
The integration of robotics education with science, technology, engineering, and mathematics (STEM) education has a great potential in future education. In recent years, numerous countries have hosted robotic competitions. This study uses a mixed research method to explore the coaches' views on student participation in the World Robot Olympiad (WRO) by incorporating the questionnaire surveys and interviews conducted at the 2019 WRO finals in Hungary. By quantitative and qualitative analyses, coaches generally agreed that participation in the WRO improved students' STEM learning skills and cultivated their patience and resilience in handling challenging tasks.
Citation: Yicong Zhang, Yanan Lu, Xianqing Bao, Feng-Kuang Chiang. Impact of participation in the World Robot Olympiad on K-12 robotics education from the coach's perspective[J]. STEM Education, 2022, 2(1): 37-46. doi: 10.3934/steme.2022002
[1] | Glen Wheeler . Convergence for global curve diffusion flows. Mathematics in Engineering, 2022, 4(1): 1-13. doi: 10.3934/mine.2022001 |
[2] | Takeyuki Nagasawa, Kohei Nakamura . Asymptotic analysis for non-local curvature flows for plane curves with a general rotation number. Mathematics in Engineering, 2021, 3(6): 1-26. doi: 10.3934/mine.2021047 |
[3] | Masashi Misawa, Kenta Nakamura, Yoshihiko Yamaura . A volume constraint problem for the nonlocal doubly nonlinear parabolic equation. Mathematics in Engineering, 2023, 5(6): 1-26. doi: 10.3934/mine.2023098 |
[4] | Jerome D. Wettstein . Half-harmonic gradient flow: aspects of a non-local geometric PDE. Mathematics in Engineering, 2023, 5(3): 1-38. doi: 10.3934/mine.2023058 |
[5] | Matteo Novaga, Marco Pozzetta . Connected surfaces with boundary minimizing the Willmore energy. Mathematics in Engineering, 2020, 2(3): 527-556. doi: 10.3934/mine.2020024 |
[6] | Anoumou Attiogbe, Mouhamed Moustapha Fall, El Hadji Abdoulaye Thiam . Nonlocal diffusion of smooth sets. Mathematics in Engineering, 2022, 4(2): 1-22. doi: 10.3934/mine.2022009 |
[7] | Jason Howell, Katelynn Huneycutt, Justin T. Webster, Spencer Wilder . A thorough look at the (in)stability of piston-theoretic beams. Mathematics in Engineering, 2019, 1(3): 614-647. doi: 10.3934/mine.2019.3.614 |
[8] | Alberto Bressan, Yucong Huang . Globally optimal departure rates for several groups of drivers. Mathematics in Engineering, 2019, 1(3): 583-613. doi: 10.3934/mine.2019.3.583 |
[9] | Giuseppe Procopio, Massimiliano Giona . Bitensorial formulation of the singularity method for Stokes flows. Mathematics in Engineering, 2023, 5(2): 1-34. doi: 10.3934/mine.2023046 |
[10] | Miyuki Koiso . Stable anisotropic capillary hypersurfaces in a wedge. Mathematics in Engineering, 2023, 5(2): 1-22. doi: 10.3934/mine.2023029 |
The integration of robotics education with science, technology, engineering, and mathematics (STEM) education has a great potential in future education. In recent years, numerous countries have hosted robotic competitions. This study uses a mixed research method to explore the coaches' views on student participation in the World Robot Olympiad (WRO) by incorporating the questionnaire surveys and interviews conducted at the 2019 WRO finals in Hungary. By quantitative and qualitative analyses, coaches generally agreed that participation in the WRO improved students' STEM learning skills and cultivated their patience and resilience in handling challenging tasks.
Geometric evolutions are a fascinating topic naturally arising from the study of dynamical models in physics and material sciences. Concrete examples are, for instance, the analysis of the behavior in time of the interfaces surfaces in phase changes of materials or in the flows of immiscible fluids. From the mathematical point of view, they describe the motion of geometric objects or structures, usually driven by systems of partial differential equations.
In this work we rethink, expand the details and present in a unified treatment the results of E. Acerbi, N. Fusco, V. Julin and M. Morini [1,2] about two of the most recent of such geometric motions, namely, the modified Mullins–Sekerka flow and the surface diffusion flow.
Both flows deal with an evolution in time of smooth subsets Et of an open set Ω⊆Rn, with d(Et,∂Ω)>0, for every t in a time interval [0,T), such that their boundaries ∂Et, which are smooth hypersurfaces, move with some "outer" normal velocity Vt that, in the first case, is obtained as solution of the following "mixed" system
{Vt=[∂νtwt]on ∂EtΔwt=0in Ω∖∂Etwt=Ht+4γvton ∂Et−Δvt=uEt−−∫ΩuEtdxin Ω(distributionally) | (mMSF) |
where γ is a nonnegative parameter, v,w:[0,T)ׯΩ→R are continuous functions such that, setting wt=w(t,⋅) and vt=v(t,⋅), the functions vt and wt are smooth in Ω∖∂Et, for every t∈[0,T); the functions νt,Ht are the "outer" normal and the relative mean curvature of ∂Et and uEt=2χEt−1; finally, the velocity of the motion is given by [∂νtwt] which denotes ∂νtw+t−∂νtw−t, that, is the "jump" of the normal derivative of wt on ∂Et, where w+t and w−t are the restrictions of wt to Ω∖¯Et and Et, respectively.
The resulting motion, called modified Mullins–Sekerka flow [46] (see also [11,33] and [22] for a very clear and nice introduction to such flow), arises as a singular limit of a nonlocal version of the Cahn–Hilliard equation [4,41,50], to describe phase separation in diblock copolymer melts (see also [49]). It has been also called Hele–Shaw model [7], or Hele–Shaw model with surface tension [19,20,21]. We mention that the adjective "modified" comes from the introduction of the parameter γ>0 in the system (mMSF), while choosing γ=0 we have the original flow proposed by Mullins and Sekerka in [46].
In the second case, we will say that a flow of sets Et as above, is a solution of the surface diffusion flow if the normal velocity is pointwise given by
Vt=ΔtHton ∂Et, | (SDF) |
where Δt is the Laplacian of the hypersurface ∂Et, for all t∈[0,T). Such flow was first proposed by Mullins in [45] to study thermal grooving in material sciences (see also [17] for a nice presentation), in particular, in the physically relevant case of three–dimensional space, it describes the evolution of interfaces between solid phases of a system, which are studied in a variety of physical settings including phase transitions, epitaxial deposition and grain growth (see for instance [34] and the references therein).
Notice that, while in this latter case, the velocity flow is immediately well defined, the system (mMSF) is clearly undetermined as it is, since the behavior of the functions wt and vt is not prescribed on the boundary of Ω (which is also possibly not bounded). By simplicity, we will consider flows in the whole Euclidean space and we assume that all the functions and sets involved are periodic with respect to the standard lattice Zn of Rn. It is then clear that this is equivalent to "ambient" the problem in the n–dimensional "flat" torus Tn=Rn/Zn, hence in the sequel we will assume Ω=Tn, modifying the definitions above accordingly. Another possibility would be asking that Ω⊆Rn is bounded, the moving sets do not "touch" the boundary of Ω and that all the functions wt and vt are subject to homogeneous (zero) Neumann boundary conditions on ∂Ω (see Subsection 4.3).
A very important property of these geometric flows is that both are the gradient flow of a functional, which clearly gives a natural "energy", decreasing in time during the evolution (the velocity Vt is minus the gradient, that is, the Euler–Lagrange equation of a functional).
Precisely, in any dimension n∈N, the modified Mullins–Sekerka flow is the H−1/2–gradient flow of the following nonlocal Area functional
J(E)=A(∂E)+γ∫Tn∫TnG(x,y)uE(x)uE(y)dxdy, | (1.1) |
under the constraint that the volume Vol(E)=Ln(E) is fixed, where (here and in the whole paper),
A(∂E)=∫∂Edμ |
is the classical Area functional that gives the area of the (n−1)–dimensional smooth boundary of any sets E (μ is the "canonical" measure associated to the Riemannian metric on ∂E induced by metric of Tn coming from the scalar product of Rn, which coincides with the n–dimensional Hausdorff measure Hn) and G is the Green function of Tn (see [41], for details).
Similarly, the surface diffusion flow can be regarded as the H−1–gradient flow of the Area functional A with fixed volume.
Then, it clearly follows that, in both cases, the volume of the evolving sets Vol(Et) is constant in time, while neither convexity (see [16] and [36]) is maintained, nor there holds the so–called "comparison property" asserting that if two initial sets are one contained into the other, they stay so during the two flows. This is due to the lack of the maximum principle for parabolic equations or systems of order larger than two. We remind that such properties are shared by the more famous mean curvature flow, which is also a gradient flow of the Area functional (without the constraint on the volume), but with respect to the L2–norm (see [42], for instance).
Parametrizing the moving smooth surfaces ∂Et by some maps (embeddings) ψt:M→Tn such that ψt(M)=∂Et, where M is a fixed smooth, compact (n−1)–dimensional differentiable manifold and νt is the outer unit normal vector to ∂Et as above, the evolution laws (mMSF) and (SDF) can be respectively expressed as
∂∂tψt=Vtνt=[∂νtwt]νt, |
and
∂∂tψt=(ΔtHt)νt. |
Due to the parabolic nature (not actually so explicit in the first case) of these systems of PDEs, it is known that for every smooth initial set E0 in Tn, with boundary described by ψ0:M→Tn, both flows with such initial data exist unique and are smooth in some positive time interval [0,T). Indeed, such short time existence and uniqueness results were proved by Escher and Simonett [19,20,21] and independently by Chen, Hong and Yi [8] for the modified Mullins–Sekerka flow and by Escher, Mayer and Simonett in [17] for the surface diffusion flow of a smooth compact hypersurface in domains of the Euclidean space of any dimension. With minor modifications, their proof can be adapted to get the same conclusion also for smooth initial hypersurfaces of Tn.
The aim of this work is to show that, in dimensions two and three, for initial data sufficiently "close" to a smooth strictly stable critical set E for the relative "energy" functional (the nonlocal or the usual Area functional) under a volume constraint, the flows exist for all positive times and asymptotically converge in some sense to a "translate" of E.
The notions of criticality and stability are as usual defined in terms of first and second variations of J and A. We say that a smooth subset E⊆Tn is critical for J (or for A, simply choosing γ=0 in formula (1.1)) if for any smooth one–parameter family of diffeomorphisms Φt:Tn→Tn, such that Vol(Φt(E))=Vol(E), for t∈(−ε,ε) and Φ0=Id (Et=Φt(E) will be called volume–preserving variation of E), we have
ddtJ(Φt(E))|t=0=0. |
We will see that this condition is equivalent to the existence of a constant λ∈R such that
H+4γvE=λon ∂E, |
where H is the mean curvature of ∂E and vE is the potential defined as
vE(x)=∫TnG(x,y)uE(y)dy, |
with G the Green function of the torus Tn and uE=χE−χTn∖E.
The second variation of J at a critical set E, leading to the central notion of stability, is more involved and, differently by the original papers, we will compute it with the tools and methods of differential/Riemannian geometry (like the first variation). We will see that at a critical set E, the second variation of J (the second derivative at t=0 of J(Et)) along a volume–preserving variation Et=Φt(E) only depends on the normal component φ on ∂E of the infinitesimal generator field X=∂Φt∂t|t=0 of the variation. The volume constraint on the admissible deformations of E implies that the functions φ must have zero integral on ∂E, hence it is natural to define a quadratic form ΠE on such space of functions which is related to the second variation of J by the following equality
ΠE(φ)=d2dt2J(Φt(E))|t=0 | (1.2) |
where Et=Φt(E) is a volume–preserving variation of E such that
⟨νE|∂Φt∂t|t=0⟩=φ |
on ∂E, with νE the outer unit normal vector of ∂E.
Because of the obvious translation invariance of the functional J, it is easy to see (by means of the formula (1.2)) that the form ΠE vanishes on the finite dimensional vector space given by the functions ψ=⟨νE|η⟩, for every vector η∈Rn. We underline that the presence of such "natural" degenerate subspace of the quadratic form ΠE (or, equivalently, the translation invariance of J) is the main reason of several technical difficulties.
We then say that a smooth critical set E⊆Tn is strictly stable if
ΠE(φ)>0 |
for all non–zero functions φ:∂E→R, with zero integral and L2–orthogonal to every function ψ=⟨νE|η⟩.
Then, the heuristic idea is that in a region around a strictly stable critical set E, we have a "potential well" for the "energy" J (and the set E is a local minimum) and, defining a suitable notion of "closedness", if one set starts close enough to E, during its evolution by (minus) the gradient of such energy, it cannot "escape" the well and asymptotically converges to a set of (local) minimal energy, which must be a translate of E. That is, the strict stability of E implies a "dynamical" stability in a neighborhood.
At the moment, this conclusion, that we state precisely below, can be shown only in dimension at most three, because of missing estimates in higher dimensions (see the discussion at the beginning of Section 4). When n>3 this and several other questions on these flows remain open. Anyway, this is sufficient for the application to some physically relevant models, since the evolution laws (mMSF) and (SDF) describe, respectively, pattern–forming processes such as the solidification in pure liquids and the evolution of interfaces between solid phases of a system, driven by surface diffusion of atoms under the action of a chemical potential (see for instance [34] and the references therein). In this paper, we will only deal with the three–dimensional case, but we underline that all the results and arguments hold, without relevant modifications, also in the two–dimensional situation of T2=R2/Z2, where the moving boundaries of the sets are curves.
Moreover, we mention here that all the results also hold in a bounded open subset Ω of R2 or R3, for moving sets which do not "touch" the boundary of Ω, imposing that the functions wt and vt in the definition of the modified Mullins–Sekerka flow satisfy a zero Neumann boundary condition (as we mentioned above), instead than choosing the "toric ambient" (see Subsection 4.3 for more details).
Theorem (Theorem 4.6 and Remark 4.7). Let E⊆T3 be a smooth strictly stable critical set for the nonlocal Area functional under a volume constraint and Nε a suitable tubular neighborhood of ∂E. For every α∈(0,1/2) there exists M>0 such that, if E0 is a smooth set satisfying
● Vol(E0)=Vol(E),
● Vol(E0△E)≤M,
● the boundary of E0 is contained in Nε and can be represented as
∂E0={y+ψE0(y)νE(y):y∈∂E}, |
for some function ψE0:∂E→R such that ‖ψE0‖C1,α(∂E)≤M,
● there holds
∫T3|∇wE0|2dx≤M, |
where w0=wE0 is the function relative to E0, as in system (mMSF),
then, there exists a unique smooth solution Et of the modified Mullins–Sekerka flow (with parameter γ≥0) starting from E0, which is defined for all t≥0. Moreover, Et→E+η exponentially fast in Ck as t→+∞, for every k∈N, for some η∈R3, with the meaning that the functions ψη,t:∂E+η→R representing ∂Et as "normal graphs" on ∂E+η, that is,
∂Et={y+ψη,t(y)νE+η(y):y∈∂E+η}, |
satisfy for every k∈N, the estimates
‖ψη,t‖Ck(∂E+η)≤Cke−βkt |
for every t∈[0,+∞), for some positive constants Ck and βk.
Theorem (Theorem 4.19 and Remark 4.20). Let E⊆T3 be a strictly stable critical set for the Area functional under a volume constraint and let Nε be a tubular neighborhood of ∂E. For every α∈(0,1/2) there exists M>0 such that, if E0 is a smooth set satisfying
● Vol(E0)=Vol(E),
● Vol(E0△E)≤M,
● the boundary of E0 is contained in Nε and can be represented as
∂E0={y+ψE0(y)νE(y):y∈∂E}, |
for some function ψE0:∂F→R such that ‖ψE0‖C1,α(∂E)≤M,
● there holds
∫∂E0|∇H0|2dμ0≤M, |
then there exists a unique smooth solution Et of the surface diffusion flow starting from E0, which is defined for all t≥0. Moreover, Et→E+η exponentially fast in Ck as t→+∞, for some η∈R3, with the same meaning as above.
We remark that the line of the proof in [1] that we are going to present, is based on suitable energy identities and compactness arguments to establish these global existence and exponential stability results. This was actually a completely new approach to manage the translation invariance of the functional J, in previous literature dealt with by means of semigroup techniques.
Summarizing, the work is organized as follows: in Section 2 we study the nonlocal Area functional (constrained or not) and we compute its first and second variation, then we discuss the notions of criticality, stability and local minimality of a set and their mutual relations, in this context. In Section 3 we introduce the modified Mullins–Sekerka and the surface diffusion flow and we analyze their basic properties. Section 4 is devoted to show the two main theorems above, while finally in Section 5, we discuss the classification of the stable and strictly stable critical sets (to whom then the two stability results apply).
We start by introducing the nonlocal Area functional and its basic properties.
In the following we denote by Tn the n–dimensional flat torus of unit volume which is defined as the Riemannian quotient of Rn with respect to the equivalence relation x∼y⟺x−y∈Zn, with Zn the standard integer lattice of Rn. Then, the functional space Wk,p(Tn), with k∈N and p≥1, can be identified with the subspace of Wk,ploc(Rn) of the functions that are 1–periodic with respect to all coordinate directions. A set E⊆Tn is of class Ck (or smooth) if its "1–periodic extension" to Rn is of class Ck (or smooth, ) which means that its boundary is locally a graph of a function of class Ck around every point. We will denote with Vol(E)=Ln(E) the volume of E⊆Tn.
Given a smooth set E⊆Tn, we consider the associated potential
vE(x)=∫TnG(x,y)uE(y)dy, | (2.1) |
where G is the Green function (of the Laplacian) of the torus Tn and uE=χE−χTn∖E. More precisely, G is the (distributional) solution of
−ΔxG(x,y)=δy−1in Tnwith∫TnG(x,y)dx=0, | (2.2) |
for every fixed y∈Tn, where δy denotes the Dirac delta measure at y∈Tn (the n–torus Tn has unit volume).
By the properties of the Green function, vE is then the unique solution of
{−ΔvE=uE−minTn(distributionally)∫TnvE(x)dx=0 | (2.3) |
where m=Vol(E)−Vol(Tn∖E)=2Vol(E)−1.
Remark 2.1. By standard elliptic regularity arguments (see [29], for instance), vE∈W2,p(Tn) for all p∈[1,+∞). More precisely, there exists a constant C=C(n,p) such that ‖vE‖W2,p(Tn)≤C, for all E⊆Tn such that Vol(E)−Vol(Tn∖E)=m.
Then, we define the following nonlocal Area functional (see [40,47,64], for instance).
Definition 2.2 (Nonlocal Area functional). Given γ≥0, the nonlocal Area functional J is defined as
J(E)=A(∂E)+γ∫Tn|∇vE(x)|2dx, | (2.4) |
for every smooth set E⊆Tn, where the function vE:Tn→R is given by formulas (2.1)–(2.3) and
A(∂E)=∫∂Edμ |
is the Area functional, where μ is the "canonical" measure associated to the Riemannian metric on ∂E induced by the metric tensor of Tn, coming from the scalar product of Rn (it is easy to see that μ coincides with the (n−1)–dimensional Hausdorff measure restricted to ∂E).
Since the nonlocal Area functional is defined adding to the Area functional a constant γ≥0 times a nonlocal term, all the results of this section will also hold for the Area functional, taking γ=0.
Multiplying by vE both sides of the first equation in system (2.3) and integrating by parts (and using also the second equation), we obtain
∫Tn|∇vE(x)|2dx=−∫TnvE(x)ΔvE(x)dx=∫TnvE(x)(uE(x)−m)dx=∫TnvE(x)uE(x)dx=∫Tn∫TnG(x,y)uE(x)uE(y)dxdy, | (2.5) |
hence, the functional J can be also written in the useful form
J(E)=A(∂E)+γ∫Tn∫TnG(x,y)uE(x)uE(y)dxdy. |
We start by computing the first variation of the functional J.
Definition 2.3. Let E⊆Tn be a smooth set. Given a smooth map Φ:(−ε,ε)×Tn→Tn, for ε>0, such that Φt=Φ(t,⋅):Tn→Tn is a one–parameter family of diffeomorphism with Φ0=Id, we say that Et=Φt(E) is the variation of E associated to Φ (or to Φt). If moreover there holds Vol(Et)=Vol(E) for every t∈(−ε,ε), we call Et a volume–preserving variation of E.
The vector field X∈C∞(Tn;Rn) defined as X=∂Φt∂t|t=0, is called the infinitesimal generator of the variation Et.
Remark 2.4. As we are going to consider only smooth sets E, it is easy to see that this definition of variation is equivalent to have a family of diffeomorphisms Φt of E only, indeed these latter can always be extended to the whole Tn. Moreover, as the relevant objects are actually the boundaries of the sets E and in view of the sequel, we could even consider only smooth "deformations" of ∂E. We chose the above definition since it is easier and more convenient for the computations that are following.
Definition 2.5. Given a variation Et of E, coming from the one–parameter family of diffeomorphism Φt, the first variation of J at E with respect to Φt is given by
ddtJ(Et)|t=0. |
We say that E is a critical set for J, if all the first variations relative to variations Et of E are zero.
