–207. 10.1007/s00205-010-0383-xSearch in Google Scholar

[27] N. Fusco and V. Julin, On the regularity of critical and minimal sets of a free interface problem, Interfaces Free Bound. 17 (2015), no. 1, 117–142. 10.4171/ifb/336Search in Google Scholar

[28] I. Fonseca and S. Müller, A-quasiconvexity, lower semicontinuity, and Young measures, SIAM J. Math. Anal. 30 (1999), no. 6, 1355–1390. 10.1137/S0036141098339885Search in Google Scholar

[29] I. Fonseca, S. Müller, and P. Pedregal, Analysis of concentration and oscillation effects generated by gradients, SIAM J. Math. Anal. 29 (1998), no. 3, 736–756. 10.1137/S0036141096306534Search in Google Scholar

[30] M. Giaquinta, Multiple integrals in the calculus of variations and nonlinear elliptic systems, Annals of Mathematical Studies 105, Princeton University Press, Princeton NJ, 1983. 10.1515/9781400881628Search in Google Scholar

[31] M. Giaquinta and G. Modica, Partial regularity of minimizers of quasiconvex integrals, Anno. Inst. H. Poincaré, Anal. Non Linéaire 3 (1986), 185–208. 10.1016/s0294-1449(16)30385-7Search in Google Scholar

[32] E. Giusti, Direct Methods in the Calculus of Variations, World Scientific, Singapore, 2003. 10.1142/9789812795557Search in Google Scholar

[33] M. Gurtin, On phase transitions with bulk, interfacial, and boundary energy, Arch. Rationo. Mech. Anal. 96 (1986), 243–264. 10.1007/BF00251908Search in Google Scholar

[34] L. Lamberti, A regularity result for minimal configurations of a free interface problem, Boll. Uno. Mat. Ital. 14 (2021), 521–539. 10.1007/s40574-021-00285-6Search in Google Scholar

[35] F.H. Lin, Variational problems with free interfaces, Calc. Var. Partial Differential Equations 1 (1993), 149–168. 10.1007/BF01191615Search in Google Scholar

[36] F.H. Lin and R.V. Kohn, Partial regularity for optimal design problems involving both bulk and surface energies, Chinese Anno. of Math. Ser. B 20 (1999), no. 2, 137–158. 10.1142/S0252959999000175Search in Google Scholar

[37] F. Maggi, Sets of Finite Perimeter and Geometric Variational Problems: An Introduction to Geometric Measure Theory, Cambridge University Press, Cambridge, 2012. 10.1017/CBO9781139108133Search in Google Scholar

[38] P. Marcellini, Approximation of quasiconvex functionals and lower semicontinuity of multiple integrals, Manuscripta Math. 51 (1985), 1–28. 10.1007/BF01168345Search in Google Scholar

[39] R. Schoen and L. Simon, A new proof of the regularity theorem for rectifiable currents which minimize parametric elliptic functionals, Indiana Univ. Math. J. 31 (1982), 415–434. 10.1512/iumj.1982.31.31035Search in Google Scholar

[40] I. Tamanini, Boundaries of Caccioppoli sets with Hölder-continuous normal vector, J. Reine Angew. Math. 334 (1982), 27–39. 10.1515/crll.1982.334.27Search in Google Scholar

Received: 2023-07-28
Revised: 2023-12-14
Accepted: 2024-05-06
Published Online: 2024-08-29

© 2024 the author(s), published by De Gruyter

This work is licensed under the Creative Commons Attribution 4.0 International License.

