Citation: François Murat, Alessio Porretta. The ergodic limit for weak solutions of elliptic equations with Neumann boundary condition[J]. Mathematics in Engineering, 2021, 3(4): 1-20. doi: 10.3934/mine.2021031
[1] | Mouhamed Moustapha Fall, Veronica Felli, Alberto Ferrero, Alassane Niang . Asymptotic expansions and unique continuation at Dirichlet-Neumann boundary junctions for planar elliptic equations. Mathematics in Engineering, 2019, 1(1): 84-117. doi: 10.3934/Mine.2018.1.84 |
[2] | David Cruz-Uribe, Michael Penrod, Scott Rodney . Poincaré inequalities and Neumann problems for the variable exponent setting. Mathematics in Engineering, 2022, 4(5): 1-22. doi: 10.3934/mine.2022036 |
[3] | Youchan Kim, Seungjin Ryu, Pilsoo Shin . Approximation of elliptic and parabolic equations with Dirichlet boundary conditions. Mathematics in Engineering, 2023, 5(4): 1-43. doi: 10.3934/mine.2023079 |
[4] | Giovanni Cupini, Paolo Marcellini, Elvira Mascolo . Local boundedness of weak solutions to elliptic equations with $ p, q- $growth. Mathematics in Engineering, 2023, 5(3): 1-28. doi: 10.3934/mine.2023065 |
[5] | Huyuan Chen, Laurent Véron . Weak solutions of semilinear elliptic equations with Leray-Hardy potentials and measure data. Mathematics in Engineering, 2019, 1(3): 391-418. doi: 10.3934/mine.2019.3.391 |
[6] | Antonio Vitolo . Singular elliptic equations with directional diffusion. Mathematics in Engineering, 2021, 3(3): 1-16. doi: 10.3934/mine.2021027 |
[7] | Feida Jiang . Weak solutions of generated Jacobian equations. Mathematics in Engineering, 2023, 5(3): 1-20. doi: 10.3934/mine.2023064 |
[8] | Mónica Clapp, Filomena Pacella . Existence of nonradial positive and nodal solutions to a critical Neumann problem in a cone. Mathematics in Engineering, 2021, 3(3): 1-15. doi: 10.3934/mine.2021022 |
[9] | Italo Capuzzo Dolcetta . The weak maximum principle for degenerate elliptic equations: unbounded domains and systems. Mathematics in Engineering, 2020, 2(4): 772-786. doi: 10.3934/mine.2020036 |
[10] | Andrea Manzoni, Alfio Quarteroni, Sandro Salsa . A saddle point approach to an optimal boundary control problem for steady Navier-Stokes equations. Mathematics in Engineering, 2019, 1(2): 252-280. doi: 10.3934/mine.2019.2.252 |
dedicato a Italo, con stima e amicizia.
Let Ω be a bounded, sufficiently regular, connected domain in RN, N≥1, and let →n denote the outward normal unit vector on the boundary ∂Ω. It is well known (see e.g., [12,15]) that, if H(x,p) is a Lipschitz function for (x,p)∈Ω×RN, then there is a unique real number λ such that the elliptic problem
{−Δu+H(x,∇u)+λ=0in Ω,∇u⋅→n=0on ∂Ω | (1.1) |
admits a solution, and this solution is unique up to a constant. The simplest example of this type of problems occurs in the linear case, when H(x,∇u)=b(x)⋅∇u−f(x). If b(x) is a Lipschitz continuous, divergence free vector field, then λ is the average value of f; if b(x) is not divergence free, the uniqueness of λ is a consequence of Fredholm theory for linear operators (see e.g., [10]), in which case λ=∫Ωf(x)φ1dx, where φ1 is the first eigenfunction of the adjoint problem (normalized so that ∫Ωφ1=1).
If H is a nonlinear function, the existence and uniqueness of λ was proved in [15] assuming that H(x,p) satisfies fairly general structure conditions with respect to p and enough regularity with respect to x.