We say that E is a critical set for J under a volume constraint, if all the first variations relative to volume–preserving variations Et of E are zero.
It is clear that if E is a minimum for J (under a volume constraint), then it is a critical set for J (under a volume constraint). We are now going to compute the first variation of J and see that it depends only on the restriction to ∂E of the infinitesimal generator X of the variation Et of E.
We briefly recall some "geometric" notations and results about the (Riemannian) geometry of the hypersurfaces in Rn, referring to [26,42,51] for instance.
In the whole work, we will adopt the convention of summing over the repeated indices.
Given any smooth immersion ψ:M→Tn of the smooth, (n−1)–dimensional, compact manifold M, representing a hypersurface ψ(M) of Tn, considering local coordinates around any p∈M, we have local bases of the tangent space TpM, which can be identified with the (n−1)–dimensional hyperplane dψp(TpM) of Rn≈Tψ(p)Tn which is tangent to ψ(M) at ψ(p), and of the cotangent space T∗pM, respectively given by vectors {∂∂xi} and 1–forms {dxj}. So we denote the vectors on M by X=Xi∂∂xi and the 1–forms by ω=ωjdxj, where the indices refer to the chosen local coordinate chart of M. With the above identification, we have clearly ∂∂xi≈∂ψ∂xi, for every i∈{1,…,n−1}.
The manifold M gets in a natural way a metric tensor g, pull–back via the map ψ of the metric tensor of Tn, coming from the standard scalar product of Rn (as Tn≈Rn/Zn), hence, turning it into a Riemannian manifold (M,g). Then, the components of g in a local chart are
gij=⟨∂ψ∂xi|∂ψ∂xj⟩ |
and the "canonical" measure μ, induced on M by the metric g is then given by μ=√detgijLn−1, where Ln−1 is the standard Lebesgue measure on Rn−1.
Thus, supposing that M has a global coordinate chart, we can write the Area functional on the hypersurface ψ(M) in the following way,
A(ψ(M))=∫Mdμ=∫M√detgij(x)dx. | (2.6) |
When this is not the case (as it is usual), we need several local charts (Uk,φk) and a subordinated partitions of unity fk:M→[0,1] (that is, the compact support of fk:M→[0,1] is contained in the open set Uk⊆M, for every k∈I), then
A(ψ(M))=∫Mdμ=∑k∈I∫Mfkdμ=∑k∈I∫Ukfk(x)√detgkij(x)dx, |
where gkij are the coefficients of the metric g in the local chart (Uk,φk).
In order to work with coordinates, in the computations with integrals in this section we will assume that all the hypersurfaces have a global coordinate chart, by simplicity. All the results actually hold also in the general case by using partitions of unity as above.
The induced Levi–Civita covariant derivative on (M,g) of a vector field X and of a 1–form ω are respectively given by
∇jXi=∂Xi∂xj+ΓijkXk,∇jωi=∂ωi∂xj−Γkjiωk, |
where Γijk are the Christoffel symbols of the connection ∇, expressed by the formula
Γijk=12gil(∂∂xjgkl+∂∂xkgjl−∂∂xlgjk). |
Moreover, the gradient ∇f of a function, the divergence divX of a tangent vector field and the Laplacian Δf at a point p∈M, are defined respectively by
g(∇f(p),v)=dfp(v)∀v∈TpM, |
divX=tr∇X=∇iXi=∂Xi∂xi+ΓiikXk |
(in a local chart) and Δf=div∇f. We then recall that by the divergence theorem for compact manifolds (without boundary), there holds
∫MdivXdμ=0, | (2.7) |
for every tangent vector field X on M, which in particular implies
∫MΔfdμ=0, |
for every smooth function f:M→R.
Assuming that we have a globally defined unit normal vector field ν:M→Rn to φ(M) (this will hold in our situation where the hypersurfaces will be boundaries of smooth sets E⊆Tn, hence we will always consider ν to be the outer unit normal vector at every point of ∂E), we define the second fundamental form B which is a symmetric bilinear form given, in a local charts, by its components
hij=−⟨∂2ψ∂xi∂xj|ν⟩ |
and whose trace is the mean curvature H=gijhij of the hypersurface (with these choices, the standard sphere of Rn has positive mean curvature).
The symmetry properties of the covariant derivative of B are given by the Codazzi–Mainardi equations
∇ihjk=∇jhik=∇khij. | (2.8) |
In the sequel, the following Gauss–Weingarten relations will be fundamental,
∂2ψ∂xi∂xj=Γkij∂ψ∂xk−hijν∂ν∂xj=hjlgls∂ψ∂xs, | (2.9) |
which imply
Δψ=gij(∂2ψ∂xi∂xj−Γkij∂ψ∂xk)=−gijhijν=−Hν. | (2.10) |
Moreover, we have the formula
Δν=∇H−|B|2ν, | (2.11) |
indeed, computing in normal coordinates at a point p∈M,
Δν=gij(∂2ν∂xi∂xj−Γkij∂ν∂xk)=gij∂∂xi(hjlgls∂ψ∂xs)=gij∇ihjlgls∂ψ∂xs+gijhjlgls∂2ψ∂xi∂xs=gij∇lhijgls∂ψ∂xs−gijhjlglshisν=∇H−|B|2ν, |
since all Γkij and ∂∂xigjk are zero at p∈M in such coordinates and we used Codazzi–Mainardi equations (2.8).
In the following, when it is clear by the context, we will write ∇, div and Δ for both the Riemannian operators on a hypersurface and the standard operators of Tn≈Rn/Zn, but these latter will be instead denoted by ∇Tn, divTn and ΔTn when they will be computed at a point of a hypersurface, in order to avoid any possibility of misunderstanding.
Theorem 2.6 (First variation of the functional J). Let E⊆Tn a smooth set and Φ:(−ε,ε)×Tn→Tn a smooth map giving a variation Et=Φt(E) with infinitesimal generator X∈C∞(Tn;Rn). Then,
ddtJ(Et)|t=0=∫∂E(H+4γvE)⟨X|νE⟩dμ | (2.12) |
where νE is the outer unit normal vector and H the mean curvature of the boundary ∂E (as defined above, relative to νE), while the function vE:Tn→R is the potential associated to E, defined by formulas (2.1)–(2.3).
In particular, the first variation of the functional J depends only on the normal component of the restriction of the infinitesimal generator X to ∂E.
Clearly, when γ=0 we get the well known first variation of the Area functional at a smooth set E,
ddtA(∂Et)|t=0=∫∂EH⟨X|νE⟩dμ. |
Proof. We start by computing the derivative of the Area functional term of J. We let ψt:∂E→Tn be the embedding given by
ψt(x)=Φ(t,x), |
for x∈∂E and t∈(−ε,ε), then ψt(∂E)=∂Et and ∂tψt|t=0=X at every point of ∂E, moreover ψ0 is simply the inclusion map of ∂E in Tn.
Denoting by gij=gij(t) the induced metrics (via ψt, as above) on the smooth hypersurfaces ∂Et and setting ψ=ψ0, in a local chart we have
∂∂tgij|t=0=∂∂t⟨∂ψt∂xi|∂ψt∂xj⟩|t=0=⟨∂X∂xi|∂ψ∂xj⟩+⟨∂X∂xj|∂ψ∂xi⟩=∂∂xi⟨X|∂ψ∂xj⟩+∂∂xj⟨X|∂ψ∂xi⟩−2⟨X|∂2ψ∂xi∂xj⟩=∂∂xi⟨Xτ|∂ψ∂xj⟩+∂∂xj⟨Xτ|∂ψ∂xi⟩−2Γkij⟨Xτ|∂ψ∂xk⟩+2hij⟨X|νE⟩, |
where we used the Gauss–Weingarten relations (2.9) in the last step and we denoted with Xτ=X−⟨X|νE⟩νE the "tangential part" of the vector field X along the hypersurface ∂E (seeing Tx∂E as a hyperplane of Rn≈TxTn).
Letting ω be the 1–form defined by ω(Y)=g(Xτ,Y), this formula can be rewritten as
∂∂tgij|t=0=∂ωj∂xi+∂ωi∂xj−2Γkijωk+2hij⟨X|νE⟩=∇iωj+∇jωi+2hij⟨X|νE⟩. | (2.13) |
Hence, by the formula
ddtdetA(t)=detA(t)tr[A−1(t)∘A′(t)], | (2.14) |
holding for any n×n squared matrix A(t) dependent on t, we get
∂∂t√detgij|t=0=√detgijgij∂∂tgij|t=02=√detgijgij(∇iωj+∇jωi+2hij⟨X|νE⟩)2=√detgij(divXτ+H⟨X|νE⟩), | (2.15) |
where the divergence is the (Riemannian) one relative to the hypersurface ∂E. Then, we conclude (recalling the discussion after formula (2.6))
∂∂tA(∂Et)|t=0=∂∂tA(ψt(∂E))|t=0=∂∂t∫∂Edμt|t=0=∂∂t∫∂E√detgijdx|t=0=∫∂E∂∂t√detgij|t=0dx=∫∂E(divXτ+H⟨X|νE⟩)√detgijdx=∫∂E(divXτ+H⟨X|νE⟩)dμ=∫∂EH⟨X|νE⟩dμ | (2.16) |
where in the last step we applied the divergence theorem, that is, formula (2.7), on ∂E.
In order to compute the derivative of the nonlocal term, we set
v(t,x)=vEt(x)=∫TnG(x,y)uEt(x)dy=∫EtG(x,y)dy−∫EctG(x,y)dy, |
where Ect=Tn∖Et. Then,
ddt(∫Tn|∇vEt(x)|2dx)|t=0=ddt(∫Tn|∇v(t,x)|2dx)|t=0=2∫Tn∇vE(x)∂∂t∇v(t,x)|t=0dx=2∫Tn(uE(x)−m)∂∂tv(t,x)|t=0dx, |
where in the last equality we used the fact that −ΔvE=uE−m and we integrated by parts. Now, we note that
∂∂tv(t,x)=∂∂t(∫EtG(x,y)dy)−∂∂t(∫EctG(x,y)dy), | (2.17) |
and, by a change of variable,
∂∂t(∫EtG(x,y)dy)|t=0=∂∂t(∫EG(x,Φ(t,z))JΦ(t,z)dz)|t=0, | (2.18) |
where JΦ(t,⋅) is the Jacobian of Φ(t,⋅). Then, as JΦ(t,z)=det[dΦ(t,z)], using again formula (2.14), we have
∂∂tJΦ(t,z)|t=0=JΦ(t,z)tr[dΦ(t,z)−1∘∂∂tdΦ(t,z)]|t=0=JΦ(t,z)tr[dΦ(t,z)−1∘d∂∂tΦ(t,z)]|t=0=trdX(z)=divX(z), |
by the definition of X and being Φ(0,z)=z. Thus, carrying the time derivative inside the integral in Eq (2.18), we obtain
∂∂t(∫EtG(x,y)dy)|t=0=∫E(⟨∇yG(x,y)|X(y)⟩+G(x,y)divX(y))dy=∫Edivy(G(x,y)X(y))dy=∫∂EG(x,y)⟨X(y)|νE(y)⟩dμ(y). |
By a very analogous computation we get
−∂∂t(∫EctG(x,y)dy)|t=0=∫∂EG(x,y)⟨X(y)|νE(y)⟩dμ(y), | (2.19) |
then, using equalities (2.1) and (2.2), we conclude
ddt∫Tn|∇vEt(x)|2dx|t=0=4∫Tn(uE(x)−m)(∫∂EG(x,y)⟨X(y)|νE(y)⟩dμ(y))dx=4∫∂E(∫TnG(x,y)(uE(x)−m)dx)⟨X(y)|νE(y)⟩dμ(y)=4∫∂EvE(y)⟨X(y)|νE(y)⟩dμ(y). | (2.20) |
Combining formulas (2.16) and (2.20), we finally obtain formula (2.12).
Given a smooth set E and any vector field X∈C∞(Tn;Rn), considering the associated smooth flow Φ:(−ε,ε)×Tn→Tn, defined by the system
{∂Φ∂t(t,x)=X(Φ(t,x)),Φ(0,x)=x | (2.21) |
for every x∈Tn and t∈(−ε,ε), for some ε>0, we have a variation Et=Φt(E) with infinitesimal generator X. We call this variation the special variation associated to X. Moreover, given any smooth vector field ¯X∈C∞(∂E;Rn), it can be extended easily to a smooth vector field X∈C∞(Tn;Rn) with X|∂E=¯X.
Hence, if E is a critical set for J there holds
∫∂E(H+4γvE)⟨X|νE⟩dμ=0, |
for every X∈C∞(Tn;Rn). Choosing a smooth vector field X∈C∞(Tn;Rn) with X|∂E=(H+4γvE)νE, we then obtain the following corollary.
Corollary 2.7. A smooth set E⊆Tn is a critical set for J if and only if the function H+4γvE is zero on ∂E.When γ=0, we recover the classical condition H=0 for a minimal surface in Rn.
It is less easy to characterize the infinitesimal generators of the volume–preserving variations of E, in order to find an analogous criticality condition on a set E, for the functional J under a volume constraint.
Given Φ:(−ε,ε)×Tn→Tn such that Vol(Φt(E))=Vol(Et)=Vol(E) for all t∈(−ε,ε), we let Xt∈C∞(Tn;Rn) be the family of the vector fields (well) defined by the formula
Xt(Φ(t,z))=∂Φ∂t(t,z), |
for every t∈(−ε,ε) and z∈Tn, hence, if t=0, the vector field X=X0 is the infinitesimal generator of the volume–preserving variation Et. Then, by changing variables, we have
0=ddtVol(Et)=ddt∫Etdx=ddt∫EJΦ(t,z)dz=∫E∂∂tJΦ(t,z)dz. | (2.22) |
As JΦ(t,z)=det[dΦ(t,z)], by means of formula (2.14), we obtain
∂∂tJΦ(t,z)=JΦ(t,z)tr[dΦ(t,z)−1∘dXt(Φ(t,z))∘dΦ(t,z)], |
since, by the definition of Xt above,
∂∂tdΦ(t,z)=d∂Φ∂t(t,z)=d[Xt(Φ(t,z))]=dXt(Φ(t,z))∘dΦ(t,z). |
Being the trace of a matrix invariant by conjugation, we conclude
∂∂tJΦ(t,z)=JΦ(t,z)tr[dXt(Φ(t,z))]=JΦ(t,z)divXt(Φ(t,z)), |
hence, by equality (2.22) and the divergence theorem (in Tn), it follows
0=∫EdivXt(Φ(t,z))JΦ(t,z)dz=∫EtdivXt(x)dx=∫∂E⟨Xt∘Φt|νEt⟩dμt, | (2.23) |
where νEt is the outer unit normal vector and μt the canonical Riemannian measure of the smooth hypersurface ∂Et, given by the embedding ψt=Φt:∂E→Tn. Thus, letting t=0,
ddtVol(Et)|t=0=∫∂E⟨X|νE⟩dμ=0 | (2.24) |
and we conclude that if X∈C∞(Tn;Rn) is the infinitesimal generator of a volume–preserving variation for E, its normal component φ=⟨X|νE⟩ on ∂E has zero integral (with respect to the measure μ).
Conversely, we have the following lemma whose proof is postponed after Lemma 2.32, since the arguments in the two proofs are very similar.
Lemma 2.8. Let φ:∂E→R a smooth function with zero integral with respect to the measure μ on ∂E. Then, there exists a smooth vector field X∈C∞(Tn;Rn) such that φ=⟨X|νE⟩, divX=0 in a neighborhood of ∂E and the flow Φ defined by system (2.21) having X as infinitesimal generator, gives a volume–preserving variation Et=Φt(E) of E.
Hence, with this characterization of the infinitesimal generators of the volume–preserving variations for E, by Theorem 2.6 we have that E is a critical set for the functional J under a volume constraint if and only if
∫∂E(H+4γvE)⟨X|νE⟩dμ=0, |
for every X∈C∞(Tn;Rn) such that ⟨X|νE⟩ has zero integral on ∂E. By Lemma 2.8, this is similarly to say that
∫∂E(H+4γvE)φdμ=0, |
for all φ∈C∞(∂E) such that ∫∂Eφdμ=0, which is equivalent to the existence of a constant λ∈R such that
H+4γvE=λon∂E. |
Remark 2.9. The parameter λ may be clearly interpreted as a Lagrange multiplier associated with the volume constraint for J.
Proposition 2.10. A smooth set E⊆Tn is a critical set for J under a volume constraint if and only if the function H+4γvE is constant on ∂E.When γ=0, we recover the classical constant mean curvature condition for hypersurfaces in Rn.
Now we deal with the second variation of the functional J.
Definition 2.11. Given a variation Et of E, coming from the one–parameter family of diffeomorphism Φt, the second variation of J at E with respect to Φt is given by
d2dt2J(Et)|t=0. |
In the following proposition we compute the second variation of the Area functional. Then, we do the same for the nonlocal term of J and we conclude with the second variation of the functional J.
Proposition 2.12 (Second variation of A). Let E⊆Tn a smooth set and Φ:(−ε,ε)×Tn→Tn a smooth map giving a variation Et=Φt(E) with infinitesimal generator X∈C∞(Tn;Rn). Then,
d2dt2A(∂Et)|t=0=∫∂E(|∇⟨X|νE⟩|2−⟨X|νE⟩2|B|2)dμ+∫∂EH(H⟨X|νE⟩2+⟨Z|νE⟩−2⟨Xτ|∇⟨X|νE⟩⟩+B(Xτ,Xτ))dμ, |
where Xτ=X−⟨X|νE⟩νE is the tangential part of X on ∂E, B and H are respectively the second fundamental form and the mean curvature of ∂E, and
Z=∂2Φ∂t2(0,⋅)=∂∂t[Xt(Φ(t,⋅))]|t=0=∂Xt∂t|t=0+dX(X), | (2.25) |
where, for every t∈(−ε,ε), the vector field Xt∈C∞(Tn;Rn) is defined by the formula
Xt(Φ(t,z))=∂Φ∂t(t,z), |
for every z∈Tn, hence, X0=X.
Proof. We let ψt=Φ(t,⋅)|∂E. By arguing as in the first part of the proof of Theorem 2.6 (without taking t=0), we have
ddtA(∂Et)=∫∂EHt⟨Xt∘Φt|νEt⟩dμt, |
where Ht is the mean curvature of ∂Et. Consequently, we have
d2dt2A(∂Et)|t=0=ddt∫∂EHt⟨Xt∘Φt|νEt⟩√detgijdx|t=0 |
where gij=gij(t).
In order to simplify the notation in the following computations, we drop the subscripts, that is, we let H(t,⋅)=Ht, ν(t,⋅)=νEt, φ(t,⋅)=⟨Xt∘Φt|νEt⟩, ψ(t,⋅)=ψt and X(t,⋅)=Xt∘Φt (by a little abuse of notation, since X is already the infinitesimal generator of the variation).
We then need to compute the derivatives
∂H∂t|t=0 and ∂∂t⟨X|ν⟩|t=0 | (2.26) |
since we already know, by formula (2.15), that
∂∂t√detgij|t=0=(divXτ+Hφ)√detgij|t=0, |
hence, this derivative gives the following contribution to the second variation,
∫∂E(φHdivXτ+φ2H2)dμ. |
Then, we compute (recalling formula (2.25))
∂⟨X|ν⟩∂t|t=0=⟨∂X∂t|ν⟩|t=0+⟨X|∂ν∂t⟩|t=0=⟨Z|ν⟩+⟨X|∂ν∂t⟩|t=0 |
and using the fact that ∂ν∂t|t=0 is tangent to ∂E, in a local coordinate chart we obtain
⟨X|∂ν∂t⟩|t=0=Xpτ⟨∂ψ∂xp|∂ν∂t⟩|t=0, |
where in the last inequality we used the notation Xτ=Xpτ∂ψ∂xp. Notice that, ⟨∂ψ∂xp|ν⟩=0 for every p∈{1,…,n−1} and t∈(−ε,ε), hence, using the Gauss–Weingarten relations (2.9),
0=∂∂t⟨∂ψ∂xp|ν⟩|t=0=⟨∂X∂xp|ν⟩+⟨∂ψ∂xp|∂ν∂t⟩|t=0=∂∂xp⟨X|ν⟩−⟨X|∂ν∂xp⟩+⟨∂ψ∂xp|∂ν∂t⟩|t=0=∂φ∂xp−⟨Xτ|∂ν∂xp⟩+⟨∂ψ∂xp|∂ν∂t⟩|t=0=∂φ∂xp−Xqτ⟨∂ψ∂xq|∂ν∂xp⟩+⟨∂ψ∂xp|∂ν∂t⟩|t=0=∂φ∂xp−Xqτ⟨∂ψ∂xq|hplgli∂ψ∂xi⟩+⟨∂ψ∂xp|∂ν∂t⟩|t=0=∂φ∂xp−Xqτhplgligqi+⟨∂ψ∂xp|∂ν∂t⟩|t=0 |
and we can conclude that
⟨∂ψ∂xp|∂ν∂t⟩|t=0=−∂φ∂xp+Xqτhpq, | (2.27) |
where hpq are the components of the second fundamental form B of ∂E in the local chart. Thus, we obtain the following identity
∂∂t⟨X|ν⟩|t=0=⟨Z|ν⟩+Xpτ⟨∂ψ∂xp|∂ν∂t⟩|t=0=⟨Z|ν⟩−∂φ∂xpXpτ+XpτXqτhpq=⟨Z|ν⟩−⟨Xτ|∇⟨X|ν⟩⟩+B(Xτ,Xτ) | (2.28) |
and the relative contribution to the second variation is given by
∫∂EH(⟨Z|ν⟩−⟨Xτ|∇⟨X|ν⟩⟩+B(Xτ,Xτ))dμ. |
Now we conclude by computing the first derivative in (2.26). To this aim, we note that
H=−⟨∂2ψ∂xi∂xj|ν⟩gij |
hence, we need the following terms
∂gij∂t|t=0 | (2.29) |
⟨∂2ψ∂xi∂xj|∂ν∂t⟩|t=0 | (2.30) |
⟨∂∂t∂2ψ∂xi∂xj|ν⟩|t=0. | (2.31) |
We start with the term (2.29), recalling that
∂gij∂t|t=0=∇iωj+∇jωi+2hij⟨X|ν⟩ |
by Eq (2.13), where ω is the 1–form defined by ω(Y)=g(Xτ,Y).