Articles in the same Issue

  1. Regular Articles
  2. Existence of normalized peak solutions for a coupled nonlinear Schrödinger system
  3. Orbital stability of the trains of peaked solitary waves for the modified Camassa-Holm-Novikov equation
  4. Global boundedness in a two-dimensional chemotaxis system with nonlinear diffusion and singular sensitivity
  5. Existence and uniqueness of solution for a singular elliptic differential equation
  6. The bounded variation capacity and Sobolev-type inequalities on Dirichlet spaces
  7. Vanishing and blow-up solutions to a class of nonlinear complex differential equations near the singular point
  8. Existence, uniqueness, localization and minimization property of positive solutions for non-local problems involving discontinuous Kirchhoff functions
  9. Concentration phenomena for a fractional relativistic Schrödinger equation with critical growth
  10. Beyond the classical strong maximum principle: Sign-changing forcing term and flat solutions
  11. Monotonicity of solutions for parabolic equations involving nonlocal Monge-Ampère operator
  12. On a nonlinear Robin problem with an absorption term on the boundary and L1 data
  13. Multiple solutions for the quasilinear Choquard equation with Berestycki-Lions-type nonlinearities
  14. Gradient estimates for a class of higher-order elliptic equations of p-growth over a nonsmooth domain
  15. Global existence and decay estimates of the classical solution to the compressible Navier-Stokes-Smoluchowski equations in ℝ3
  16. Global existence and finite-time blowup for a mixed pseudo-parabolic r(x)-Laplacian equation
  17. Multiple positive solutions for a class of concave-convex Schrödinger-Poisson-Slater equations with critical exponent
  18. A minimization problem with free boundary for p-Laplacian weakly coupled system
  19. Multiplicity of semiclassical solutions for a class of nonlinear Hamiltonian elliptic system
  20. k-convex solutions for multiparameter Dirichlet systems with k-Hessian operator and Lane-Emden type nonlinearities
  21. Infinitely many solutions for Hamiltonian system with critical growth
  22. On optimal control in a nonlinear interface problem described by hemivariational inequalities
  23. Variational–hemivariational system for contaminant convection–reaction–diffusion model of recovered fracturing fluid
  24. Potential and monotone homeomorphisms in Banach spaces
  25. Critical fractional Schrödinger-Poisson systems with lower perturbations: the existence and concentration behavior of ground state solutions
  26. Supercritical Hénon-type equation with a forcing term
  27. Concentration of blow-up solutions for the Gross-Pitaveskii equation
  28. Mountain-pass-type solutions for Schrödinger equations in R2 with unbounded or vanishing potentials and critical exponential growth nonlinearities
  29. Regularity for critical fractional Choquard equation with singular potential and its applications
  30. Double phase anisotropic variational problems involving critical growth
  31. Normalized solutions of NLS equations with mixed nonlocal nonlinearities
  32. Nontrivial solutions for resonance quasilinear elliptic systems
  33. Standing waves for Choquard equation with noncritical rotation
  34. Low regularity conservation laws for Fokas-Lenells equation and Camassa-Holm equation
  35. Modified quasilinear equations with strongly singular and critical exponential nonlinearity
  36. Limit profiles and the existence of bound-states in exterior domains for fractional Choquard equations with critical exponent
  37. Nests of limit cycles in quadratic systems
  38. Regularity of minimizers for double phase functionals of borderline case with variable exponents
  39. Boundedness, existence and uniqueness results for coupled gradient dependent elliptic systems with nonlinear boundary condition
  40. Ground state solutions for magnetic Schrödinger equations with polynomial growth
  41. Nonoccurrence of Lavrentiev gap for a class of functionals with nonstandard growth
  42. Dynamics for wave equations connected in parallel with nonlinear localized damping
  43. Boussinesq's equation for water waves: Asymptotics in Sector I
  44. Optimal global second-order regularity and improved integrability for parabolic equations with variable growth
  45. Normalized solutions of Schrödinger equations involving Moser-Trudinger critical growth
  46. Normalized solutions for the double-phase problem with nonlocal reaction
  47. Normalized solutions for Sobolev critical fractional Schrödinger equation
  48. Well-posedness of Cauchy problem of fractional drift diffusion system in non-critical spaces with power-law nonlinearity
  49. Improved results on planar Klein-Gordon-Maxwell system with critical exponential growth
  50. Existence and multiplicity of solutions for a new p(x)-Kirchhoff equation
  51. Quasiconvex bulk and surface energies: C1,α regularity
  52. Time decay estimates of solutions to a two-phase flow model in the whole space
  53. Bounds for the sum of the first k-eigenvalues of Dirichlet problem with logarithmic order of Klein-Gordon operators
  54. The properties of a new fractional g-Laplacian Monge-Ampère operator and its applications
  55. A fractional profile decomposition and its application to Kirchhoff-type fractional problems with prescribed mass
  56. Cauchy problem for a non-Newtonian filtration equation with slowly decaying volumetric moisture content
  57. Multiplicity of normalized semi-classical states for a class of nonlinear Choquard equations
  58. The second Yamabe invariant with boundary
  59. Peaked solitary waves and shock waves of the Degasperis-Procesi-Kadomtsev-Petviashvili equation
  60. Normalized solutions for the Choquard equations with critical nonlinearities
  61. Boundedness and long-time behavior in a parabolic-elliptic system arising from biological transport networks
  62. Initial boundary value problem and exponential stability for the planar magnetohydrodynamics equations with temperature-dependent viscosity
  63. Nonexistence of mean curvature flow solitons with polynomial volume growth immersed in certain semi-Riemannian warped products
  64. Analysis of a vector-borne disease model with vector-bias mechanism in advective heterogeneous environment
  65. Normalized solutions for the Kirchhoff equation with combined nonlinearities in ℝ4
  66. Nonexistence of the compressible Euler equations with space-dependent damping in high dimensions
  67. Multiplicity of solutions for a nonhomogeneous quasilinear elliptic equation with concave-convex nonlinearities
  68. On semilinear inequalities involving the Dunkl Laplacian and an inverse-square potential outside a ball
  69. Besov regularity for the elliptic p-harmonic equations in the non-quadratic case
  70. Uniqueness and nondegeneracy of ground states for Δ u + u = ( I α u 2 ) u in R 3 when α is close to 2
  71. Existence of solutions for a class of quasilinear Schrödinger equations with Choquard-type nonlinearity
  72. Weighted Hardy-Adams inequality on unit ball of any even dimension
  73. Existence and regularity for a p-Laplacian problem in ℝN with singular, convective, and critical reaction
  74. Temporal periodic solutions of non-isentropic compressible Euler equations with geometric effects
  75. Nodal solutions for a zero-mass Chern-Simons-Schrödinger equation
  76. The modified quasi-boundary-value method for an ill-posed generalized elliptic problem
  77. Nonlocal heat equations with generalized fractional Laplacian
  78. Choquard equations with recurrent potentials
  79. Local and global well-posedness of the Maxwell-Bloch system of equations with inhomogeneous broadening
  80. The Riemann problem for two-layer shallow water equations with bottom topography
  81. Global stability and asymptotic profiles of a partially degenerate reaction diffusion Cholera model with asymptomatic individuals
  82. On Kirchhoff-Schrödinger-Poisson-type systems with singular and critical nonlinearity
  83. Energy-variational solutions for viscoelastic fluid models
  84. Stability on 3D Boussinesq system with mixed partial dissipation
  85. Existence and multiplicity results for non-autonomous second-order Hamiltonian systems
  86. Erratum
  87. Erratum to “Normalized solutions for the Kirchhoff equation with combined nonlinearities in ℝ4
Downloaded on 7.9.2025 from https://www.degruyterbrill.com/document/doi/10.1515/anona-2024-0021/html
Scroll to top button