The real number λ appearing in this kind of problems is sometimes called additive eigenvalue and is definitively a critical value which plays a role in many different contexts. If H(x,p) is convex in p, then λ can be interpreted as the optimal value of an ergodic stochastic control problem; we refer the reader to [2,3,12,15] and especially to [1] for an extensive presentation of the ergodic stochastic control setting. In that framework, it is natural to obtain λ as the limit of εuε, where uε solves the approximating coercive problem
{−Δu+H(x,∇u)+εu=0in Ω,∇u⋅→n=0on ∂Ω. | (1.2) |
Indeed, from Bellman's principle, (1.2) is the equation solved by the value function of an infinite horizon stochastic control problem, where ε is the discount factor. The vanishing discount limit leads, through time averaging, to the ergodic control problem represented by (1.1), and λ=limε→0εuε. This interpretation awarded to λ the name of ergodic constant, and to the singular limit of solutions of (1.2), as ε→0, the name of ergodic limit.
The constant λ is also a critical value for the long time behavior of the evolution problem, since it represents the asymptotic speed of time-dependent solutions: typically, a solution of the evolution problem vt−Δv+H(x,∇v)=0 (with Neumann conditions) satisfies v(x,t)t→∞≃u(x)+λt for some stationary solution u of (1.1). Again, this is consistent with the ergodicity property of the underlying controlled stochastic trajectory, but of course the long time convergence itself does not need the convexity of H, at least for Lipschitz nonlinearities. Finally, λ also plays a crucial role in homogenization problems (in which context problem (1.1) is referred to as the cell problem), see [4].
A huge literature has been devoted so far to the existence and characterization of ergodic constants, as well as to the study of ergodic limits and of the long time behavior of evolution problems, at the point that it is impossible here to recall such a long list of contributions. Most papers concerned with the above issues treat the problem in the framework of viscosity solutions' theory, for both second and first order Hamilton-Jacobi-Bellman equations. This explains why this kind of results were proved under many different structure conditions on the "Hamiltonian" H(x,∇u) as well as on the second order operator, but mostly assuming a regular dependence with respect to x. This regularity is often required for verification theorems, whenever the application to stochastic control is the main motivation.
The purpose of this note is to give a prototype result of existence, uniqueness of the ergodic constant and a characterization of the ergodic limit under natural assumptions for elliptic operators in divergence form, replacing the L∞ framework (and most times continuity, needed for viscosity solutions) with the L2-setting which is natural for weak solutions in the Sobolev space H1(Ω). To be precise, we consider the elliptic problem
{−div(A(x)∇u)+λ=H(x,∇u)+χin Ω,A(x)∇u⋅→n=0on ∂Ω, | (1.3) |
where χ∈(H1(Ω))′ (the dual space of H1(Ω)), A(x) is a measurable matrix such that
A(x)∈L∞(Ω)N×N,A(x)ξ⋅ξ≥α|ξ|2, | (1.4) |
for some α>0, and where H(x,p) is a Carathéodory function (measurable in x, continuous in p) satisfying the following linear growth condition (this is for N≥3)
|H(x,p)|≤b(x)|p|+f(x),for some b(x)∈LN(Ω), f(x)∈L2NN+2(Ω), | (1.5) |
for almost every x∈Ω and for every p∈RN.
Let us stress that f(x) (and therefore H(x,0)) is not assumed to be bounded, and only belongs to L(2∗)′(Ω), where 2∗=2NN−2 is the Sobolev exponent for N≥3. The case N=1,2 is mentioned later, see Remark 2.4. Due to Sobolev embedding, the condition b∈LN(Ω) is the usual threshold for Lebesgue summability of drift terms in elliptic equations, see e.g., [11]; in fact b(x)|∇u|∈L2NN+2(Ω) whenever ∇u∈L2(Ω) and b(x)∈LN(Ω).