Using the fact that gijgjk=0, we obtain
0=∂gij∂t|t=0gjk+gij∂gjk∂t|t=0=gjk(∇iωj+∇jωi+2hij⟨X|ν⟩)+gij∂gjk∂t|t=0 |
then,
∂gpk∂t|t=0=−gjpgik(∇iωj+∇jωi+2hij⟨X|ν⟩)=−∇pXkτ−∇kXpτ−2hpkφ. | (2.32) |
We then proceed with the computation of the term (2.30), by means of Eq (2.27),
⟨∂2ψ∂xi∂xj|∂ν∂t⟩|t=0=Γkij⟨∂ψ∂xk|∂ν∂t⟩|t=0=Γkij(−∂φ∂xk+Xqτhqk) |
and finally we compute the term (2.31),
⟨∂∂t∂2ψ∂xi∂xj|ν⟩=⟨∂2X∂xi∂xj|ν⟩|t=0=⟨∂2(φν)∂xi∂xj|ν⟩+⟨∂2Xτ∂xi∂xj|ν⟩. |
We have
⟨∂2(φν)∂xi∂xj|ν⟩=∂2φ∂xi∂xj+⟨∂2ν∂xi∂xj|ν⟩φ=∂2φ∂xi∂xj+⟨∂∂xi(hjlglp∂ψ∂xp)|ν⟩φ=∂2φ∂xi∂xj+hjlglp⟨∂2ψ∂xi∂xj|ν⟩φ=∂2φ∂xi∂xj+φhjlglphip |
and
⟨∂2Xτ∂xi∂xj|ν⟩=∂∂xi⟨∂Xτ∂xj|ν⟩−⟨∂Xτ∂xj|∂ν∂xi⟩=∂∂xi⟨∂∂xj(Xpτ∂ψ∂xp)|ν⟩−⟨∂Xτ∂xj|∂ν∂xi⟩=∂∂xi[Xpτ⟨∂2ψ∂xj∂xp|ν⟩]−⟨∂Xτ∂xj|∂ν∂xi⟩=−∂∂xi(Xpτhpj)−⟨∂Xτ∂xj|∂ν∂xi⟩=−∂∂xi(Xpτhpj)−⟨∂∂xj(Xpτ∂ψ∂xp)|∂ν∂xi⟩=−∂∂xi(Xpτhpj)−Xpτ⟨∂2ψ∂xj∂xp|∂ν∂xi⟩−∂Xpτ∂xj⟨∂ψ∂xp|∂ν∂xi⟩=−∂∂xi(Xpτhpj)−XpτΓkjp⟨∂ψ∂xk|∂ν∂xi⟩−∂Xpτ∂xj⟨∂ψ∂xp|∂ν∂xi⟩=−∂∂xi(Xpτhpj)−XpτΓkjphilglqgkq−∂Xp∂xjhilglqgpq=−∂∂xi(Xpτhpj)−XpτΓkjphik−∂Xk∂xjhik. |
Hence, we finally get
∂H∂t|t=0=−2hij∇iXjτ−2⟨X|ν⟩|B|2−gij∂2φ∂xi∂xj+gijΓkij∂φ∂xk+|B|2⟨X|ν⟩−gijΓkijhkqXqτ+gij∂∂xi(Xpτhpj)+hij∇iXjj=−|B|2⟨X|ν⟩−hij∇iXjτ−Δφ+gij[∂∂xi(Xpτhpj)−Γkij(Xpτhpk)]=−φ|B|2−Δφ−hij∇iXjτ+gij∇i(Xpτhpj)=−φ|B|2−Δφ−hij∇iXjτ+div(Xpτhpj)=−φ|B|2−Δφ+⟨Xτ|divB⟩=−φ|B|2−Δφ+⟨Xτ|∇H⟩, | (2.33) |
where in the last equality we used the Codazzi–Mainardi equations (see [42]). We conclude that the contribution of the first term in (2.26) is then
∫∂Eφ(−φ|B|2−Δφ+⟨Xτ|∇H⟩)dμ. |
Putting all these contributions together, we obtain the second variation of the Area functional,
d2dt2A(∂Et)|t=0=∫∂E[−φΔφ−φ2|B|2+φ⟨Xτ|∇H⟩+φHdivXτ+φ2H2+H(⟨Z|ν⟩−⟨Xτ|∇φ⟩+B(Xτ,Xτ))]dμ. |
Integrating by parts, we have
∫∂Eφ⟨Xτ|∇H⟩dμ=−∫∂E[H⟨Xτ|∇φ⟩+HφdivXτ]dμ |
and we can conclude
d2dt2A(∂Et)|t=0=∫∂E[|∇φ|2−φ2|B|2+φ2H2+H(⟨Z|ν⟩−2⟨Xτ|∇φ⟩+B(Xτ,Xτ))]dμ, |
which is the formula we wanted.
Proposition 2.13 (Second variation of the nonlocal term). Let E⊆Tn, Φ, Et, X, Xτ, Xt, H, B and Z as in the previous proposition. Then, setting
N(t)=∫Tn|∇vEt(x)|2dx, |
where vEt:Tn→R is the function defined by formulas (2.1)–(2.3) and ∂νEvE=⟨∇TnvE|νE⟩, the following formula holds
d2dt2N(t)|t=0=8∫∂E∫∂EG(x,y)⟨X(x)|νE(x)⟩⟨X(y)|νE(y)⟩dμ(x)dμ(y)+4∫∂E[vE(H⟨X|νE⟩2+⟨Z|νE⟩−2⟨Xτ|∇⟨X|νE⟩⟩+B(Xτ,Xτ))+∂νEvE⟨X|νE⟩2]dμ, | (2.34) |
giving the second variation of the nonlocal term of J.
Proof. By arguing as in the second part of the proof of Theorem 2.6 (equations (2.17)–(2.20)), we have
ddtN(t)=4∫∂EvEt⟨Xt∘Φt|νEt⟩dμt=4∫∂EvEt⟨Xt∘Φt|νEt⟩√detgijdx. |
Setting v(t,x)=vEt(x), vt=∂v∂t(0,⋅), vi=∂v∂xi(0,⋅) and adopting the same notation of the proof of the previous proposition, that is, we let H(t,⋅)=Ht, ν(t,⋅)=νEt and X(t,⋅)=Xt∘Φt, we have
d2dt2N(t)|t=0=4ddt∫∂Ev⟨X|ν⟩√detgijdx|t=0=4∫∂E[vt⟨X|ν⟩+viXi⟨X|ν⟩+v⟨X|ν⟩divXτ+vH⟨X|ν⟩2+v∂∂t⟨X|ν⟩|t=0]dμ=4∫∂E[vt⟨X|ν⟩+viXi⟨X|ν⟩+v⟨X|ν⟩divXτ+v(H⟨X|ν⟩2+⟨Z|ν⟩−⟨Xτ|∇⟨X|ν⟩⟩+B(Xτ,Xτ))]dμ, |
by formulas (2.15) and (2.28). Then, integrating by parts the divergence, we obtain
d2dt2N(t)|t=0=4∫∂E[vt⟨X|ν⟩+viXi⟨X|ν⟩−⟨∇v|Xτ⟩⟨X|ν⟩+v(H⟨X|ν⟩2+⟨Z|ν⟩−2⟨Xτ|∇⟨X|ν⟩⟩+B(Xτ,Xτ))]dμ=4∫∂E[vt⟨X|ν⟩+∂νv⟨X|ν⟩2+v(H⟨X|ν⟩2+⟨Z|ν⟩−2⟨Xτ|∇⟨X|ν⟩⟩+B(Xτ,Xτ))]dμ |
where ∂νv=⟨∇Tnv|ν⟩.
Now, by Eqs (2.17)–(2.19), there holds
vt(0,x)=2∫∂EG(x,y)⟨X(y)|ν(y)⟩dμ(y), | (2.35) |
hence, substituting this expression for vt in the equation above we have formula (2.34).
Putting together Propositions 2.12 and 2.13, we then obtain the second variation of the nonlocal Area functional J.
Theorem 2.14 (Second variation of the functional J). Let E⊆Tn a smooth set and Φ:(−ε,ε)×Tn→Tn a smooth map giving a variation Et with infinitesimal generator X∈C∞(Tn;Rn). Then,
d2dt2J(Et)|t=0=∫∂E(|∇⟨X|νE⟩|2−⟨X|νE⟩2|B|2)dμ+8γ∫∂E∫∂EG(x,y)⟨X|νE(x)⟩⟨X|νE(y)⟩dμ(x)dμ(y)+4γ∫∂E∂νEvE⟨X|νE⟩2dμ+R, | (2.36) |
with the "remainder term" R given by
R=∫∂E(H+4γvE)(H⟨X|ν⟩2+⟨Z|ν⟩−2⟨Xτ|∇⟨X|ν⟩⟩+B(Xτ,Xτ))dμ=∫∂E(H+4γvE)[⟨X|νE⟩divTnX−div(⟨X|νE⟩Xτ)+⟨∂Xt∂t|t=0|νE⟩]dμ |
where νE is the outer unit normal vector to ∂E, Xτ=X−⟨X|νE⟩νE is the tangential part of X on ∂E, vE:Tn→R is the function defined by formulas (2.1)–(2.3), ∂νEvE=⟨∇TnvE|νE⟩, B and H are respectively the second fundamental form and the mean curvature of ∂E, the vector field Xt∈C∞(Tn;Rn) is defined by the formulaXt(Φ(t,z))=∂Φ∂t(t,z) for every t∈(−ε,ε) and z∈Tn, and
Z=∂2Φ∂t2(0,⋅)=∂∂t[Xt(Φ(t,⋅))]|t=0=∂Xt∂t|t=0+dX(X). |
Proof. Formula (2.36) and the first equality for R follows simply adding (after multiplying the nonlinear term by γ) the expressions for d2dt2A(∂Et)|t=0 and d2dt2∫Tn|∇vEt|2dx|t=0 we found in Propositions 2.12 and 2.13.
If now we show that
H⟨X|νE⟩2+⟨Z|νE⟩−2⟨Xτ|∇⟨X|νE⟩⟩+B(Xτ,Xτ)=⟨X|νE⟩divTnX−div(⟨X|νE⟩Xτ)+⟨∂Xt∂t|t=0|νE⟩, | (2.37) |
we clearly obtain the second expression for R.
We note that, being every derivative of νE a tangent vector field,
⟨Xτ|∇⟨X|νE⟩⟩=⟨νE|dX(Xτ)⟩+⟨X|⟨Xτ|∇νE⟩⟩=⟨νE|dX(Xτ)⟩+⟨Xτ|⟨Xτ|∇νE⟩⟩=⟨νE|dX(Xτ)⟩+B(Xτ,Xτ), |
by the Gauss–Weingarten relations (2.9).
Therefore, since Z−∂Xt∂t|t=0=dX(X), we have
H⟨X|νE⟩2+⟨Z|νE⟩−2⟨Xτ|∇⟨X|νE⟩⟩+B(Xτ,Xτ)−⟨∂Xt∂t|t=0|ν⟩=H⟨X|νE⟩2+⟨νE|dX(X)⟩−⟨Xτ|∇⟨X|νE⟩⟩−⟨νE|dX(Xτ)⟩=H⟨X|νE⟩2+⟨νE|dX(⟨X|νE⟩νE)⟩−⟨Xτ|∇⟨X|νE⟩⟩=H⟨X|νE⟩2+⟨X|νE⟩⟨νE|dX(νE)⟩+⟨X|νE⟩divXτ−div(⟨X|νE⟩Xτ). | (2.38) |
Now we notice that, choosing an orthonormal basis e1,…,en−1,en=νE of Rn at a point p∈∂E and letting X=Xiei, we have
⟨ei|∇⊤Xi⟩=⟨ei|∇TnXi−⟨∇TnXi|νE⟩νE⟩=divTnX−⟨νE|dX(νE)⟩, |
where the symbol ∇⊤f denotes the projection on the tangent space to ∂E of the gradient ∇Tnf of a function, called tangential gradient of f and coincident with the gradient operator of ∂E applied to the restriction of f to the hypersurface, while ⟨ei|∇⊤Xi⟩ is called tangential divergence of X, usually denoted with div⊤X and coincident with the (Riemannian) divergence of ∂E if X is a tangent vector field, as we will see below (see [63]). Moreover, if we choose a local parametrization of ∂E such that ∂ψ∂xi(p)=ei, for i∈{1,…,n−1}, we have eji=∂ψj∂xi=gij=δij at p and
⟨ei|∇⊤Xi⟩=div⊤X=⟨ei|∇⊤Xiτ⟩+⟨ei|∇⊤(⟨X|νE⟩νiE)⟩=⟨ei|∇Xiτ⟩+⟨X|νE⟩⟨ei|∇TnνiE⟩=⟨ei|∇Xiτ⟩+⟨X|νE⟩∂ψj∂xihjlgls∂ψi∂xs=∇eiXiτ+⟨X|νE⟩hii=divXτ+⟨X|νE⟩H, |
where we used again the Gauss–Weingarten relations (2.9) and the fact that the covariant derivative of a tangent vector field along a hypersurface of Rn can be obtained by differentiating in Rn (a local extension of) the vector field and projecting the result on the tangent space to the hypersurface (see [26], for instance). Hence, we get
⟨νE|dX(νE)⟩=divTnX−⟨ei|∇⊤Xi⟩=divTnX−divXτ−⟨X|νE⟩H |
and Eq (2.37) follows by substituting this left term in formula (2.38).
Remark 2.15. We are not aware of the presence in literature of this "geometric" line in deriving the (first and) second variation of J, moreover, in [9,Theorem 2.6,Step 3,Eq 2.67], this latter is obtained only at a critical set, while in [6,Theorem 3.6] the methods are strongly "analytic" and in our opinion less straightforward. These two papers are actually the ones on which is based the computation in [2,Theorem 3.1] of the second variation of J at a general smooth set E⊆Tn. Anyway, in this last paper, the variations of E are all special variations, that is, they are given by the flows in system (2.21), indeed, the term with the time derivative of Xt is missing (see formulas 3.1 and 7.2 in [2]).
Notice that the second variation in general does not depend only on the normal component ⟨X|νE⟩ of the restriction to ∂E of the infinitesimal generator X of a variation Φ (this will anyway be true at a critical set E, see below), due to the presence of the Z–term and of B(Xτ,Xτ) depending also on the tangential component of X and of its behavior around ∂E. Even if we restrict ourselves to the special variations coming from system (2.21), with a normal infinitesimal generator X, which imply that all the vector fields Xt are the same and coinciding with X, hence Z=dX(X) and Xτ=0, the second variation still depends also on the behavior of X in a neighborhood of ∂E (as Z). However, there are very particular case in which it depend only on ⟨X|νE⟩, for instance when the variation is special and X is normal with zero divergence (of Tn) on ∂E (in particular, if divTnX=0 in a neighborhood of ∂E or in the whole Tn), as it can be seen easily by the second form of the remainder term R in the above theorem.
We see now how the second variation behaves at a critical set of J.
Corollary 2.16. If E⊆Tn is a critical set for J, there holds
d2dt2J(Et)|t=0=∫∂E(|∇⟨X|νE⟩|2−⟨X|νE⟩2|B|2)dμ+8γ∫∂E∫∂EG(x,y)⟨X|νE(x)⟩⟨X|νE(y)⟩dμ(x)dμ(y)+4γ∫∂E∂νEvE⟨X|νE⟩2dμ, |
for every variation Et of E, hence, the second variation of J at E depends only on the normal component of the restriction of the infinitesimal generator X to ∂E, that is, on ⟨X|νE⟩.
When γ=0 we get the well known second variation of the Area functional at a smooth set E such that ∂E is a minimal surface in Rn,
d2dt2A(∂Et)|t=0=∫∂E(|∇⟨X|νE⟩|2−⟨X|νE⟩2|B|2)dμ. |
Proof. The thesis follows immediately, recalling that there holds H+4γvE=0, by Corollary 2.7, hence the remainder term R in formula (2.36) is zero.
Finally, we see that the second variation has the same form (that is, R=0) also for J under a volume constraint, at a critical set.
Proposition 2.17. If E⊆Tn is a critical set for J under a volume constraint, there holds
d2dt2J(Et)|t=0=∫∂E(|∇⟨X|νE⟩|2−⟨X|νE⟩2|B|2)dμ+8γ∫∂E∫∂EG(x,y)⟨X|νE(x)⟩⟨X|νE(y)⟩dμ(x)dμ(y)+4γ∫∂E∂νEvE⟨X|νE⟩2dμ, |
for every volume–preserving variation Et of E, hence, the second variation of J at E depends only on the normal component of the restriction of the infinitesimal generator X to ∂E, that is, on ⟨X|νE⟩.
When γ=0 we get the second variation of the Area functional under a volume constraint, at a smooth set E such that ∂E has constant mean curvature,
d2dt2A(∂Et)|t=0=∫∂E(|∇⟨X|νE⟩|2−⟨X|νE⟩2|B|2)dμ. |
Proof. By Proposition 2.10, the function H+4γvE is equal to a constant λ∈R on ∂E, then the remainder term R in formula (2.36) becomes
R=λ∫∂E(H⟨X|ν⟩2+⟨Z|ν⟩−2⟨Xτ|∇⟨X|ν⟩⟩+B(Xτ,Xτ))dμ. |
Computing, in the same hypotheses and notations of Proposition 2.13, the second derivative of the (constant) volume of Et, by Eqs (2.22)–(2.23) we have (recalling formulas (2.15), (2.28) and using the divergence theorem)
0=d2dt2Vol(Et)|t=0=ddt∫EtdivXt(x)dx|t=0=ddt∫∂E⟨X|νEt⟩dμt|t=0=∫∂E[divXτ⟨X|νE⟩+H⟨X|νE⟩2+⟨Z|νE⟩−⟨Xτ|∇⟨X|νE⟩⟩+B(Xτ,Xτ)]dμ=∫∂E[H⟨X|νE⟩2+⟨Z|νE⟩−2⟨Xτ|∇⟨X|νE⟩⟩+B(Xτ,Xτ)]dμ, | (2.39) |
hence R=0 and we are done.
Remark 2.18. Notice that by the previous computation and relation (2.37), it follows
d2dt2Vol(Et)|t=0=∫∂E[⟨X|νE⟩divTnX+⟨∂Xt∂t|t=0|ν⟩]dμ=0, | (2.40) |
for every volume–preserving variation Et of E. Hence, if we restrict ourselves to the special (volume–preserving) variations coming from system (2.21), as in [2], we have
d2dt2Vol(Et)|t=0=∫∂E⟨X|νE⟩divTnXdμ=0, |
indeed, for such variations we have Xt=X, for every t∈(−ε,ε). One can clearly use equality (2.40) to show the above proposition, as the term R reduces (using the second form in Theorem 2.14) to
R=λ∫∂E[⟨X|νE⟩divTnX+⟨∂Xt∂t|t=0|ν⟩]dμ, |
by the divergence theorem.
Moreover, we see that if we have a special variation generated by a vector field X such that divTnX=0 on ∂E, then d2dt2Vol(Et)|t=0=0 and if E is a critical set, R=0. This is then true for the special volume–preserving variations coming from Lemma 2.8 and when X is a constant vector field, hence the associated special variation Et is simply a translation of E (clearly, in this case J(Et) is constant and the first and second variations are zero).
By Proposition 2.17, the second variation of the functional J under a volume constraint at a smooth critical set E is a quadratic form in the normal component on ∂E of the infinitesimal generator X∈C∞(Tn;Rn) of a volume–preserving variation, that is, on φ=⟨X|νE⟩. This and the fact that the infinitesimal generators of the volume–preserving variations are "characterized" by having zero integral of such normal component on ∂E, by Lemma 2.8 and the discussion immediately before, motivate the following definition.
Definition 2.19. Given any smooth open set E⊆Tn we define the space of (Sobolev) functions (see [5])
˜H1(∂E)={φ:∂E→R:φ∈H1(∂E) and ∫∂Eφdμ=0}, |
and the quadratic form ΠE:˜H1(∂E)→R as
ΠE(φ)=∫∂E(|∇φ|2−φ2|B|2)dμ+8γ∫∂E∫∂EG(x,y)φ(x)φ(y)dμ(x)dμ(y)+4γ∫∂E∂νEvEφ2dμ, | (2.41) |
with the notations of Theorem 2.14.
Remark 2.20. Letting for φ∈˜H1(∂E),
vφ(x)=∫∂EG(x,y)φ(y)dμ(y), |
it follows (from the properties of the Green's function) that vφ satisfies distributionally −Δvφ=φμ in Tn, indeed,
∫Tn⟨∇vφ(x)|∇ψ(x)⟩dx=−∫Tnvφ(x)Δψ(x)dx=−∫Tn∫∂EG(x,y)φ(y)Δψ(x)dμ(y)dx=−∫∂Eφ(y)∫TnG(x,y)Δψ(x)dxdμ(y)=−∫∂Eφ(y)∫TnΔG(x,y)ψ(x)dxdμ(y)=∫∂Eφ(y)[ψ(y)−∫Tnψ(x)dx]dμ(y)=∫∂Eφ(y)ψ(y)dμ(y), |
for all ψ∈C∞(Tn), since ∫∂Eφ(y)dμ(y)=0. Therefore, taking ψ=vφ, we have
∫Tn|∇vφ(x)|2dx=∫∂Eφ(y)vφ(y)dμ(y), |
hence, the following identity holds
∫∂E∫∂EG(x,y)φ(x)φ(y)dμ(x)dμ(y)=∫∂Eφ(y)vφ(y)dμ(y)=∫Tn|∇vφ(x)|2dx, |
and we can write
ΠE(φ)=∫∂E(|∇φ|2−φ2|B|2)dμ+8γ∫Tn|∇vφ|2dx+4γ∫∂E∂νEvEφ2dμ, | (2.42) |
for every φ∈˜H1(∂E).
Definition 2.21. Given any smooth open set E⊆Tn, we say that a smooth vector field X∈C∞(Tn;Rn) is admissible for E if the function φ:∂E→R given by φ=⟨X|νE⟩ belongs to ˜H1(∂E), that is, has zero integral on ∂E.
Remark 2.22. Clearly, if X∈C∞(Tn;Rn) is the infinitesimal generator of a volume–preserving variation for E, then X is admissible, by the discussion after Corollary 2.7.
Remark 2.23. By what we said above, if E is a smooth critical set for J under a volume constraint, we can from now on consider only the special variations Et=Φt(E) associated to admissible vector fields X, given by the flow Φ defined by system (2.21), hence
ddtJ(Et)|t=0=∫∂E⟨X|νE⟩dμ=0 |
and
d2dt2J(Et)|t=0=ΠE(⟨X|νE⟩) |
where ΠE is the quadratic form defined by formula (2.41).
We notice that every constant vector field X=η∈Rn is clearly admissible, as
∫∂E⟨η|νE⟩dμ=∫Edivηdx=0 |
and the associated flow is given by Φ(t,x)=x+tη, then, by the translation invariance of the functional J, we have J(Et)=J(E) and
0=d2dt2J(Et)|t=0=ΠE(⟨η|νE⟩), |
that is, the form ΠE is zero on the vector subspace
T(∂E)={⟨η|νE⟩:η∈Rn}⊆˜H1(∂E) |
of dimension clearly less than or equal to n. We split
˜H1(∂E)=T(∂E)⊕T⊥(∂E), | (2.43) |
where T⊥(∂E)⊆˜H1(∂E) is the vector subspace L2–orthogonal to T(∂E) (with respect to the measure μ on ∂E), that is,
T⊥(∂E)={φ∈˜H1(∂E):∫∂EφνEdμ=0}={φ∈H1(∂E):∫∂Eφdμ=0 and ∫∂EφνEdμ=0} |
and we give the following "stability" conditions.
Definition 2.24 (Stability). We say that a critical set E⊆Tn for J under a volume constraint is stable if
ΠE(φ)≥0for all φ∈˜H1(∂E) |
and strictly stable if moreover
ΠE(φ)>0for all φ∈T⊥(∂E)∖{0}. |
Remark 2.25. Introducing the symmetric bilinear form associated (by polarization) to ΠE on ˜H1(∂E),
bE(φ,ψ)=ΠE(φ+ψ)−ΠE(φ−ψ)4 |
at a critical set E⊆Tn, it can be seen that actually T(∂E) is a degenerate vector subspace of ˜H1(∂E) for bE, that is, bE(φ,ψ)=0 for every φ∈˜H1(∂E) and ψ∈T(∂E). Indeed, we observe that by formula (2.1) and the properties of the Green function, we get
∇vE(x)=∫Tn∇xG(x,y)uE(x)dy=∫E∇xG(x,y)dy−∫Ec∇xG(x,y)dy=−∫E∇yG(x,y)dy+∫Ec∇yG(x,y)dy=−2∫∂EG(x,y)νE(y)dμ(y), | (2.44) |
where in the last passage we applied the divergence theorem.
By means of formula (2.11)
ΔνE=∇H−|B|2νE, |
since E (being critical) satisfies H+4γvE=λ for some constant λ∈R, we have
−ΔνE−|B|2νE=∇(4γvE−λ)=∇Tn(4γvE−λ)−∂νE(4γvE−λ)=−4γ(∂νEvE)νE−8γ∫∂EG(x,y)νE(y)dμ(y) |
on ∂E, by formula (2.44).
This equation can be written as L(νi)=0, for every i∈{1,…,n}, where L is the self–adjoint, linear operator defined as
L(φ)=−Δφ−|B|2φ+4γ∂νEvEφ+8γ∫∂EG(x,y)φ(y)dμ(y), |
which clearly satisfies
bE(φ,ψ)=∫∂E⟨L(φ)|ψ⟩dμ and ΠE(φ)=∫∂E⟨L(φ)|φ⟩dμ. |
Then, if we "decompose" a smooth function φ∈˜H1(∂E) as φ=ψ+⟨η|νE⟩, for some η∈Rn and ψ∈T⊥(∂E), we have (recalling formula (2.41))
ΠE(φ)=∫∂E⟨L(φ)|φ⟩dμ=∫∂E⟨L(ψ)|ψ⟩dμ+2∫∂E⟨L(⟨η|νE⟩)|ψ⟩dμ+∫∂E⟨L(⟨η|νE⟩)|⟨η|νE⟩⟩dμ=ΠE(ψ). |
By approximation with smooth functions, we conclude that this equality holds for every function in ˜H1(∂E).
The initial claim about the form bE then easily follows by its definition. Moreover, if E is a strictly stable critical set there holds
ΠE(φ)>0for everyφ∈˜H1(∂E)∖T(∂E). | (2.45) |
Remark 2.26. We observe that there exists an orthonormal frame {e1,…,en} of Rn such that
∫∂E⟨νE|ei⟩⟨νE|ej⟩dμ=0, | (2.46) |
for all i≠j, indeed, considering the symmetric n×n–matrix A=(aij) with components aij=∫∂EνiEνjEdμ, where νiE=⟨νE|εi⟩ for some basis {ε1,…,εn} of Rn, we have
∫∂E(OνE)i(OνE)jdμ=(OAO−1)ij, |
for every O∈SO(n). Choosing O such that OAO−1 is diagonal and setting ei=O−1εi, relations (2.46) are clearly satisfied.
Hence, the functions ⟨νE|ei⟩ which are not identically zero are an orthogonal basis of T(∂E). We set
IE={i∈{1,…,n}:⟨νE|ei⟩ is not identically zero} |
and
OE=Span{ei:i∈IE}, | (2.47) |
then, given any φ∈˜H1(∂E), its projection on T⊥(∂E) is
π(φ)=φ−∑i∈IE∫∂Eφ⟨νE|ei⟩dμ‖⟨νE|ei⟩‖2L2(∂E)⟨νE|ei⟩. | (2.48) |
From now on we will extensively use Sobolev spaces on smooth hypersurfaces. Most of their properties hold as in Rn, standard references are [3] in the Euclidean space and [5] when the ambient is a manifold.
Given a smooth set E⊆Tn, for ε>0 small enough, we let (d is the "Euclidean" distance on Tn)
Nε={x∈Tn:d(x,∂E)<ε} | (2.49) |
to be a tubular neighborhood of ∂E such that the orthogonal projection map πE:Nε→∂E giving the (unique) closest point on ∂E and the signed distance function dE:Nε→R from ∂E
dE(x)={d(x,∂E)ifx∉E−d(x,∂E)ifx∈E | (2.50) |
are well defined and smooth in Nε (for a proof of the existence of such tubular neighborhood and of all the subsequent properties, see [43] for instance). Moreover, for every x∈Nε, the projection map is given explicitly by
πE(x)=x−∇d2E(x)/2=x−dE(x)∇dE(x) | (2.51) |
and the unit vector ∇dE(x) is orthogonal to ∂E at the point πE(x)∈∂E, indeed actually
∇dE(x)=∇dE(πE(x))=νE(πE(x)), | (2.52) |
which means that the integral curves of the vector field ∇dE are straight segments orthogonal to ∂E.
This clearly implies that the map
∂E×(−ε,ε)∋(y,t)↦L(y,t)=y+t∇dE(y)=y+tνE(y)∈Nε | (2.53) |
is a smooth diffeomorphism with inverse
Nε∋x↦L−1(x)=(πE(x),dE(x))∈∂E×(−ε,ε). |
Moreover, denoting with JL its Jacobian (relative to the hypersurface ∂E), there holds
0<C1≤JL(y,t)≤C2 |
on ∂E×(−ε,ε), for a couple of constants C1,C2, depending on E and ε.
By means of such tubular neighborhood of a smooth set E⊆Tn and the map L, we can speak of "Wk,p–closedness" (or "Ck,α–closedness") to E of another smooth set F⊆Tn, asking that for some δ>0 "small enough", we have Vol(E△F)<δ and that ∂F is contained in a tubular neighborhood Nε of E, as above, described by
∂F={y+ψ(y)νE(y):y∈∂E}, |
for a smooth function ψ:∂E→R with ‖ψ‖Wk,p(∂E)<δ (resp. ‖ψ‖Ck,α(∂E)<δ). That is, we are asking that the two sets E and F differ by a set of small measure and that their boundaries are "close" in Wk,p (or Ck,α) as graphs.
Notice that
ψ(y)=π2∘L−1(∂E∩{y+λνE(y):λ∈R}), |
where π2:∂E×(−ε,ε)→R is the projection on the second factor.
Moreover, given a sequence of smooth sets Fi⊆Tn, we will write Fi→E in Wk,p (resp. Ck,α) if for every δ>0, there hold Vol(Fi△E)<δ, the smooth boundary ∂Fi is contained in some Nε, relative to E and it is described by
∂Fi={y+ψi(y)νE(y):y∈∂E}, |
for a smooth function ψi:∂E→R with ‖ψi‖Wk,p(∂E)<δ (resp. ‖ψi‖Ck,α(∂E)<δ), for every i∈N large enough.
From now on, in all the rest of the work, we will refer to the volume–constrained nonlocal Area functional J (and Area functional A), sometimes without underlining the presence of such constraint, by simplicity. Moreover, with Nε we will always denote a suitable tubular neighborhood of a smooth set, with the above properties.
Definition 2.27. We say that a smooth set E⊆Tn is a local minimizer for the functional J (for the Area functional A) if there exists δ>0 such that
J(F)≥J(E)(A(F)≥A(E)) |
for all smooth sets F⊆Tn with Vol(F)=Vol(E) and Vol(E△F)<δ.
We say that a smooth set E⊆Tn is a W2,p–local minimizer if there exists δ>0 and a tubular neighborhood Nε of E, as above, such that
J(F)≥J(E)(A(F)≥A(E)) |
for all smooth sets F⊆Tn with Vol(F)=Vol(E), Vol(E△F)<δ and ∂F contained in Nε, described by
∂F={y+ψ(y)νE(y):y∈∂E}, |
for a smooth function ψ:∂E→R with ‖ψ‖W2,p(∂E)<δ.
Clearly, any local minimizer is a W2,p–local minimizer.
We immediately show a necessary condition for W2,p–local minimizers.
Proposition 2.28. Let the smooth set E⊆Tn be a W2,p–local minimizer of J, then E is a critical set and
ΠE(φ)≥0for all φ∈˜H1(∂E), |
in particular, E is stable.
Proof. If E is a W2,p–local minimizer of J, given any φ∈C∞(∂E)∩˜H1(∂E), we consider the admissible vector field X∈C∞(Tn;Rn) given by Lemma 2.8 and the associated flow Φ. Then, the variation Et=Φt(E) of E is volume–preserving, that is, Vol(Et)=Vol(E) and for every δ>0, there clearly exists a tubular neighborhood Nε of E and ¯ε>0 such that for t∈(−¯ε,¯ε) we have
Vol(E△Et)<δ |
and
∂Et={y+ψ(y)νE(y):y∈∂E}⊆Nε |
for a smooth function ψ:∂E→R with ‖ψ‖W2,p(∂E)<δ. Hence, the W2,p–local minimality of E implies
J(E)≤J(Et), |
for every t∈(−¯ε,¯ε). It follows
0=ddtJ(Et)|t=0=∫∂E(H+4γvE)φdμ, |
by Theorem 2.6, which implies that E is a critical set, by the subsequent discussion and
0≤d2dt2J(Et)|t=0=ΠE(φ), |
by Proposition 2.17 and Remark 2.23.
Then, the thesis easily follows by the density of C∞(∂E)∩˜H1(∂E) in ˜H1(∂E) (see [5], for instance) and the definition of ΠE, formula (2.41).
The rest of this section will be devoted to show that the strict stability (see Definition 2.24) is a sufficient condition for the W2,p–local minimality. Precisely, we will prove the following theorem.
Theorem 2.29. Let p>max{2,n−1} and E⊆Tn a smooth strictly stable critical set for the nonlocal Area functional J (under a volume constraint), with Nε a tubular neighborhood of ∂E as in formula (2.49). Then, there exist constants δ,C>0 such that
J(F)≥J(E)+C[α(E,F)]2, |
for all smooth sets F⊆Tn such that Vol(F)=Vol(E), Vol(F△E)<δ, ∂F⊆Nε and
∂F={y+ψ(y)νE(y):y∈∂E}, |
for a smooth function ψ with ‖ψ‖W2,p(∂E)<δ, where the "distance" α(E,F) is defined as
α(E,F)=minη∈RnVol(E△(F+η)). |
As a consequence, E is a W2,p–local minimizer of J. Moreover, if F is W2,p–close enough to E and J(F)=J(E), then F is a translate of E, that is, E is locally the unique W2,p–local minimizer, up to translations.
Remark 2.30. We could have introduced the definitions of strict local minimizer or strict W2,p–local minimizer for the nonlocal Area functional, by asking that the inequalities J(F)≤J(E) in Definition 2.27 are equalities if and only if F is a translate of E. With such notion, the conclusion of this theorem is that E is actually a strict W2,p–local minimizer (with a "quantitative" estimate of its minimality).
Remark 2.31. With some extra effort, it can be proved that in the same hypotheses of Theorem 2.29, the set F is actually a local minimizer (see [2]). Since in the analysis of the modified Mullins–Sekerka and surface diffusion flow in the next sections we do not need such a stronger result, we omitted to prove it.
For the proof of this result we need some technical lemmas. We underline that most of the difficulties are due to the presence of the degenerate subspace T(∂E) of the form ΠE (where it is zero), related to the translation invariance of the nonlocal Area functional (recall the discussion after Definition 2.19).
In the next key lemma we are going to show how to construct volume–preserving variations (hence, admissible smooth vector fields) "deforming" a set E to any other smooth set with the same volume, which is W2,p–close enough. By the same technique we will also prove Lemma 2.8 immediately after, whose proof was postponed from Subsection 2.1.
Lemma 2.32. Let E⊆Tn be a smooth set and Nε a tubular neighborhood of ∂E as above, in formula (2.49). For all p>n−1, there exist constants δ,C>0 such that if ψ∈C∞(∂E) and ‖ψ‖W2,p(∂E)≤δ, then there exists a vector field X∈C∞(Tn;Rn) with divX=0 in Nε and the associated flow Φ, defined by system (2.21), satisfies
Φ(1,y)=y+ψ(y)νE(y),for all y∈∂E. | (2.54) |
Moreover, for every t∈[0,1]
‖Φ(t,⋅)−Id‖W2,p(∂E)≤C‖ψ‖W2,p(∂E). | (2.55) |
Finally, if Vol(E1)=Vol(E), then the variation Et=Φt(E) is volume–preserving, that is, Vol(Et)=Vol(E) for all t∈[−1,1] and the vector field X is admissible.
Proof. We start considering the vector field ˜X∈C∞(Nε;Rn) defined as
˜X(x)=ξ(x)∇dE(x) | (2.56) |
for every x∈Nε, where dE:Nε→R is the signed distance function from E and ξ:Nε→R is the function defined as follows: for all y∈∂E, we let fy:(−ε,ε)→R to be the unique solution of the ODE
{f′y(t)+fy(t)ΔdE(y+tνE(y))=0fy(0)=1 |
and we set
ξ(x)=ξ(y+tνE(y))=fy(t)=exp(−∫t0ΔdE(y+sνE(y))ds), |
recalling that the map (y,t)↦x=y+tνE(y) is a smooth diffeomorphism between ∂E×(−ε,ε) and Nε (with inverse x↦(πE(x),dE(x)), where πE is the orthogonal projection map on E, defined by formula (2.51)). Notice that the function f is always positive, thus the same holds for ξ and ξ=1, ∇dE=νE, hence ˜X=νE on ∂E.
Our aim is then to prove that the smooth vector field X defined by
X(x)=∫ψ(πE(x))0dsξ(πE(x)+sνE(πE(x)))˜X(x) | (2.57) |
for every x∈Nε and extended smoothly to all Tn, satisfies all the properties of the statement of the lemma.
Step 1. We saw that ˜X|∂E=νE, now we show that div˜X=0 and analogously divX=0 in Nε.
Given any x=y+tνE(y)∈Nε, with y∈∂E, we have
div˜X(x)=div[ξ(x)∇dE(x)]=⟨∇ξ(x)|∇dE(x)⟩+ξ(x)ΔdE(x)=∂∂t[ξ(y+tνE(y))]+ξ(y+tνE(y))ΔdE(y+tνE(y))=f′y(t)+fy(t)ΔdE(y+tνE(y))=0, |
where we used the fact that f′y(t)=⟨∇ξ(y+tνE(y))|νE(y)⟩ and ∇dE(y+tνE(y))=νE(y), by formula (2.52).
Since the function
x↦θ(x)=∫ψ(πE(x))0dsξ(πE(x)+sν(πE(x)) |
is clearly constant along the segments t↦x+t∇dE(x), for every x∈Nε, it follows that
0=∂∂t[θ(x+t∇dE(x))]|t=0=⟨∇θ(x)|∇dE(x)⟩, |
hence,
divX=⟨∇θ|∇dE⟩ξ+θdiv˜X=0. |
Step 2. Recalling that ψ∈C∞(∂E) and p>n−1, we have
‖ψ‖L∞(∂E)≤‖ψ‖C1(∂E)≤CE‖ψ‖W2,p(∂E), |
by Sobolev embeddings (see [5]). Then, we can choose δ<ε/CE such that for all x∈∂E we have that x±ψ(x)νE(x)∈Nε.
To check that equation (2.54) holds, we observe that
θ(x)=∫ψ(πE(x))0dsξ(πE(x)+sνE(πE(x))) |
represents the time needed to go from πE(x) to πE(x)+ψ(πE(x))νE(πE(x)) along the trajectory of the vector field ˜X, which is the segment connecting πE(x) and πE(x)+ψ(πE(x))νE(πE(x)), of length ψ(πE(x)), parametrized as
s↦πE(x)+sψ(πE(x))νE(πE(x)), |
for s∈[0,1] and which is traveled with velocity ξ(πE(x)+sνE(πE(x)))=fπE(x)(s). Therefore, by the above definition of X=θ˜X and the fact that the function θ is constant along such segments, we conclude that
Φ(1,y)−Φ(0,y)=ψ(y)νE(y), |
that is, Φ(1,y)=y+ψ(y)νE(y), for all y∈∂E.
Step 3. To establish inequality (2.55), we first show that
‖X‖W2,p(Nε)≤C‖ψ‖W2,p(∂E) | (2.58) |
for a constant C>0 depending only on E and ε. This estimate will follow from the definition of X in Eq (2.57) and the definition of W2,p–norm, that is,
‖X‖W2,p(Nε)=‖X‖Lp(Nε)+‖∇X‖Lp(Nε)+‖∇2X‖Lp(Nε). |
As |∇dE|=1 everywhere and the positive function ξ satisfies 0<C1≤ξ≤C2 in Nε, for a pair of constants C1 and C2, we have
‖X‖pLp(Nε)=∫Nε|∫ψ(πE(x))0dsξ(πE(x)+sνE(πE(x)))ξ(x)∇dE(x)|pdx≤‖ξ‖pL∞(Nε)∫Nε|∫ψ(πE(x))0dsξ(πE(x)+sνE(πE(x)))|pdx≤Cp2Cp1∫Nε|ψ(πE(x))|pdx=Cp2Cp1∫∂E∫ε−ε|ψ(πE(y+tνE(y)))|pJL(y,t)dtdμ(y)=Cp2Cp1∫∂E|ψ(y)|p∫ε−εJL(y,t)dtdμ(y)≤C∫∂E|ψ(y)|pdμ(y)=C‖ψ‖pLp(∂E), |
where L:∂E×(−ε,ε)→Nε the smooth diffeomorphism defined in formula (2.53) and JL its Jacobian. Notice that the constant C depends only on E and ε.