Here and below, we assume that Ω is a Lipschitz bounded and connected domain in RN; the Lipschitz regularity being just one possible condition which ensures that the Sobolev embedding (and the Poincaré-Wirtinger inequality) hold true.
Eventually, for the purpose of uniqueness, we will also assume the following Lipschitz condition upon H, namely that
|H(x,p)−H(x,q)|≤b(x)|p−q|,b(x)∈LN(Ω), | (1.6) |
for almost every x∈Ω and every p,q∈RN.
The first main result that we prove in this note is the following.
Theorem 1.1. Let N≥3. Assume that A(x) satisfies (1.4) and that H(x,p) satisfies (1.5). Then there exist a constant λ∈R and a function u∈H1(Ω) which satisfy the elliptic problem (1.3) in the weak sense, i.e.,
∫ΩA(x)∇u∇φ+λ∫Ωφ=∫ΩH(x,∇u)φ+⟨χ,φ⟩∀φ∈H1(Ω). |
In addition, if H satisfies (1.6), problem (1.3) is solvable for a unique constant λ and the corresponding weak solution u is unique up to addition of a constant.
Even if the result of Theorem 1.1 is quite simple, it seems new to the best of our knowledge, except for the linear case, which was treated in [10] through Fredholm theory. As usual, the existence of the constant λ is proved by considering the singular ergodic limit, as ε→0, of solutions uε of
{−div(A(x)∇uε)+εuε=H(x,∇uε)+χin Ω,A(x)∇uε⋅→n=0on ∂Ω. | (1.7) |
Here the main difference, compared to the classical case ([15]), is that the uniform estimate of εuε, usually given by the maximum principle, is not available because of the more singular x-dependence of the Hamiltonian. In fact, we directly estimate ‖∇uε‖L2(Ω) as a first, and then crucial, step; this is done with a similar strategy as suggested in [20] for the Dirichlet problem.
As a consequence of Theorem 1.1, and of our structure conditions, we eventually give a complete description of the limit of uε, assuming further that H(x,p) is differentiable with respect to p. In that case, the limit of uε can be fully characterized in terms of the additive eigenvalue of the linearized problem: this is the (non homogeneous) linear problem
{−div(A(x)∇w)+θ=Hp(x,∇ˉu)⋅∇w−ˉuin Ω,A(x)∇w⋅→n=0on ∂Ω, | (1.8) |
where ˉu is the unique solution of (1.3) such that ∫Ωˉu=0.
Theorem 1.2. Assume that H(x,p) satisfies (1.5), (1.6) and H is differentiable with respect to p with Hp(x,p):=∂H(x,p)∂p being continuous in p, for a.e. x∈Ω. Let uε be the unique solution of (1.7) and (λ,ˉu) be the unique solution of (1.3) such that ∫Ωˉu=0. Then we have
limε→0(uε−λε−ˉu)=θ |
where the limit is in H1(Ω), and θ is the unique constant for which problem (1.8) is solvable. Moreover, we also have (in L2(Ω))
∇uε=∇ˉu+ε∇w+o(ε)asε→0, |
where w is the unique solution of (1.8) with zero average.
The equivalent of this result for much more general diffusion operators and Hamiltonians, but with a smooth dependence on x, is proved in many recent papers through viscosity solutions' methods, see e.g., [13,17] for second order problems, where this is called the selection problem, since the constant θ selects the limit of uε among all possible solutions of the ergodic problem (1.3). Of course, this is much simpler for elliptic equations rather than for first order (or degenerate elliptic) problems, as treated e.g., in [9] or in the pioneering paper [8].
The proofs of Theorem 1.1 and Theorem 1.2 are given in the next Section. Later we briefly address some extension of our results to nonlinear divergence form operators (Theorem 2.5) and to more singular x-dependence, including the case of data f(x) in Lebesgue spaces Lm(Ω), m≥1 (see Theorem 2.9 and Theorem 2.10).