Now we estimate the Lp–norm of ∇X. We compute
∇X=∇ψ(πE(x))dπE(x)ξ(πE(x)+ψ(πE(x))νE(πE(x)))ξ(x)∇dE(x)−[∫ψ(πE(x))0∇ξ(πE(x)+sνE(πE(x)))ξ2(πE(x)+sνE(πE(x)))dπE(x)Idds]ξ(x)∇dE(x)−[∫ψ(πE(x))0∇ξ(πE(x)+sνE(πE(x)))ξ2(πE(x)+sνE(πE(x)))dπE(x)sdνE(πE(x))ds]ξ(x)∇dE(x)+∫ψ(πE(x))0dsξ(πE(x)+sνE(πE(x)))(∇ξ(x)∇dE(x)+ξ(x)∇2dE(x)) |
and we deal with the integrals in the three terms as before, changing variable by means of the function L. That is, since all the functions dπE, dνE, ∇2dE, ξ, 1/ξ, ∇ξ are bounded by some constants depending only on E and ε, we easily get (the constant C could vary from line to line)
‖∇X‖pLp(Nε)≤C∫Nε|∇ψ(πE(x))|pdx+C∫Nε|ψ(πE(x))|pdx=C∫∂E∫ε−ε|∇ψ(πE(y+tνE(y)))|pJL(y,t)dtdμ(y)+C∫∂E∫ε−ε|ψ(πE(y+tνE(y)))|pJL(y,t)dtdμ(y)=C∫∂E(|ψ(y)|p+|∇ψ(y)|p)∫ε−εJL(y,t)dtdμ(y)≤C‖ψ‖pLp(∂E)+C‖∇ψ‖pLp(∂E)≤C‖ψ‖pW1,p(∂E). |
A very analogous estimate works for ‖∇2X‖pLp(Nε) and we obtain also
‖∇2X‖pLp(Nε)≤C‖ψ‖pW2,p(∂E), |
hence, inequality (2.58) follows with C=C(E,ε).
Applying now Lagrange theorem to every component of Φ(⋅,y) for any y∈∂E and t∈[0,1], we have
Φi(t,y)−yi=Φi(t,y)−Φi(0,y)=tXi(Φ(s,y)), |
for every i∈{1,…,n}, where s=s(y,t) is a suitable value in (0,1). Then, it clearly follows
‖Φ(t,⋅)−Id‖L∞(∂E)≤C‖X‖L∞(Nε)≤C‖X‖W2,p(Nε)≤C‖ψ‖W2,p(∂E) | (2.59) |
by estimate (2.58), with C=C(E,ε) (notice that we used Sobolev embeddings, being p>n−1, the dimension of ∂E).
Differentiating the equations in system (2.21), we have (recall that we use the convention of summing over the repeated indices)
{∂∂t∇iΦj(t,y)=∇kXj(Φ(t,y))∇iΦk(t,y)∇iΦj(0,y)=δij | (2.60) |
for every i,j∈{1,…,n}. It follows,
∂∂t|∇iΦj(t,y)−δij|2≤2|(∇iΦj(t,y)−δij)∇kXj(Φ(t,y))∇iΦk(t,y)|≤2‖∇X‖L∞(Nε)|∇iΦj(t,y)−δij|2+2‖∇X‖L∞(Nε)|∇iΦj(t,y)−δij| |
hence, for almost every t∈[0,1], where the following derivative exists,
∂∂t|∇iΦj(t,y)−δij|≤C‖∇X‖L∞(Nε)(|∇iΦj(t,y)−δij|+1). |
Integrating this differential inequality, we get
|∇iΦj(t,y)−δij|≤etC‖∇X‖L∞(Nε)−1≤eC‖X‖W2,p(Nε)−1, |
as t∈[0,1], where we used Sobolev embeddings again. Then, by inequality (2.58), we estimate
∑1≤i,j≤n‖∇iΦj(t,⋅)−δij‖L∞(∂E)≤C(eC‖ψ‖W2,p(∂E)−1)≤C‖ψ‖W2,p(∂E), | (2.61) |
as ‖ψ‖W2,p(∂E)≤δ, for any t∈[0,1] and y∈∂E, with C=C(E,ε,δ).
Differentiating Eq (2.60), we obtain
{∂∂t∇ℓ∇iΦj(t,y)=∇s∇kXj(Φ(t,y))∇iΦk(t,y)∇ℓΦs(t,y)∂∂t∇ℓ∇iΦj(t,y)=+∇kXj(Φ(t,y))∇ℓ∇iΦk(t,y)∇ℓ∇iΦ(0,y)=0 |
(where we sum over s and k), for every t∈[0,1], y∈∂E and i,j,ℓ∈{1,…,n}.
This is a linear non–homogeneous system of ODEs such that, if we control C‖ψ‖W2,p(∂E), the smooth coefficients in the right side multiplying the solutions ∇ℓ∇iΦj(⋅,y) are uniformly bounded (as in estimate (2.61), Sobolev embeddings then imply that ∇X is bounded in L∞ by C‖ψ‖W2,p(∂E)). Hence, arguing as before, for almost every t∈[0,1] where the following derivative exists, there holds
∂∂t|∇2Φ(t,y)|≤C‖∇X‖L∞(Nε)|∇2Φ(t,y)|+C|∇2X(Φ(t,y))|≤Cδ|∇2Φ(t,y)|+C|∇2X(Φ(t,y))|, |
by inequality (2.58) (notice that inequality (2.61) gives an L∞–bound on ∇Φ, not only in Lp, which is crucial). Thus, by means of Gronwall's lemma (see [52], for instance), we obtain the estimate
|∇2Φ(t,y)|≤C∫t0|∇2X(Φ(s,y))|eCδ(t−s)ds≤C∫t0|∇2X(Φ(s,y))|ds, |
hence,
‖∇2Φ(t,⋅)‖pLp(∂E)≤C∫∂E(∫t0|∇2X(Φ(s,y))|ds)pdμ(y)≤C∫t0∫∂E|∇2X(Φ(s,y))|pdμ(y)ds=C∫Nε|∇2X(x)|pJL−1(x)dx≤C‖∇2X‖pLp(Nε)≤C‖X‖pW2,p(Nε)≤C‖ψ‖pW2,p(∂E), | (2.62) |
by estimate (2.58), for every t∈[0,1], with C=C(E,ε,δ).
Clearly, putting together inequalities (2.59), (2.61) and (2.62), we get the estimate (2.55) in the statement of the lemma.
Step 4. Finally, computing as in formula (2.39) and Remark 2.18, we have
d2dt2Vol(Et)=∫∂E⟨X|νEt⟩divTnXdμt, |
for every t∈[−1,1], hence, since by Step 1 we know that divTnX=0 in Nε (which contains each ∂Et), we conclude that d2dt2Vol(Et)=0 for all t∈[−1,1], that is, the function t↦Vol(Et) is linear.
If then Vol(E1)=Vol(E)=Vol(E0), it follows that Vol(Et)=Vol(E), for all t∈[−1,1] which implies that X is admissible, by Remark 2.22.
With an argument similar to the one of this proof, we now prove Lemma 2.8.
Proof of Lemma 2.8. Let φ:∂E→R a C∞ function with zero integral, then we define the following smooth vector field in Nε,
X(x)=φ(πE(x))˜X(x), |
where ˜X is the smooth vector field defined by formula (2.56) and we extend it to a smooth vector field X∈C∞(Tn;Rn) on the whole Tn. Clearly, by the properties of ˜X seen above,
⟨X(y)|νE(y)⟩=φ(y)⟨˜X(y)|νE(y)⟩=φ(y) |
for every y∈∂E.
As the function x↦φ(πE(x)) is constant along the segments t↦x+t∇dE(x), for every x∈Nε, it follows, as in Step 1 of the previous proof, that divX=0 in Nε. Then, arguing as in Step 4, the flow Φ defined by system (2.21) having X as infinitesimal generator, gives a variation Et=Φt(E) of E such that the function t↦Vol(Et) is linear, for t in some interval (−δ,δ). Since, by Eq (2.24), there holds
ddtVol(Et)|t=0=∫∂E⟨X|νE⟩dμ=∫∂Eφdμ=0, |
such function t↦Vol(Et) must actually be constant.
Hence, Vol(Et)=Vol(E), for all t∈(−δ,δ) and the variation Et is volume–preserving.
The next lemma gives a technical estimate needed in the proof of Theorem 2.29.
Lemma 2.33. Let p>max{2,n−1} and E⊆Tn a strictly stable critical set for the (volume–constrained) functional J. Then, in the hypotheses and notation of Lemma 2.32, there exist constants δ,C>0 such that if ‖ψ‖W2,p(∂E)≤δ then |X|≤C|⟨X|νEt⟩| on ∂Et and
‖∇X‖L2(∂Et)≤C‖⟨X|νEt⟩‖H1(∂Et) | (2.63) |
(here ∇ is the covariant derivative along Et), for all t∈[0,1], where X∈C∞(Tn;Rn) is the smooth vector field defined in formula (2.57).
Proof. Fixed ε>0, from inequality (2.55) it follows that there exist δ>0 such that if ‖ψ‖W2,p(∂E)≤δ there holds
|νEt(Φ(t,y))−νE(y)|≤ε |
for every y∈∂E, hence, as ∇dE=νE on ∂E, we have
|∇dE(Φ−1(t,x))−νEt(x)|=|νE(Φ−1(t,x))−νEt(x)|≤ε |
for every x∈∂Et. Then, if ‖ψ‖W2,p(∂E) is small enough, Φ−1(t,⋅) is close to the identity, thus
|∇dE(Φ−1(t,x))−∇dE(x)|≤ε |
on ∂Et and we conclude
‖∇dE−νEt‖L∞(∂Et)≤2ε. |
Moreover, using again the inequality (2.55) and following the same argument above, we also obtain
‖∇2dE−∇νEt‖L∞(∂Et)≤2ε. | (2.64) |
We estimate Xτt=X−⟨X|νEt⟩νEt (recall that X=⟨X|∇dE⟩∇dE),
|Xτt|=|X−⟨X|νEt⟩νEt|=|⟨X|∇dE⟩∇dE−⟨X|νEt⟩νEt|=|⟨X|∇dE⟩∇dE−⟨X|νEt⟩∇dE+⟨X|νEt⟩∇dE−⟨X|νEt⟩νEt|≤|⟨X|(∇dE−νEt)⟩∇dE|+|⟨X|νEt⟩(∇dE−νEt)|≤2|X||∇dE−νEt|≤4ε|X|, |
then
|Xτt|≤4ε|Xτt+⟨X|νEt⟩νEt|≤4ε|Xτt|+|⟨X|νEt⟩|, |
hence,
|Xτt|≤C|⟨X|νEt⟩|. | (2.65) |
We now estimate the covariant derivative of Xτt along ∂Et, that is,
|∇Xτt|=|∇X−∇(⟨X|νEt⟩νEt)|=|∇(⟨X|∇dE⟩∇dE)−∇(⟨X|νEt⟩νEt)|=|∇(⟨X|∇dE⟩∇dE)−∇(⟨X|νEt⟩∇dE)+∇(⟨X|νEt⟩∇dE)−∇(⟨X|νEt⟩νEt)|≤|∇(⟨X|(∇dE−νEt)⟩∇dE)|+|∇(⟨X|νEt⟩(∇dE−νEt))|≤Cε[|∇X|+|∇⟨X|νEt⟩|]+C|X|[|∇(∇dE)|+|∇νEt|]≤Cε[|∇(⟨X|νEt⟩νEt+Xτt)|+|∇⟨X|νEt⟩|]+C(|⟨X|νEt⟩|+|Xτt|)[|∇2dE|+|∇νEt|] |
hence, using inequality (2.65) and arguing as above, there holds
|∇Xτt|≤C|∇⟨X|νEt⟩|+C|⟨X|νEt⟩|[|∇2dE|+|∇νEt|]. |
Then, we get
‖∇Xτt‖2L2(∂Et)≤C‖∇⟨X|νEt⟩‖2L2(∂Et)+C∫∂Et|⟨X|νEt⟩|2[|∇2dE|+|∇νEt|]2dμ≤C‖⟨X|νEt⟩‖2H1(∂Et)+C‖⟨X|νEt⟩‖2L2pp−2(∂Et)‖|∇2dE|+|∇νEt|‖2Lp(∂Et)≤C ‖⟨X|νEt⟩‖2H1(∂Et) |
where in the last inequality we used as usual Sobolev embeddings, as p>max{2,n−1} and the fact that ‖∇νEt‖Lp(∂Et) is bounded by the inequality (2.64) (as ‖∇2dE‖Lp(∂Et)).
Considering the covariant derivative of X=Xτt+⟨X|νEt⟩νEt, by means of this estimate, the trivial one
‖∇⟨X|νEt⟩‖L2(∂Et)≤‖⟨X|νEt⟩‖H1(∂Et) |
and inequality (2.65), we obtain estimate (2.63).
We now show that any smooth set E sufficiently W2,p–close to another smooth set F, can be "translated" by a vector η∈Rn such that ∂E−η={y+φ(y)νF(y):y∈∂F}, for a function φ∈C∞(∂F) having a suitable small "projection" on T(∂F) (see the definitions and the discussion after Remark 2.23).
Lemma 2.34. Let p>n−1 and F⊆Tn a smooth set with a tubular neighborhood Nε as above, in formula (2.49). For any τ>0 there exist constants δ,C>0 such that if another smooth set E⊆Tn satisfies Vol(E△F)<δ and ∂E={y+ψ(y)νF(y):y∈∂F}⊆Nε for a function ψ∈C∞(R) with ‖ψ‖W2,p(∂F)<δ, then there exist η∈Rn and φ∈C∞(∂F) with the following properties:
∂E−η={y+φ(y)νF(y):y∈∂F}⊆Nε, |
|η|≤C‖ψ‖W2,p(∂F),‖φ‖W2,p(∂F)≤C‖ψ‖W2,p(∂F) |
and
|∫∂FφνFdμ|≤τ‖φ‖L2(∂F). |
Proof. We let dF to be the signed distance function from ∂F. We underline that, throughout the proof, the various constants will be all independent of ψ:∂F→R.
We recall that in Remark 2.26 we saw that there exists an orthonormal basis {e1,…,en} of Rn such that the functions ⟨νF|ei⟩ are orthogonal in L2(∂F), that is,
∫∂F⟨νF|ei⟩⟨νF|ej⟩dμ=0, | (2.66) |
for all i≠j and we let IF to be the set of the indices i∈{1,…,n} such that ‖⟨νF|ei⟩‖L2(∂F)>0. Given a smooth function ψ:∂F→R, we set η=∑ni=1ηiei, where
ηi={1‖⟨νF|ei⟩‖2L2(∂F)∫∂Fψ(x)⟨νF(x)|ei⟩dμifi∈IF,ηi=0otherwise. | (2.67) |
Note that, from Hölder inequality, it follows
|η|≤C1‖ψ‖L2(∂F). | (2.68) |
Step 1. Let Tψ:∂F→∂F be the map
Tψ(y)=πF(y+ψ(y)νF(y)−η). |
It is easily checked that there exists ε0>0 such that if
‖ψ‖W2,p(∂F)+|η|≤ε0≤1, | (2.69) |
then Tψ is a smooth diffeomorphism, moreover,
‖JTψ−1‖L∞(∂F)≤C‖ψ‖C1(∂F) | (2.70) |
(here JTψ is the Jacobian relative to ∂F) and
‖Tψ−Id‖W2,p(∂F)+‖T−1ψ−Id‖W2,p(∂F)≤C(‖ψ‖W2,p(∂F)+|η|). | (2.71) |
Therefore, setting ˆE=E−η, we have
for some function which is linked to by the following relation: for all , we let such that
then,
that is, and
Thus, using inequality (2.71), we have
(2.72) |
for some constant . We now estimate
(2.73) |
where
(2.74) |
by inequality (2.70).
On the other hand,
(2.75) |
where
(2.76) |
In turn, recalling inequality (2.68), we get
(2.77) |
Since in , by Eq (2.51), we have , it follows
thus, for all , there holds
From this identity and equalities (2.73), (2.75) and (2.76), we conclude
As the integral at the right–hand side vanishes by relations (2.66) and (2.67), estimates (2.74) and (2.77) imply
(2.78) |
where in the last passage we used a well–known interpolation inequality, with depending only on (see [5,Theorem 3.70]).
Step 2. The previous estimate does not allow to conclude directly, but we have to rely on the following iteration procedure. Fix any number and assume that is such that (possibly considering a smaller )
(2.79) |
Given , we set and we denote by the vector defined as in (2.67). We set and denote by the function such that . As before, satisfies
Since and , by inequalities (2.68), (2.72) and (2.79) we have
(2.80) |
Using again that , by estimate (2.78) we obtain
where we have .
We now distinguish two cases.
If , from the previous inequality and (2.79), we get
thus, the conclusion follows with .
In the other case,
(2.81) |
We then repeat the whole procedure: we denote by the vector defined as in formula (2.67) with replaced by , we set and we consider the corresponding which satisfies
Since
the map is a diffeomorphism, thanks to formula (2.69) (having chosen and small enough).
Thus, by applying inequalities (2.72) (with ), (2.68), (2.79) and (2.81), we get
as , analogously to conclusion (2.80). On the other hand, by estimates (2.68), (2.80) and (2.81),
hence, also the map is a diffeomorphism satisfying inequalities (2.69) and (2.70). Therefore, arguing as before, we obtain
Since by inequality (2.81), if the conclusion follows with . Otherwise, we iterate the procedure observing that
This construction leads to three (possibly finite) sequences , and such that
If for some we have , the construction stops, since, arguing as before,
and the conclusion follows with and . Otherwise, the iteration continues indefinitely and we get the thesis with
(notice that the series is converging), which actually means that .
We are now ready to show the main theorem of this first part of the work.
Proof of Theorem 2.29.
Step 1. We first want to see that
(2.82) |
To this aim, we consider a minimizing sequence for the above infimum and we assume that weakly in , then (since it is a closed subspace of ) and if , there holds
due to the strict stability of and the lower semicontinuity of (recall formula (2.41) and the fact that the weak convergence in implies strong convergence in by Sobolev embeddings). On the other hand, if instead , again by the strong convergence of in , by looking at formula (2.41), we have
since .
Step 2. Now we show that there exists a constant such that if is like in the statement and , with , and , then
(2.83) |
We argue by contradiction assuming that there exists a sequence of sets with with and , and a sequence of functions with and , such that
We then define the following sequence of smooth functions
(2.84) |
which clearly belong to . Setting , as , by the Sobolev embeddings, in and in , hence, the sequence is bounded in and if is the special orthonormal basis found in Remark 2.26, we have uniformly for all . Thus,
as , indeed,
and
as the Jacobians (notice that are Jacobians "relative" to the hypersurface ) uniformly and we assumed .
Hence, using expression (2.48), for the projection map on , it follows
as and
(2.85) |
since , thus , by looking at the definition of the functions in formula (2.84).
Note now that the – convergence of to (the second fundamental form of is "morally" the Hessian of ) implies
as , then, by Sobolev embeddings again (in particular for any , with which is larger than ) and the –convergence of to , we get
Standard elliptic estimates for the problem (2.3) (see [23], for instance) imply the convergence of the potentials
for , hence arguing as before,
Setting, as in Remark 2.20,
where , as , and
it is easy to check (see [2,pages 537-538], for details) that
Finally, recalling expression (2.42), we conclude
since we have
which easily follows again by looking at the definition of the functions in formula (2.84) and taking into account that , hence limits (2.85) imply
By the previous conclusion and Sobolev embeddings, it this then straightforward, arguing as above, to get also
hence,
Since we assumed that , we conclude that for , large enough there holds
which is a contradiction to Step , as .
Step 3. Let us now consider such that , and
with where is smaller than given by Step .
Taking a possibly smaller , we consider the field and the associated flow found in Lemma 2.32. Hence, in and , for all , that is, which implies and . Then the special variation is volume–preserving, for and the vector field is admissible, by the last part of such lemma.
By Lemma 2.34, choosing an even smaller if necessary, possibly replacing with a translate for some if needed, we can assume that
(2.86) |
We now claim that
(2.87) |
To this aim, we write
with appropriate and (see below).
By the definition of in formula (2.57) (in the proof of Lemma 2.32), the bounds and (by inequality (2.55) and Sobolev embeddings, as , we have ), the following inequality holds
(2.88) |
for every .
We want now to prove that for every , choosing a suitably small we have the estimate
(2.89) |
First,
then, since by equality (2.54), it follow that for every the two terms
can be made (uniformly in ) small as we want, if is small enough, by using inequality (2.88), we obtain
Then we estimate, by means of inequality (2.54) and where ,
where in the last inequality we use Eq (2.88). Hence, using equality (2.58) and Sobolev embeddings, as , we get
then, since , we obtain
if is small enough.
Arguing similarly, recalling the definition of given by formula (2.57), we also obtain , hence estimate (2.89) follows. We can then conclude that, for small enough, we have
for any , where in the last inequality we used the assumption (2.86), thus choosing we get
Along the same line, it is then easy to prove that
(2.90) |
for any , hence claim (2.87) follows.
As a consequence, since , being admissible for (recalling computation 2.23) and can be described as a graph over with a function with small norm in (by estimate (2.55) of Lemma 2.32), we can apply Step with to the function , concluding
(2.91) |
By means of Lemma 2.33, for small enough, we now show the following inequality on (here is the divergence operator and is a tangent vector field on ), for any ,
(2.92) |
where we used the Sobolev embedding , as .
Then, we compute (here is the tangent component of , is the mean curvature and the potential relative to defined by formula (2.1))
by Theorem 2.14 and the definition of in formula (2.41), considering the second form of the remainder term , relative to and taking into account that in and that , as the variation is special.
Hence, by estimate (2.91), we have (recall that constant, as is a critical set)
by estimate (2.92). If is sufficiently small, as is –close to (recall the definition of in formula (2.1)), we have
hence,
Then, we can conclude the proof of the theorem with the following series of inequalities, holding for a suitably small as in the statement,
where the first inequality is due to the –closedness of to , the second one by the very expression (2.57) of the vector field on ,
the third follows by a straightforward computation (involving the map defined by formula (2.53) and its Jacobian), as is a "normal graph" over with as "height function", finally the last one simply by the definition of the "distance" , recalling that we possibly translated the "original" set by a vector , at the beginning of this step.
We conclude this section by proving two propositions that will be used later. The first one says that when a set is sufficiently –close to a strictly stable critical set of the functional , then the quadratic form (2.41) remains uniformly positive definite (on the orthogonal complement of its degenerate subspace, see the discussion at the end of the previous subsection).
Proposition 2.35. Let and be a smooth strictly stable critical set with a tubular neighborhood of , as in formula . Then, for every there exist such that if a smooth set is –close to , that is, and with
for a smooth with , there holds
(2.93) |
for all satisfying
where is defined by formula (2.47).
Proof. Step 1. We first show that for every there holds
(2.94) |
Indeed, let be a minimizing sequence for this infimum and assume that weakly in .
If , as the weak convergence in implies strong convergence in by Sobolev embeddings, for every we have
hence,
thus, we conclude and
where the last inequality follows from estimate (2.45) in Remark 2.25.
If , then again by the strong convergence of in , by looking at formula (2.41), we have
since .
Step 2. In order to finish the proof it is enough to show the existence of some such that if and with , then
(2.95) |
where is defined by formula (2.94), with in place of . Assume by contradiction that there exist a sequence of smooth sets , with and , and a sequence , with and , such that
(2.96) |
Let us suppose first that and observe that by Sobolev embeddings for some , thus, since the functions are uniformly bounded in for , recalling formula (2.41), it is easy to see that
which is a contradiction with assumption (2.96).
Hence, we may assume that
(2.97) |
The idea now is to write every as a function on . We define the functions , given by
for every .
As in , we have in particular that
moreover, note also that in and thus in for a suitable , depending on , by Sobolev embeddings. Using this fact and taking into account the third limit above and inequality (2.97), one can easily show that
Hence, for large enough, we have
then, in turn, by Step , we infer
(2.98) |
Arguing now exactly like in the final part of Step in the proof of Theorem 2.29, we have that all the terms of are asymptotically close to the corresponding terms of , thus
which is a contradiction, by inequalities (2.96) and (2.98). This establishes inequality (2.95) and concludes the proof.
The following final result of this section states the fact that close to a strictly stable critical set there are no other smooth critical sets (up to translations).
Proposition 2.36. Let and be as in Proposition 2.35. Then, there exists such that if is a smooth critical set with , , and
for a smooth with , then is a translate of .
Proof. In Step of the proof of Theorem 2.29, it is shown that under these hypotheses on and , if is small enough, we may find a small vector and a volume–preserving variation such that, and
for all , where is a positive constant independent of .
Assume that is a smooth critical set as in the statement, which is not a translate of , then , but from the above formula it follows , which implies that cannot be critical, hence neither , which is a contradiction. Indeed, is a volume–preserving variation for and
showing that is not critical.
We start with the notion of smooth flow of sets.
Definition 3.1. Let for be a one-parameter family of sets, then we say that it is a smooth flow if there exists a smooth reference set and a map such that is a smooth diffeomorphism from to and , for all .
The velocity of the motion of any point of the set , with , is then given by
(notice that, in general, the smooth vector field , defined in the whole by for every , is not independent of ).
When , we define the outer normal velocity of the flow of the boundaries , which are smooth hypersurfaces of , as
for every , where is the outer normal vector to .
For more clarity and to simplify formulas and computations, from now on we will denote with
for every , where in the second integral is the canonical Riemannian measure induced on the hypersurface , parametrized by , by the flat metric of (coinciding with the Hausdorff –dimensional measure). Moreover, in the same spirit we set .
Before giving the definition of the modified Mullins–Sekerka flow (first appeared in [46] – see also [11,33] and [22] for a very clear and nice introduction to such flow), we need some notation. Given a smooth set and , we denote by the unique solution in of the following problem
(3.1) |
where is the potential introduced in (2.3) and is the mean curvature of . Moreover, we denote by and the restrictions and , respectively. Finally, denoting as usual by the outer unit normal to , we set
that is the "jump" of the normal derivative of on .
Definition 3.2. Let be a smooth set. We say that a smooth flow such that , is a modified Mullins–Sekerka flow with parameter , on the time interval and with initial datum , if the outer normal velocity of the moving boundaries is given by
(3.2) |
where (with the above definitions) and we used the simplified notation in place of .
Remark 3.3. The adjective "modified" comes from the introduction of the parameter in the problem, while considering we have the original flow proposed by Mullins and Sekerka in [46] (see also [11,33]), which has been also called Hele–Shaw model [7], or Hele–Shaw model with surface tension [19,20,21], which arises as a singular limit of a nonlocal version of the Cahn–Hilliard equation [4,41,50], to describe phase separation in diblock copolymer melts (see also [49]).
Parametrizing the smooth hypersurfaces of by some smooth embeddings such that (here is a fixed smooth differentiable –dimensional manifold and the map is smooth), the geometric evolution law (3.2) can be expressed equivalently as
(3.3) |
where we denoted by the outer unit normal to .
Moreover, as the moving hypersurfaces are compact, it is always possible to smoothly reparametrize them with maps (that we still call) such that
(3.4) |
in describing such flow. This follows by the invariance by tangential perturbations of the velocity, shared by the flow due to its geometric nature and can be proved following the line in Section 1.3 of [42], where the analogous property is shown in full detail for the (more famous) mean curvature flow. Roughly speaking, the tangential component of the velocity of the points of the moving hypersurfaces, does not affect the global "shape" during the motion.
Like the nonlocal Area functional (see Definition 2.2), the flow is obviously invariant by translations, or more generally under any isometry of (or ). Moreover, if is a modified Mullins–Sekerka flow of hypersurfaces, in the sense of equation (3.3) and is a time–dependent family of smooth diffeomorphisms of , then it is easy to check that the reparametrization defined as is still a modified Mullins–Sekerka flow (again in the sense of equation (3.3)). This property can be reread as "the flow is invariant under reparametrization", suggesting that the really relevant objects are actually the subsets of .
We show now that the volume of the sets is preserved during the evolution. We remark that instead, other geometric properties shared for instance by the mean curvature flow (see [42,Chapter 2]), like convexity are not necessarily maintained (see [16]), neither there holds the so–called "comparison property" asserting that if two initial sets are one contained in the other, they stay so during the two respective flows.
This volume–preserving property can be easily proved, arguing as in the computation leading to equation (2.23). Indeed, if is a modified Mullins–Sekerka flow, described by , with an associated smooth vector field as above, we have
(3.5) |
where the last equality follows from the divergence theorem and the fact that is harmonic in .
Another important property of the modified Mullins–Sekerka flow is that it can be regarded as the –gradient flow of the functional under the constraint that the volume is fixed, that is, the outer normal velocity is minus such –gradient of the functional (see [41]).
For any smooth set , we let the space to be the dual of (the functions in with zero integral) with the Gagliardo –seminorm (see [3,14,48,61], for instance)
(it is a norm for since the functions in it have zero integral) and the pairing between and simply being the integral of the product of the functions on .
We define the linear operator on the smooth functions with zero integral on as follows: we consider the unique smooth solution of the problem
and we denote by and the restrictions and , respectively, then we set
which is another smooth function on with zero integral. Then, we have
and such quantity turns out to be a norm equivalent to the one given by the Gagliardo seminorm on above (this is related to the theory of trace spaces for which we refer to [3,25]), see [41]. Hence, it induces the dual norm
for every smooth function . By polarization, we have the –scalar product between a pair of smooth functions with zero integral,
This scalar product, extended to the whole space , makes it a Hilbert space (see [27]), hence, by Riesz representation theorem, there exists a function such that, for every smooth function , there holds
by Theorem 2.6, where is the potential introduced in (2.3) and is the mean curvature of .
Then, by the fundamental lemma of calculus of variations, we conclude
for a constant , that is, recalling the definition of in problem (3.1) and of the operator above,
It clearly follows that the outer normal velocity of the moving boundaries is minus the –gradient of the volume–constrained functional .
We deal now with the surface diffusion flow.
Definition 3.4. Let be a smooth set. We say that a smooth flow , for , with , is a surface diffusion flow starting from if the outer normal velocity of the moving boundaries is given by
(3.6) |
where is the (rough) Laplacian associated to the hypersurface , with the Riemannian metric induced by (that is, by ).
Such flow was first proposed by Mullins in [45] to study thermal grooving in material sciences and first analyzed mathematically more in detail in [17]. In particular, in the physically relevant case of three–dimensional space, it describes the evolution of interfaces between solid phases of a system, driven by surface diffusion of atoms under the action of a chemical potential (see for instance [34]).
With the same argument used for the modified Mullins–Sekerka flow, representing the smooth hypersurfaces in with a family of smooth embeddings , we can describe the flow as
and also simply as
(3.7) |
Remark 3.5. This is actually the more standard way to define the surface diffusion flow, in the more general situation of smooth and possibly immersed–only hypersurfaces (usually in ), without being the boundary of any set.
By means of Eq (2.10), the system (3.7) can be rewritten as
(3.8) |
and it can be seen that it is a fourth order, quasilinear and degenerate, parabolic system of PDEs. Indeed, it is quasilinear, as the coefficients (as second order partial differential operator) of the Laplacian associated to the induced metrics on the evolving hypersurfaces, that is,
depend on the first order derivatives of , as (and the coefficient of on the third order derivatives). Moreover, the operator at the right hand side of system (3.7) is degenerate, as its symbol (the symbol of the linearized operator) admits zero eigenvalues due to the invariance of the Laplacian by diffeomorphisms.
Arguing as in computation (3.5), using the Eq (3.6) in place of (3.2), it can be seen that also the surface diffusion flow of boundaries of sets is volume–preserving. Moreover, analogously to the modified Mullins–Sekerka flow (see the discussion above), it does not preserve convexity (see [36]), nor the embeddedness (in the "stand–alone" formulation of motion of hypersurfaces, as in formula (3.7), see [28]), indeed it also does not have a "comparison principle", while it is invariant by isometries of , reparametrizations and tangential perturbations of the velocity of the motion. In addition, it can be regarded as the –gradient flow of the volume–constrained Area functional, in the following sense (see [27], for instance). For a smooth set , we let the space to be the dual of with the norm and the pairing between and simply being the integral of the product of the functions on .
Then, it follows easily that the norm of a smooth function is given by
and, by polarization, we have the –scalar product between a pair of smooth functions with zero integral,
integrating by parts.
This scalar product, extended to the whole space , make it a Hilbert space, hence, by Riesz representation theorem, there exists a function such that, for every smooth function , there holds
by Theorem 2.6 (with ).
Then, by the fundamental lemma of calculus of variations, we conclude
for a constant , that is,
It clearly follows that the outer normal velocity of the moving boundaries of a surface diffusion flow is minus the –gradient of the volume–constrained functional .
Remark 3.6. It is interesting to notice that the (unmodified, that is, with ) Mullins–Sekerka flow is the –gradient flow and the surface diffusion flow the –gradient flow of the Area functional on the boundary of the sets, under a volume constraint, while considering the unconstrained Area functional, its –gradient flow is the mean curvature flow.
It follows that, in a way, the unmodified Mullins–Sekerka flow, representing the moving hypersurfaces as of smooth embeddings , can be described as
showing its parabolic nature (differently by the surface diffusion flow, in this case the equation is nonlocal, due to the fractional Laplacian involved, even if the functional is still simply the Area, hence implying that the flow depends only on the hypersurface) – again quasilinear and degenerate – and suggesting the problem of analyzing (and eventually generalizing the existing results) the nonlocal evolutions of hypersurfaces given by the laws
when , arising from considering, as above, the –gradient of the Area functional on the boundary of the sets (under a volume constraint).
Up to our knowledge, these flows are not present in literature and it would be also interesting to compare them to the fractional mean curvature flows arising considering the gradient flows associated to the fractional Area functionals on the boundary of a set (in this case such functionals are "strongly" nonlocal), see [35,38] and references therein, for instance.
To state the short time existence and uniqueness results for the two flows, we give the following definition which is actually fundamental for the discussion of the global existence in the next section.
Definition 3.7. Given a smooth set and a tubular neighborhood of , as in formula (2.49), for any (recall the discussion in Subsection 2.2 about the notion of "closedness" of sets), we denote by , the class of all smooth sets such that and
(3.9) |
for some , with (hence, ). For every and , we also denote by the collection of sets such that .
The following existence/uniqueness theorem of classical solutions for the modified Mullins–Sekerka flow was proved by Escher and Simonett [19,20,21] and independently by Chen, Hong and Yi [8] (see also [18]). The original version deals with the flow in domains of , but it can be easily adapted to hold also when the ambient is the flat torus .
Theorem 3.8. Let be a smooth set and a tubular neighborhood of , as in formula , Then, for every and small enough, there exists such that if there exists a unique smooth modified Mullins–Sekerka flow with parameter , starting from , in the time interval .
We now state the analogous result (and also of dependence on the initial data) for the surface diffusion flow starting from a smooth hypersurface, proved by Escher, Mayer and Simonett in [17], which should be expected by the explicit parabolic nature of the system (3.7), as shown by the formula (3.8). As before, it deals with the evolution in the whole space of a generic hypersurface, even only immersed, hence possibly with self–intersections. It is then straightforward to adapt the same arguments to our case, when the ambient is the flat torus and the hypersurfaces are the boundaries of the sets , as in Definition 3.4, getting a (unique) surface diffusion flow in a positive time interval , for every initial smooth set .
Theorem 3.9. Let be a smooth and compact, immersed hypersurface. Then, there exists a unique smooth surface diffusion flow , starting from and solving system , for some maximal time of existence .
Moreover, such flow and the maximal time of existence depend continuously on the norm of the initial hypersurface.
As an easy consequence, we have the following proposition (analogous to Theorem 3.8), better suited for our setting.
Proposition 3.10. Let be a smooth set and a tubular neighborhood of , as in formula , Then, for every and small enough, there exists such that if there exists a unique smooth surface diffusion flow, starting from , in the time interval .
In the same paper [17], Escher, Mayer and Simonett also showed that if the initial set is in , where is a ball with the same volume and is small enough (that is, is –close to the ball ), then the smooth flow exists for every time and smoothly converges to a translate of the ball .
The analogous result for the (unmodified, that is, with ) Mullins–Sekerka flow, was proved by Escher and Simonett in [22] (moving by their previous work [20]), generalizing to any dimension the two dimensional case shown by Chen in [7].
The next section will be devoted to present the generalization by Acerbi, Fusco, Julin and Morini in [1] (in dimensions two and three) of this stability result for the surface diffusion and modified Mullins–Sekerka flow, to every strictly stable critical set (as it is every ball for the Area functional under a volume constraint, by direct check – see the last section).
We conclude mentioning another interesting result by Elliott and Garcke [15] (which is not present in literature for the modified Mullins–Sekerka flow, up to our knowledge) is that if the initial curve in of the surface diffusion flow is closed to a circle, then the flow exists for all times and converges, up to translations, to a circle in the plane with the same volume. This is clearly related to the fact that the unique bounded strictly stable critical sets for the Area functional under a volume constraint in the plane are the disks (see the last section).
In this section we show the proof by Acerbi, Fusco, Julin and Morini in [1], in dimensions two and three of the toric ambient, that if the "initial" sets is "close enough" to a strictly stable critical set of the respectively relative functional, then the surface diffusion and the modified Mullins–Sekerka flow exist for all times and smoothly converge to a translate of . Heuristically, this shows that a strictly stable critical set is in a way like the equilibrium configuration of a system at the bottom of a potential well "attracting" the close enough smooth sets.
We will deal here with the (more difficult) case of dimension three. When the dimension is two, the "exponents" in the functional spaces involved in the estimates (in particular the ones in the interpolation inequalities, which are very dimension–dependent) change but the same proof still works (roughly speaking, we have the necessary "compactness" of the sequences of hypersurfaces – see Lemma 4.5 and 4.18), modifying suitably the statements. If the dimension of the toric ambient is larger than three, the analogous (mostly, interpolation) estimates are too weak to conclude and this proof does not work. It is indeed a challenging open problem to extend these results to such higher dimensions.
For both flows, we will have a subsection with the necessary technical lemmas and then one with the proof of the main theorem. Moreover, for the modified Mullins–Sekerka flow, we also briefly discuss the "Neumann case", in Subsection 4.3.
In order to simplify the notation, for a smooth set we will write and in place of and , for the function uniquely defined by problem (3.1). Moreover, we will also denote with the smooth potential function associated to by formula (2.3).
We start with the following lemma holding in all dimensions.
Lemma 4.1 (Energy identities). Let be a modified Mullins–Sekerka flow as in Definition . Then, the following identities hold:
(4.1) |
and
(4.2) |
where is the quadratic form defined in formula (2.41).
Proof. Let the smooth family of maps describing the flow as in formula (3.4). By formula (2.15), where is the smooth (velocity) vector field along , hence (as usual is the outer normal unit vector of ), following the computation in the proof of Theorem (2.6), we have
where the last equality follows integrating by parts, as is harmonic in . This establishes relation (4.1).
In order to get identity (4.2), we compute
(4.3) |
where we interchanged time and space derivatives and we applied the divergence theorem, taking into account that is harmonic in .
Then, we need to compute on . We know that
on , hence, (totally) differentiating in time this equality, we get
that is,
where we used computation (2.33).
Therefore from Eqs (4.3) and (2.35) we get
Computing analogously for in and adding the two results, we get
where we integrated by parts the very first term of the right hand side, recalled Definition (2.41) and in the last step we used the identity
Hence, also Eq (4.2) is proved.
From now on, we restrict ourselves to the three–dimensional case, that is, we will consider smooth subsets of with boundaries which then are smooth embedded (–dimensional) surfaces. As we said at the beginning of the section, this is due to the dependence on the dimension of several of the estimates that follow.
In the estimates in the following series of lemmas, we will be interested in having uniform constants for the families , given a smooth set and a tubular neighborhood of as in formula (2.49), for any and . This is guaranteed if the constants in the Sobolev, Gagliardo–Nirenberg interpolation and Calderón–Zygmund inequalities, relative to all the smooth hypersurfaces boundaries of the sets , are uniform, as it is proved in detail in [13].
We remind that in all the inequalities, the constants may vary from one line to another.
The next lemma provides some boundary estimates for harmonic functions.
Lemma 4.2 (Boundary estimates for harmonic functions). Let be a smooth set and . Let with zero integral on and let be the (distributional) solution of
with zero integral on . Let and and assume that and are of class up to the boundary . Then, for every there exists a constant , such that:
for all , with depending also on .
Moreover, if , then for every there exists a constant , such that
Proof. We are not going to underline it every time, but it is easy to check that all the constants that will appear in the proof will depend with only on , , and sometimes , recalling the previous discussion about the "uniform" inequalities holding for the families of sets .
Recalling Remark 2.20, we have
It is well known that it is always possible to write where is smooth away from , one–periodic and in a neighborhood of , while is smooth and one–periodic. The conclusion then follows since for there holds
with .
We are going to adapt the proof of [37] to the periodic setting. First observe that since is harmonic in we have
(4.4) |
Moreover, there exist constants , and , depending only on , , , such that we may cover with balls , with every and
(4.5) |
for every that .