Let us recall that, in the following, Ω is a bounded connected domain in RN, N≥3, with Lipschitz boundary, and we denote by →n the outward normal unit vector to the boundary ∂Ω. The Sobolev space is denoted by H1(Ω) and its dual by (H1(Ω))′. We start with a characterization of all possible weak subsolutions of a Neumann elliptic problem.
Lemma 2.1. Let w∈H1(Ω) and λ∈R satisfy
{−div(A(x)∇w)=B(x)+λinΩ,A(x)∇w⋅→n=0on∂Ω, | (2.1) |
where the matrix A(x) satisfies (1.4) and the function B(x)∈L2NN+2(Ω) satisfies
|B(x)|≤b(x)|∇w(x)|forsomeb∈LN(Ω). | (2.2) |
Then, we have λ=0 and the function w(x) is constant in Ω.
Proof. We divide the proof in three steps.
Step 1. We prove that w∈L∞(Ω).
This is standard, but we sketch the argument for the reader's convenience, following [21]. For k>0 we use Gk(w):=(w−k)+ as test function in problem (2.1). Then, using the ellipticity of A(x) and condition (2.2), we get
α∫Ω|∇Gk(w)|2dx≤∫Ω[b(x)|∇w|+λ]Gk(w)dx | (2.3) |
Let us set Ak:={x∈Ω:w(x)>k}. Since the integral in the right-hand side is restricted in the set Ak, we deduce, using Hölder inequality,
α∫Ω|∇Gk(w)|2dx≤(∫Ak∩{|∇w|≠0}|b|N)1N‖∇Gk(w)‖L2(Ω)‖Gk(w)‖L2∗(Ω)+|λ|‖Gk(w)‖L2∗(Ω)|Ak|1−12∗. |
This readily implies
∫Ω|∇Gk(w)|2dx≤c(∫Ak∩{|∇w|≠0}|b|N)2N‖Gk(w)‖2L2∗(Ω)+c|λ|‖Gk(w)‖L2∗(Ω)|Ak|1−12∗, | (2.4) |
where, here and below, c denotes possibly different numbers only depending on α,N,Ω. By Poincaré-Wirtinger inequality we deduce
‖Gk(w)−∫Ω−Gk(w)‖2L2∗(Ω)≤c(∫Ak|b|N)2N‖Gk(w)‖2L2∗(Ω)+c|λ|‖Gk(w)‖L2∗(Ω)|Ak|1−12∗ |
which implies
‖Gk(w)‖2L2∗(Ω)≤c(∫Ak|b|N)2N‖Gk(w)‖2L2∗(Ω)+c|λ|‖Gk(w)‖L2∗(Ω)|Ak|1−12∗+c|∫ΩGk(w)|2. | (2.5) |
We estimate last term as
|∫ΩGk(w)|2≤‖Gk(w)‖2L2∗(Ω)|Ak|2−22∗≤‖Gk(w)‖2L2∗(Ω)(‖w‖L1(Ω)k)2−22∗. |
Using this estimate in (2.5), we obtain that
‖Gk(w)‖2L2∗(Ω)[1−c(∫Ak|b|N)2N−c(‖w‖L1(Ω)k)2−22∗]≤c|λ|‖Gk(w)‖L2∗(Ω)|Ak|1−12∗. |
Let us take k0 sufficiently large such that, for every k≥k0,
1−c(∫Ak|b|N)2N−c(‖w‖L1(Ω)k)2−22∗≥12. |
Then we have
‖Gk(w)‖L2∗(Ω)≤2c|λ||Ak|1−12∗∀k≥k0. |
Hence
∫ΩGk(w)≤‖Gk(w)‖L2∗(Ω)|Ak|1−12∗≤2c|λ||Ak|2−22∗. |
Recall that 2−22∗=1+2N, and, for a.e. k, we have ddk∫ΩGk(w)=−|Ak|. This means that the function φ(k):=∫ΩGk(w) is a non increasing function which satisfies φ≤2c|λ|(−φ′)1+2N for all k≥k0. It follows that φ(k1)=0 for some k1>k0. Hence w(x)≤k1 a.e. in Ω. Repeating the argument for −w, we conclude that w∈L∞(Ω).