If then is a smooth function with compact support in such that in and , by integrating the function
in and using equality (4.4), we get
hence,
Using the Poincaré inequality on the torus (recall that has zero integral) and estimate (4.5), this inequality implies
Putting together all the above estimates and repeating the argument on , we get
The thesis then follows by observing that
Let us define
We want to show that
(4.6) |
By the decomposition recalled at the point , we have , where , for small enough and is smooth. Thus, by a standard partition of unity argument we may localize the estimate and reduce to show that if and is a bounded domain setting and
for every , where is the "upper" normal to the graph , then is well defined at every and
In order to show this we observe that we may write
where we used the fact that
and then that
Therefore,
Thus, inequality (4.6) follows from a standard convolution estimate.
For we have
hence, for there holds
We claim that
(4.7) |
for every , then the result follows from inequality (4.6) and this limit, together with the analogous identity for .
To show equality (4.7) we first observe that
(4.8) |
(4.9) |
Indeed, using Definition (2.3), we have
then,
which clearly implies Eq (4.8). Equality (4.9) instead follows by an approximation argument, after decomposing the Green function as at the beginning of the proof of point , , with in a neighborhood of and a smooth function.
Therefore, we may write, for and (remind that is the outer unit normal vector, hence ),
(4.10) |
by equality (4.8).
Let us now prove that
observing that since is of class then for sufficiently small we have
(4.11) |
Then, in view of the decomposition of above, it is enough show that
which follows from the dominated convergence theorem, after observing that due to the –Hölder continuity of and to inequality (4.11), the absolute value of both integrands can be estimated from above by for some constant .
Arguing analogously, we also get
Then, letting in equality (4.10), for every , we obtain
where we used equality (4.9), then limit (4.7) holds and the thesis follows.
Fixed and , as before, due to the properties of the Green's function, it is sufficient to establish the statement for the function
For , we have
In turn, by an elementary inequality, we have
thus, by Hölder inequality we have
where we set
with .
For the second part of the lemma, we start by observing that
If we have, by Gagliardo–Nirenberg interpolation inequalities (see [5,Theorem 3.70]),
Therefore, by combining the two previous inequalities we get that, for , there holds
Hence, the thesis follows once we show
To this aim, let us fix and with a little abuse of notation denote its harmonic extension to still by . Then, by integrating by parts twice and by point , we get
Therefore,
and we are done. For any smooth set , the fractional Sobolev space , usually obtained via local charts and partitions of unity, has an equivalent definition considering directly the Gagliardo –seminorm of a function , for , as follows
and setting (we refer to [3,14,48,61] for details). As it is customary, we set and .
Then, it can be shown that for all the sets , given a smooth set and a tubular neighborhood of as in formula (2.49), for any and , the constants giving the equivalence between this norm above and the "standard" norm of can be chosen to be uniform, independent of . Moreover, as for the "usual" (with integer order) Sobolev spaces, all the constants in the embeddings of the fractional Sobolev spaces are also uniform for this family. This is related to the possibility, due to the closeness in and the graph representation, of "localizing" and using partitions of unity "in a single common way" for all the smooth hypersurfaces boundaries of the sets , see [13] for details.
Then, we have the following technical lemma.
Lemma 4.3. Let be a smooth set and . For every , there exists a constant such that if and , then
Proof. We estimate with Hölder inequality, noticing that , as , hence there exists such that ,
Hence the thesis follows noticing that all the constants above depend only on , , and , by the previous discussion, before the lemma.
As a corollary we have the following estimate.
Lemma 4.4. Let be a smooth set and . Then, for small enough, there holds
where is the mean curvature of and the function is defined by formula (3.9).
Proof. By a standard localization/partition of unity/straightening argument, we may reduce ourselves to the case where the function is defined in a disk and . Fixed a smooth cut–off function with compact support in and equal to one on a smaller disk , we have
(4.12) |
where the remainder term is a smooth Lipschitz function. Then, using Lemma 4.3 with and recalling that , we estimate
We now use the fact that, by a simple integration by part argument, if is a smooth function with compact support in , there holds
hence,
then, if is small enough, we have
(4.13) |
as
(4.14) |
where we used the continuous embedding of in (see for instance Theorem in [48], with , and ).
By the Calderón–Zygmund estimates (holding uniformly for every hypersurface , with , see [13]),
(4.15) |
and the expression of the mean curvature
we obtain
(4.16) |
Hence, possibly choosing a smaller , we conclude
(4.17) |
again by inequality (4.14).
Thus, by estimate (4.15), we get
(4.18) |
and using this inequality in estimate (4.13),
hence,
The inequality in the statement of the lemma then easily follows by this inequality, estimate (4.18) and , with a standard covering argument.
We are now ready to prove the last lemma of this section.
Lemma 4.5 (Compactness). Let be a smooth set and a sequence of smooth sets such that
where are the functions associated to by problem (3.1).
Then, if and is small enough, there exists a smooth set such that, up to a (non relabeled) subsequence, in for all (recall the definition of convergence of sets at the beginning of Subsection 2.2).
Moreover, if
then is critical for the volume–constrained nonlocal Area functional and the convergence is in .
Proof. Throughout all the proof we write , , and instead of , , and , respectively. Moreover, we denote by and we set and .
First, we recall that
(4.19) |
by standard elliptic estimates. We want to show that
(4.20) |
To this aim, we recall that for every constant
then,
The above equality vanishes if and only if , hence,
and inequality (4.20) follows by the definition of and the observation on the Gagliardo seminorms just before Lemma 4.3.
Then, from the trace inequality (see [23]), which holds with a "uniform" constant , for all the sets (see [13]), we obtain
(4.21) |
with a constant independent of .
We claim now that
(4.22) |
To see this note that by the uniform –bounds on , we may find a fixed solid cylinder of the form , with a ball centered at the origin and functions , with
(4.23) |
such that with respect to a suitable coordinate frame (depending on ). Then,
(4.24) |
where is the canonical (standard) measure on the circle .
Hence, recalling the uniform bound (4.23) and the fact that are equibounded thanks to inequalities (4.19) and (4.21), we get that are also equibounded (by a standard "localization" argument, "uniformly" applied to all the hypersurfaces ). Therefore, the claim (4.22) follows.
By applying the Sobolev embedding theorem on each connected component of , we have that
for a constant independent of .
Now, by means of Calderón–Zygmund estimates, it is possible to show (see [13]) that there exists a constant depending only on , , and such that for every , there holds
(4.25) |
Then, if we write
we have , for all . Thus, by the Sobolev compact embedding , up to a subsequence (not relabeled), there exists a set such that
for all and .
From estimate (4.22) and Lemma 4.4 (possibly choosing a smaller ), we have then that the functions are bounded in . Hence, possibly passing to another subsequence (again not relabeled), we conclude that in for every , by the Sobolev compact embedding (see for instance Theorem in [48], with , and , applied to ).
If moreover we have
then, the above arguments yield the existence of and a subsequence (not relabeled) such that in . In turn,
in , where is the mean curvature of . Hence is critical.
To conclude the proof we then only need to show that converge to in . Fixed , arguing as in the proof of Lemma 4.4, we reduce ourselves to the case where the functions are defined on a disk , are bounded in , converge in for all to and . Then, fixed a smooth cut–off function with compact support in and equal to one on a smaller disk , we have
where is a smooth Lipschitz function. Then, using Lemma 4.3 with , an argument similar to the one in the proof of Lemma 4.4 shows that
Using Lemma 4.3 again to estimate with the seminorm on the left hand side of the previous inequality and arguing again as in the proof of Lemma 4.4, we finally get
hence,
from which the conclusion follows, by the first part of the lemma with and a standard covering argument.
We are ready to prove the long time existence/stability result.
Theorem 4.6. Let be a smooth strictly stable critical set for the nonlocal Area functional under a volume constraint and (with ) a tubular neighborhood of , as in formula . For every there exists such that, if is a smooth set in satisfying and
where is the function relative to as in problem , then the unique smooth solution of the modified Mullins–Sekerka flow (with parameter ) starting from , given by Theorem 3.8, is defined for all . Moreover, exponentially fast in as (recall the definition of convergence of sets at the beginning of Subsection 2.2), for some , with the meaning that the functions representing as "normal graphs" on , that is,
satisfy
for every , for some positive constants and .
Remark 4.7. With some extra effort, arguing as in the proof of Theorem 5.1 in [24] (last part – see also Theorem 4.4 in the same paper), it can be shown that the convergence of is actually smooth (see also Remark 4.20). Indeed, by means of standard parabolic estimates and interpolation (and Sobolev embeddings) the exponential decay in implies analogous estimates in , for every ,
for every , for some positive constants and .
Remark 4.8. We already said that the property of a set to belong to is a "closedness" condition in of and and in of their boundaries. The extra condition in the theorem on the –smallness of the gradient of (see the second part of Lemma 4.5 and its proof) implies that the quantity on is "close" to be constant, as it is the analogous quantity for the set (or actually for any critical set). Notice that this is a second order condition for the boundary of , in addition to the first order one .
Proof of Theorem 4.6. Throughout the whole proof will denote a constant depending only on , and , whose value may vary from line to line.
Assume that the modified Mullins–Sekerka flow is defined for in the maximal time interval , where and let the moving boundaries be represented as "normal graphs" on as
for some smooth functions . As before we set , and .
We recall that, by Theorem 3.8, for every , the flow is defined in the time interval , with .
We show the theorem for the smooth sets satisfying
for some positive constants , then we get the thesis by setting .
For any set we introduce the following quantity
(4.26) |
where is the signed distance function defined in formula (2.50). We observe that
for a constant depending only on and, as ,
Moreover,
where the smooth diffeomorphism defined in formula (2.53) and its Jacobian. As we already said, the constant depends only on and . This clearly implies
(4.27) |
Hence, by this discussion, the initial smooth set satisfies (having chosen ).
By rereading the proof of Lemma 4.5, it follows that for small enough, if and
then,
(4.28) |
where is a positive nondecreasing function (defined on ) such that as . Hence,
(4.29) |
for a function with the same properties of . Both and only depend on and , for small enough.
We split the proof of the theorem into steps.
Step 1 (Stopping–time). Let be the maximal time such that
(4.30) |
for all . Hence,
for all , as in formula (4.28). Note that such a maximal time is clearly positive, by the hypotheses on .
We claim that by taking small enough, we have .
Step 2 (Estimate of the translational component of the flow). We want to see that there exists a small constant such that
(4.31) |
where is defined by formula (2.47).
If is small enough, clearly there exists a constant such that, for every , we have , holding . It is then easy to show that the vector realizing such minimum is unique and satisfies
(4.32) |
where is a function –orthogonal (with respect to the measure on ) to the vector subspace of spanned by , with , where is the orthonormal basis of given by Remark 2.26. Moreover, the inequality
(4.33) |
holds, with a constant depending only on , and .
We now argue by contradiction, assuming . First, by formula (2.6) and the translation invariance of the functional , we have
It follows that, by multiplying equality (4.32) by , with and integrating over , we get
Note that in the second and the third equality above we have used the fact that and have zero integral on .
By the trace inequality (see [23]), we have
(4.34) |
hence, by the previous estimate, we conclude
(4.35) |
Let us denote with the harmonic extension of to , we then have
(4.36) |
where the first inequality comes by standard elliptic estimates (holding with a constant , see [13] for details), the second is trivial and the last one follows by inequalities (4.29) and (4.33).
Thus, by equality (4.32) and estimates (4.35) and (4.36), we get
If then is chosen so small that in the last inequality, then we have a contradiction with equality (4.32) and the fact that , as they imply (by –orthogonality) that
All this argument shows that for such a choice of condition (4.31) holds.
By Propositions 2.35 and 2.36, there exist positive constants and with the following properties: for any set such that , there holds
for all such that and if is critical, with , then
(4.37) |
for a suitable vector . We then assume that are small enough such that
(4.38) |
where is the function introduced in formula (4.28).
Step 3 (The stopping time is equal to the maximal time ). We show now that, by taking smaller if needed, we have .
By the previous point and the suitable choice of made in its final part, formula (4.31) holds, hence we have
In turn, by Lemma 4.1 we may estimate
for every .
It is now easy to see that
then, by point of Lemma 4.2, we estimate the last term as
thus, the last estimate in the statement of Lemma 4.2 implies
Therefore, combining the last three estimates, we get
(4.39) |
for every , where in the last inequality we used the trace inequality (4.34)
possibly choosing a smaller such that .
This argument clearly says that the quantity is nonincreasing in time, hence, if are small enough, the inequality is preserved during the flow. If we assume by contradiction that , then it must happen that or . Before showing that this is not possible, we prove that actually the quantity decreases (non increases) exponentially.
Computing as in the previous step,
where we used again the trace inequality (4.34). Then,
and combining this inequality with estimate (4.39), we obtain
for every and for a suitable constant . Integrating this differential inequality, we get
(4.40) |
for every .
Then, we assume that or . Recalling formula (4.26) and denoting by the velocity field of the flow (see Definition 3.1 and the subsequent discussion), we compute
where denotes the harmonic extension of to . Note that, by standard elliptic estimates and the properties of the signed distance function , we have
then, by the previous equality and formula (4.40), we get
for every . By integrating this differential inequality over and recalling estimate (4.27), we get
(4.41) |
as , provided that are chosen suitably small. This shows that cannot happen if we chose .
By arguing as in Lemma 4.5 (keeping into account inequality (4.30) and formula (4.28)), we can see that the –estimate (4.41) implies a –bound on with a constant going to zero, keeping fixed , as , hence, by estimate (4.40), as . Then, by Sobolev embeddings, the same holds for , hence, if is small enough, we have a contradiction with .
Thus, and
(4.42) |
for every , by choosing small enough.
Step 4 (Long time existence). We now show that, by taking smaller if needed, we have , that is, the flow exists for all times.
We assume by contradiction that and we recall that, by estimate (4.39) and the fact that , we have
for all . Integrating this differential inequality over the interval
where is given by Theorem 3.8, as we said at the beginning of the proof, we obtain
where the last inequality follows from estimate (4.42). Thus, by the mean value theorem there exists such that
Note that for any smooth set , we have , for some "absolute" constant and that is constant, then, since embeds into for all , by Lemma 4.2, we in turn infer that
where and stand for the –Hölder seminorms on and , respectively and remind that are the potentials, defined by formula (2.1), associated to and .
By means of Schauder estimates (as Calderón–Zygmund inequality implied estimate (4.25)), it is possible to show (see [13]) that there exists a constant depending only on , , and such that for every , choosing even smaller , there holds
Hence, by the above discussion, we can conclude that . Therefore, the maximal time of existence of the classical solution starting from is at least , which means that the flow can be continued beyond , which is a contradiction.
Step 5 (Convergence, up to subsequences, to a translate of ). Let , then, by estimates (4.42), the sets satisfy the hypotheses of Lemma 4.5, hence, up to a (not relabeled) subsequence we have that there exists a critical set such that in . Due to formulas (4.28) and (4.38) we also have and for some (small) (equality (4.37)).
Step 6 (Exponential convergence of the full sequence). Consider now
The very same calculations performed in Step show that
for all , moreover, by means of the previous step, it follows . In turn, by integrating this differential inequality and writing
we get
(4.43) |
Since by the previous steps is bounded, we infer from this inequality and interpolation estimates that also decays exponentially for all . Then, setting , we have, by estimates (4.43) and (4.27) (and standard elliptic estimates),
(4.44) |
for all . Denoting the average of on by , as by estimates (4.34) and (4.40) (recalling the argument to show inequality (4.20)), we have that
It follows, taking into account inequality (4.44), that
(4.45) |
exponentially fast, as , where and stand for the averages of on and of on , respectively.
Since (up to a subsequence) in , it is easy to check that which decays exponentially, therefore, thanks to limit (4.45), we have
exponentially fast.
The conclusion then follows arguing as at the end of Step .
Let be a smooth bounded open subset of . As before we consider the nonlocal Area functional
for every with , where is a real parameter and is the potential defined as follows, similarly to problem (2.3),
with , and the outer unit normal to .
As in formula (2.5), we have
where is the (distributional) solution of
for every .
Note that, unlike the "periodic" case (when the ambient is the torus ), the functional is not translation invariant, therefore several arguments simplify. The calculus of the first and second variations of , under a volume constraint, is exactly the same as for , then we say that a smooth set , with , is a critical set, if it satisfies the Euler–Lagrange equation
for a constant , instead, since is not translation invariant, the spaces , , and the decomposition (2.43) are no longer needed and, defining the same quadratic form as in formula (2.41), we say that a smooth critical set is strictly stable if
Naturally, is a local minimizer if there exists a such that
for all , , and . Then, as in the periodic case, we have a local minimality result with respect to small –perturbations. Precisely, the following (cleaner) counterpart to Theorem 2.29 holds (see also [39]).
Theorem 4.9. Let and a smooth strictly stable critical set for the nonlocal Area functional (under a volume constraint) with a tubular neighborhood of as in formula . Then there exist constants such that
for all smooth sets such that , , and
for a smooth with .
As a consequence, is a –local minimizer of (as defined above). Moreover, if is –close enough to and , then , that is, is locally the unique –local minimizer.
Sketch of the proof. Following the line of proof of Theorem 2.29, since the functional is not translation invariant we do not need Lemma 2.34 and inequality (2.83), proved in Step 2 of the proof of such theorem, simplifies to
where is the constant defined in formula (2.82). The proof of this inequality then goes exactly as there.
Coming to Step 3 of the proof of Theorem 2.29, we do not need inequality (2.86), thus we do not need to replace by a suitable translated set . Instead, we only need to observe that inequality (2.90) is still satisfied. The rest of the proof remains unchanged.
The short time existence and uniqueness Theorem 3.8, proved in [18] in any dimension, holds also in the "Neumann case" for the modified Mullins–Sekerka flow with parameter , obtained (as in Definition 3.2) by letting the outer normal velocity of the moving boundaries given by
where and is the unique solution in of the problem
with the potential defined above and, as before, is the jump of the outer normal derivative of on .
Then, we conclude by stating the following analogue of Theorem 4.6 (taking into account Remark 4.7).
Theorem 4.10. Let be an open smooth subset of and let be a smooth strictly stable critical set for the nonlocal Area functional under a volume constraint, with and (with ) a tubular neighborhood of , as in formula . Then, for every there exists such that, if is a smooth set in satisfying and
where is the function relative to as in problem (with in place of ), then, the unique smooth solution to the Mullins–Sekerka flow (with parameter ) starting from , given by Theorem 3.8, is defined for all . Moreover, exponentially fast in , as , for every , with the meaning that the functions representing as "normal graphs" on , that is,
satisfy, for every ,
for every , for some positive constants and .
The proof of this result is similar to the one of Theorem 4.6 and actually it is simpler since we do not need the argument used in Step 2 of such proof, where we controlled the translational component of the flow. Note also that in the statement of Proposition 2.35, in this case, inequality (2.93) holds for all . Finally, observe that under the hypotheses of Proposition 2.36 we may actually conclude that , that is, there are no other critical sets close to .
As for the modified Mullins–Sekerka flow, we start with the technical lemmas for the global existence result.
Lemma 4.11 (Energy identities). Let be a surface diffusion flow. Then, the following identities hold:
(4.46) |
and
(4.47) |
where is the quadratic form defined in formula (with ).
Proof. Let the smooth family of maps describing the flow as in formula (3.7). By formula (2.15), where is the smooth (velocity) vector field along , hence (as usual is the outer normal unit vector of ), following computation (2.16), we have
where the last equality follows integrating by parts. This establishes relation (4.46).
In order to get relation (4.47) we also need the time derivatives of the evolving metric and of the mean curvature of , that we already computed in formulas (2.13), (2.32) and (2.33) (where the function in this case is equal to and ), that is,
Then, we compute
which is formula (4.47), recalling the definition of in formula (2.41).
From now on, as before due to the dimension–dependence of the estimates that follow, we restrict ourselves to the three–dimensional case.
The following lemma is an easy consequence of Theorem 3.70 in [5], with , , and , taking into account the previous discussion.
Lemma 4.12 (Interpolation on boundaries). Let be a smooth set. In the previous notations, for every there exists a constant such that for every set and , we have
with .
Moreover, the following Poincaré inequality holds
where , if belongs to a connected component of .
Then, we have the following mixed "analytic–geometric" estimate.
Lemma 4.13 (–estimates on boundaries). Let be a smooth set. Then there exists a constant such that if and with , then and
Proof. We first claim that the following inequality holds,
(4.48) |
Indeed, if we integrate by parts the left–hand side, we obtain (the Hessian of a function is symmetric)
Hence, interchanging the covariant derivatives and integrating by parts, we get
thus, inequality (4.48) holds (in the last passage we applied Cauchy–Schwarz inequality and the well known relation , then ).
We now estimate the last term in formula (4.48) by means of Lemma 4.12 (which is easily extended to vector valued functions ) with and :
Hence, expanding the product on the last line, using Peter–Paul (Young) inequality on the first term of such expansion and "adsorbing" in the left hand side of inequality (4.48) the small fraction of the term that then appears, we obtain
(4.49) |
By the fact that has zero average on each connected component of , there holds
where we used Lemma 4.12 again, hence,
(4.50) |
Thus, from inequality (4.49), we deduce
(4.51) |
Now, by means of Calderón–Zygmund estimates, it is possible to show (see [13]) that there exists a constant depending only on , , and such that for every , there holds
(4.52) |
Then, since it is easy to check that also all the other constant in the previous inequalities (and the ones coming from Lemma 4.12 also) depend only on , , and , if , substituting this estimate, with , in formula (4.51), the thesis of the lemma follows.
The following lemma provides a crucial "geometric interpolation" that will be needed in the proof of the main theorem.
Lemma 4.14 (Geometric interpolation). Let be a smooth set. Then there exists a constant such that the following estimates holds
for every .
Proof. First, by a standard application of Hölder inequality, we have
Then, using the Poincaré inequality stated in Lemma 4.12 and the fact that has zero average on each connected component of , we get
Now, we use Hölder inequality again
and we apply Lemma 4.12 with ,
Combining all these inequalities, we conclude
By Lemma 4.13 and estimate (4.50), with in place of , the right–hand side of the previous inequality can be bounded from above by
Hence, using again Poincaré inequality and estimate (4.52) with , we have
and
Finally, using this relations and Hölder inequality, we obtain the thesis
We now remind that since can be disconnected (as in the case of lamellae), the Poincaré inequality could fail for . However, if is sufficiently close to a stable critical set then it is true for the mean curvature of .
Lemma 4.15 (Geometric Poincaré inequality). Fixed and a smooth strictly stable critical set , let be the constant provided by Proposition 2.35, with . Then, for small enough, there exists a constant such that
(4.53) |
for every set such that with
for a smooth function with .
Proof. Since
there holds
for all . Choosing , we may then apply Proposition 2.35 with and , obtaining
The following lemma is straightforward.
Lemma 4.16. Let be a smooth set. If and , then
for a constant independent of .
Proof. We estimate with Cauchy–Schwarz inequality,
hence the thesis follows.
As a consequence, we prove the following result.
Lemma 4.17. Let be a smooth set and . Then, for small enough, there holds
where is the mean curvature of (the function is defined by formula ).
Proof. As we do in Lemma 4.4, by a standard localization/partition of unity/straightening argument, we may reduce ourselves to the case where the function is defined in a disk and . Fixed a smooth cut–off function with compact support in and equal to one on a smaller disk , we have again relation (4.12) (see also [42]).
Then, using Lemma 4.16 and recalling that , we estimate
We now use the fact that, by a simple integration by part argument, if is a smooth function with compact support in , there holds
hence,
then, if is small enough, we have
(4.54) |
as
(4.55) |
by Theorem 3.70 in [5].
By the Calderón–Zygmund estimates (holding uniformly for every hypersurface , with , see [13]), we have again the inequality (4.15) and the most useful estimation (4.16).
Hence, possibly choosing a smaller , we conclude (as in inequality (4.17))
again by inequality (4.55).
Thus, by estimate (4.15), we get
(4.56) |
and using this inequality in estimate (4.54),
hence,
The inequality in the statement of the lemma then easily follows by this inequality, estimate (4.56) and , with a standard covering argument.
Now, we state a compactness result whose proof is very close in spirit to the proof of Lemma 4.5, however we present it explicitly in order to show how the lemmas above come differently into play.
Lemma 4.18 (Compactness). Let be a smooth set and a sequence of smooth sets such that
Then, if and is small enough, there exists a smooth set such that, up to a (non relabeled) subsequence, in for all .
Moreover, if inequality holds for every set with a constant independent of and
then is critical for the volume–constrained Area functional and the convergence is in .
Proof. We first claim that
(4.57) |
We set , then, by the "geometric" Poincaré inequality of Lemma 4.15, which holds with a "uniform" constant , for all the sets (see [13]), if is small enough, we have
with a constant independent of .
Then, we note that, as in Lemma 4.5, by the uniform –bounds on , we may find a solid cylinder of the form , with a ball centered at the origin and functions , with
(4.58) |
such that with respect to a suitable coordinate frame (depending on ). Hence, recalling the formula (4.24), the uniform bound (4.58) and the fact that are equibounded, we get that are also equibounded (by a standard "localization" argument, "uniformly" applied to all the hypersurfaces ). Therefore, the claim (4.57) follows.
By applying the Sobolev embedding theorem on each connected component of , we have that
for a constant independent of .
Now, as before, we obtain
for every with a uniform constant . Then, if we write
we have , for all .
Thus, by the Sobolev compact embedding , up to a subsequence (not relabeled), there exists a set such that
for all and .
From estimate (4.57) and Lemma 4.17 (possibly choosing a smaller ), we have then that the functions are bounded in . Hence, possibly passing to another subsequence (again not relabeled), we conclude that in for every , by the Sobolev compact embeddings.
For the second part of the lemma, we first observe that if
then there exists and a subsequence (not relabeled) such that
in , where is the mean curvature of . Hence is critical.
To conclude the proof we only need to show that converge to in .
Fixed , arguing as in the proof of Lemma 4.17, we reduce ourselves to the case where the functions are defined on a disk , are bounded in , converge in for all to and . Then, fixed a smooth cut–off function with compact support in and equal to one on a smaller disk , we have
where is a smooth Lipschitz function.
Then, using Lemma 4.16, an argument similar to the one of the proof of Lemma 4.17 shows that
Being constant, that is , by using Lemma 4.16 again and arguing as in the proof of Lemma 4.17, we finally get
hence,
from which the conclusion follows, by the first part of the lemma and a standard covering argument.
We now show the global existence result for the surface diffusion flow, whose proof is very similar to the one of Theorem 4.6. However, in order to make it clear, we present it in a detailed way.
Theorem 4.19. Let be a strictly stable critical set for the Area functional under a volume constraint and let be a tubular neighborhood of , as in formula . For every there exists such that, if is a smooth set in satisfying and
then the unique smooth solution of the surface diffusion flow starting from , given by Proposition 3.10, is defined for all . Moreover, exponentially fast in as (recall the definition of convergence of sets in Subsection 2.2), for some , with the meaning that the functions representing as "normal graphs" on , that is,
satisfy
for every , for some positive constants and .
Remark 4.20. The convergence of is actually smooth, that is, for every , there holds
for every , for some positive constants and . This is a particular case of Theorem 5.1 in [24], proved by means of standard parabolic estimates and interpolation (and Sobolev embeddings), using the exponential decay in , analogously to the modified Mullins–Sekerka flow (Remark 4.7).
Remark 4.21. The extra condition in the theorem on the –smallness of the gradient of (see the second part of Lemma 4.18 and its proof) implies that the mean curvature of is "close" to be constant, as it is for the set or actually for any critical set (recall Remark 4.8).
Proof of Theorem 4.19. As in proof of Theorem 4.6, will denote a constant depending only on , and , whose value may vary from line to line.
Assume that the surface diffusion flow is defined for in the maximal time interval , where and let the moving boundaries be represented as "normal graphs" on as
for some smooth functions .
We recall that, by Proposition 3.10, for every , the flow is defined in the time interval , with .
As before, we show the theorem for the smooth sets satisfying
(4.59) |
for some positive constants , then we get the thesis by setting . For any set , we define quantity in (4.26) and by the same arguments we obtain estimation (4.27).
Hence, by this discussion, the initial smooth set satisfies (having chosen ).
By rereading the proof of Lemma 4.18, it follows that for small enough, if
and
then
(4.60) |
where is a positive nondecreasing function (defined on ) such that as . This clearly implies
for a function with the same properties of (also in this case, and only depend on and , for small enough). Moreover, thanks to Lemma 4.15, there exists such that, choosing small enough, in order that is small enough, we have
(4.61) |
where, as usual, is the average of over .
We again split the proof of the theorem into steps.
Step 1 (Stopping–time). Let be the maximal time such that
(4.62) |
for all . Hence,
(4.63) |
for all , as in formula (4.60).
As before, we claim that by taking small enough, we have .
Step 2 (Estimate of the translational component of the flow). We want to show that there exists a small constant such that
(4.64) |
where is defined by formula (2.47).
If is small enough, clearly there exists a constant such that, for every , we have , holding . It is then easy to show that the vector realizing such minimum is unique and satisfies
(4.65) |
where is chosen as in relation (4.32). Moreover, the inequality
(4.66) |
holds, with a constant depending only on , and .
We now argue by contradiction, assuming .
First we recall that has zero average. Then, setting , and recalling relation (4.61), we get
(4.67) |
Hence, we conclude
(4.68) |
Since, there holds
by multiplying relation (4.65) by , integrating over , and using inequality (4.68), we get
Recalling now estimate (4.66), as is orthogonal to , computing as in the first three lines of formula (4.67), we have
where in the last inequality we estimated with and we used inequality (4.63).
If then is chosen so small that in the last inequality, then we have a contradiction with equality (4.65) and the fact that , as they imply (by –orthogonality) that
All this argument shows that for such a choice of condition (4.64) holds.
Then, we can conclude as in Step of Theorem 4.6, by replacing the –norm on with the –norm on the same boundary.
Step 3 (The stopping time is equal to the maximal time ). We show now that, by taking smaller if needed, we have .
By the previous point and the suitable choice of made in its final part, formula (4.64) holds, hence we have
In turn, by Lemma 4.11 and 4.14 we may estimate
(4.69) |
for every , where in the last step we used relations (4.62) and (4.63).
Noticing that from formulas (4.67) and (4.68) it follows
keeping fixed and choosing a suitably small , we conclude
This argument clearly says that the quantity is nonincreasing in time, hence, if are small enough, the inequality is preserved during the flow. As before, if we assume by contradiction that , then it must happen that or .
Before showing that this is not possible, we prove that actually the quantity decreases (non increases) exponentially. Indeed, integrating the differential inequality above and recalling proprieties (4.59), we obtain
(4.70) |
for every . Then, we assume that or . Recalling formula (4.26) and denoting by the velocity field of the flow (see Definition 3.1 and the subsequent discussion), we compute
for all , where the last inequality clearly follows from inequality (4.70).
By integrating this differential inequality over and recalling estimate (4.27), we get
(4.71) |
as , provided that are chosen suitably small. This shows that cannot happen if we chose .
By arguing as in Lemma 4.18 (keeping into account inequality (4.62) and formula (4.60)), we can see that the –estimate (4.71) implies a –bound on with a constant going to zero, keeping fixed , as , hence, by estimate (4.70), as . Then, by Sobolev embeddings, the same holds for , hence, if is small enough, we have a contradiction with .
Thus, and
(4.72) |
for every , by choosing small enough.
Step 4 (Long time existence). We now show that, by taking smaller if needed, we have , that is, the flow exists for all times.
We assume by contradiction that and we notice that, by computation (4.69) and the fact that , we have
for all . Integrating this differential inequality over the interval , where is given by Proposition 3.10, as we said at the beginning of the proof, we obtain
where the last inequality follows from estimate (4.72). Thus, by the mean value theorem there exists such that
Then, by Lemma 4.13
where in the last inequality we also used the curvature bounds provided by formula (4.63). In turn, for large enough, we get
where stands for the –Hölder seminorm on and in the last inequality we used the previous estimate.
Then, arguing as in Step of Theorem 4.6, it is possible to show that flow exists beyond , which is a contradiction.
Step 5 (Convergence, up to subsequences, to a translate of ). Let , then, by estimates (4.72), the sets satisfy the hypotheses of Lemma 4.18, hence, up to a (not relabeled) subsequence we have that there exists a critical set such that in . Due to formulas (4.60) (and estimation (4.38), that also holds in this case) we have and for some (small) .
Step 6 (Exponential convergence of the full sequence). Consider now
The very same calculations performed in Step show that
for all , moreover, by means of the previous step, it follows . In turn, by integrating this differential inequality and writing
we get
Since by the previous steps is bounded, we infer from this inequality, Sobolev embeddings and standard interpolation estimates that also decays exponentially for .
Denoting the average of on by , as by estimates (4.67) and (4.70), we have that
It follows that
(4.73) |
exponentially fast, as , where stands for the average of on .
Since (up to a subsequence) in , it is easy to check that which decays exponentially, therefore, thanks to limit (4.73), we have
exponentially fast.
The conclusion then follows arguing as at the end of Step of Theorem 4.6.
In this final section, we are going to discuss the classes of smooth sets to which Theorems 4.6 and 4.19 can be applied, hence, "dynamically exponentially stable" for the modified Mullins–Sekerka and surface diffusion flow. Much is known for the stable and strictly stable critical sets (or of ) of the Area functional (hence, for the unmodified Mullins–Sekerka and surface diffusion flows), characterized by having constant mean curvature and satisfying respectively
for every and
for every , according to Definition 2.24. Instead, considerably less can be said for the "nonlocal case", relative to the modified (with ) Mullins–Sekerka flow, for which in the above formulas we need to consider analogously the positivity properties of form
on the critical sets (in or in domains of , with "Neumann conditions" at the boundary) satisfying on .
Concentrating for a while on the Area functional, we observe that it is easy to see that (by a dilation/contraction argument) any strictly stable smooth critical set must be connected, but actually, being the normal velocity of the surface diffusion flow at every point defined by the local quantity , it follows that Theorem 4.19 can be applied also to finite unions of boundaries of strictly stable critical sets (see [24] and the Figure 1 below). Moreover, by the very definition above, if in is composed by flat pieces, hence its second fundamental form is identically zero, the set is critical and stable and with a little effort, actually strictly stable. It is a little more difficult to show that any ball in any dimension is strictly stable (it is obviously a critical set), which is connected to the study of the eigenvalues of the Laplacian on the sphere , see [30,Theorem 5.4.1], for instance. The same then holds for all the "cylinders" , bounding after taking their quotient by the same equivalence relation defining , determined by the standard integer lattice of .
Notice that if , it follows that the only bounded strictly stable critical sets of the (in this case) Length functional in the plane are the disks and in they are the disks and the "strips" with straight borders. This is clearly in agreement with the two–dimensional convergence/stability result of Elliott and Garcke [15], mentioned at the end of Section 3.
In the three–dimensional case, a first classification of the smooth stable "periodic" critical sets for the volume–constrained Area functional, was given by Ros in [59], where it is shown that in the flat torus , they are balls, –tori, gyroids or lamellae.
Notice that, despite their name, the lamellae are (after taking the quotient) parallel planar –tori and the –tori are quotients of circular cylinders in . As we said, with the balls, these surfaces are actually strictly stable, while in [31,32,60] the authors established the strict stability of gyroids only in some cases. To give an example, we refer to [32] where Grosse–Brauckmann and Wohlgemuth showed the strictly stability of the gyroids that are fixed with respect to translations. We remind that the gyroids, that were discovered by the crystallographer Schoen in the (see [62]), are the unique non–trivial embedded members of the family of the Schwarz P surfaces and then conjugate to the D surfaces, that are the simplest and most well–known triply–periodic minimal surfaces (see [60]).
For the case , that is, for the nonlocal Area functional, a complete classification of the stable periodic structures is instead, up to now, still missing.
It is worth to mention what is shown in [2] about the minimizers of . The authors proved that if a horizontal strip is the unique global minimizer of the Area functional in , then it is also the unique global minimizer of the nonlocal Area functional under a volume constraint, provided that is sufficiently small. Precisely, the following result holds.
Theorem 5.1. Assume that is the unique, up to rigid motions, global minimizer of the Area functional, under a volume constraint. Then the same set is also the unique global minimizer of the nonlocal Area functional , provided that is sufficiently small.
This theorem then allows to conclude that the global minimizers are lamellae in several cases in low dimensions (two and three), for suitable parameters and volume constraint. Moreover, in [2], it is also shown that lamellae with multiple strips are local minimizers of the functional , if the number of strips is large enough.
Finally, we conclude by citing the papers [9,10,12,44,53,54,55,56,57,58] with related and partial results on the classification problem which is at the moment fully open.
We wish to thank Nicola Fusco for many discussions about his work on the topic and several suggestions. We also thank the anonymous referee for the careful reading and several suggestions.
The authors declare no conflict of interest.
[1] |
Schreiner, C., Henriksen, E.K. and Kirkeby Hansen, P.J., Climate Education: Empowering Today's Youth to Meet Tomorrow's Challenges. Studies in Science Education, 2005, 41(1): 3‒49. https://doi.org/10.1080/03057260508560213. doi: 10.1080/03057260508560213![]() |
[2] |
Kopcha, T.J., McGregor, J., Shin, S., Qian, Y., Choi, J., Hill, R., Mativo, J. and Choi, I., Developing an Integrative STEM Curriculum for Robotics Education Through Educational Design Research. Journal of Formative Design in Learning, 2017, 1(1): 31‒44. https://doi.org/10.1007/s41686-017-0005-1. doi: 10.1007/s41686-017-0005-1![]() |
[3] |
Barnes, J., Fakhrhosseini, S.M., Vasey, E., Park, C.H. and Jeon, M., Child-Robot Theater: Engaging Elementary Students in Informal STEAM Education Using Robots. IEEE Pervasive Computing, 2020, 19(1): 22‒31. https://doi.org/10.1109/MPRV.2019.2940181. doi: 10.1109/MPRV.2019.2940181![]() |
[4] |
Plaza, P., Sancristobal, E., Carro, G., Blazquez, M. and Castro, M., Scratch as Driver to Foster Interests for STEM and Educational Robotics. Revista Iberoamericana de Tecnologias del Aprendizaje, 2019, 14(4): 117‒126. https://doi.org/10.1109/RITA.2019.2950130. doi: 10.1109/RITA.2019.2950130![]() |
[5] |
Chiang, F. K., Liu, Y. Q., Feng, X., Zhuang, Y. and Sun, Y., Effects of the world robot Olympiad on the students who participate: a qualitative study. Interactive Learning Environments, 2020, (3): 1‒12. https://doi.org/10.1080/10494820.2020.1775097. doi: 10.1080/10494820.2020.1775097![]() |
[6] |
Eguchi, A., RoboCupJunior for promoting STEM education, 21st century skills, and technological advancement through robotics competition - ScienceDirect. Robotics and Autonomous Systems, 2016, 75: 692‒699. https://doi.org/10.1016/j.robot.2015.05.013. doi: 10.1016/j.robot.2015.05.013![]() |
[7] |
Menekse, M., Higashi, R., Schunn, C. D. and Baehr, E., The Role of Robotics Teams' Collaboration Quality on Team Performance in a Robotics Tournament. Journal of Engineering Education, 2017,106(4): 564‒584. https://doi.org/10.1002/jee.20178. doi: 10.1002/jee.20178![]() |
[8] |
Ten Huang, Y., Liu, E. Z.-F., Lin, C.H. and Liou, P.-Y., Developing and Validating a High School Version of the Robotics Motivated Strategies for Learning Questionnaire. International Journal of Online Pedagogy and Course Design, 2017, 7(2): 20‒34. https://doi.org/10.4018/ijopcd.2017040102. doi: 10.4018/ijopcd.2017040102![]() |
[9] |
Kaji, Y., Kawata, J. and Fujisawa, S., Educational Effect of Participation in Robot Competition on Experience-Based Learning. Journal of Robotics and Mechatronics, 2019, 31(3): 383‒390. https://doi.org/10.20965/jrm.2019.p0383. doi: 10.20965/jrm.2019.p0383![]() |
[10] |
Çetin, M. and Demircan, H. Ö., Empowering technology and engineering for STEM education through programming robots: a systematic literature review. Early Child Development and Care, 2020,190(9): 1323‒1335. https://doi.org/10.1080/03004430.2018.1534844. doi: 10.1080/03004430.2018.1534844![]() |
[11] |
Hendricks, C., Alemdar, M. and Ogletree, T., The Impact of Participation in VEX Robotics Competition on Middle and High School Students' Interest in Pursuing STEM Studies and STEM-related Careers. in 2012 ASEE Annual Conference & Exposition. San Antonio, 2012: 25.1312.1‒25.1312.16. https://doi.org/10.18260/1-2--22069. |
[12] |
Witherspoon, E.B., Schunn, C.D., Higashi, R.M. and Baehr, E.C., Gender, interest, and prior experience shape opportunities to learn programming in robotics competitions. International Journal of Stem Education, 2016, 3(1): 1‒12. https://doi.org/10.1186/s40594-016-0052-1. doi: 10.1186/s40594-016-0052-1![]() |
[13] |
Lin, C.H., Liu, E.Z.F. and Huang, Y.Y., Exploring parents' perceptions towards educational robots: Gender and socio‐economic differences. British Journal of Educational Technology, 2012, 43(1): E31‒E34. https://doi.org/10.1111/j.1467-8535.2011.01258.x. doi: 10.1111/j.1467-8535.2011.01258.x![]() |
[14] |
Chiang, F.K. and Feng, X., A pilot study of the World Robot Olympiad's Effect on the Participants. in BERA Conference 2018. Newcastle, 2018. |
1. | Dario Mazzoleni, Benedetta Pellacci, Calculus of variations and nonlinear analysis: advances and applications, 2023, 5, 2640-3501, 1, 10.3934/mine.2023059 | |
2. | Daniele De Gennaro, Antonia Diana, Andrea Kubin, Anna Kubin, Stability of the surface diffusion flow and volume-preserving mean curvature flow in the flat torus, 2024, 390, 0025-5831, 4429, 10.1007/s00208-024-02863-3 |
The 2018 and 2019 dimension themes mentioned by coaches