Step 2. We prove that λ=0.
To this purpose, we reason by contradiction. Suppose that λ<0. Since w∈L∞(Ω), for ε sufficiently small we have λ+εw≤0. Hence w satisfies
{−div(A(x)∇w)+εw≤b(x)|∇w|,in Ω,A(x)∇w⋅→n=0on ∂Ω. |
This implies (with the same proof as e.g., [20, Proposition 2.1]) that w≤0. Since w+c is still a solution of (2.1), whatever is the constant c∈R, we easily get a contradiction. Of course the same argument applies if λ>0. We conclude that λ=0.
Step 3. We now prove that w is a constant. To this purpose, we recall that the median of a function u∈H1(Ω) is defined as
med(u):=sup{k∈R:meas({u>k})≥|Ω|2}. |
As for the average of u, a Poincaré type inequality holds for u−med(u), see e.g., [22]. Namely there exists a constant C (only depending on N,Ω) such that
‖u−med(u)‖L2(Ω)≤C‖∇u‖L2(Ω)∀u∈H1(Ω) |
and therefore, by Sobolev inequality, we have, for a possibly different constant C,
‖u−med(u)‖L2∗(Ω)≤C‖∇u‖L2(Ω)∀u∈H1(Ω). | (2.6) |
We now normalize our solution w so that
med(w)=0. |
This implies that med(ψ(w))=0 for every nondecreasing Lipschitz function ψ:R→R such that ψ(0)=0. In particular, we have that med(Gk(w))=0 for all k>0, where now Gk(s)=(|s|−k)+sign(s). Defining Ak={x:|w(x)|>k}, we obtain as in (2.4) (with the additional information now that λ=0):
∫Ω|∇Gk(w)|2≤c(∫Ak∩{|∇w|≠0}|b|N)2N‖Gk(w)‖2L2∗(Ω) |
and so by (2.6) we deduce
‖Gk(w)‖2L2∗(Ω)≤c(∫Ak∩{|∇w|≠0}|b|N)2N‖Gk(w)‖2L2∗(Ω). |
This inequality now yields that w=0. In fact, assume by contradiction that M:=sup|w|>0 and take a sequence of k<M, k↑M. Since k<M, we have ‖Gk(w)‖2L2∗(Ω)≠0 (otherwise this contradicts the definition of M), so the previous inequality implies
1≤c(∫Ak∩{|∇w|≠0}|b|N)2N. |
But, as k↑M, we have |Ak∩{|∇w|≠0}|→0, because ∇w=0 a.e. in the set {|w|=M}. Hence we reach a contradiction. This implies that w=0, so w coincides with its median, and it is a constant.
Remark 2.2. We notice that this lemma remains true for a nonlinear divergence form operator −div(a(x,∇u)) which satisfies
a(x,p)⋅p≥α|p|2forsomeα>0. | (2.7) |
We now analyze the limit, as ε→0, of the elliptic problem (1.7). For the purposes of Theorem 1.2, it is convenient to state this kind of result in a slightly more general form, where the Hamiltonian H(x,p) may possibly depend on ε as well.
Lemma 2.3. Assume that A(x) satisfies (1.4), and that Hε(x,p) is a sequence of Carathéodory functions satisfying
|Hε(x,p)|≤b(x)|p|+f(x),forsomeb(x)∈LN(Ω),$f(x)∈L2NN+2(Ω)$, | (2.8) |
for almost every x∈Ω and for every p∈RN.
For χ∈(H1(Ω))′, and ε>0, let uε be solutions of the elliptic problem
{−div(A(x)∇uε)+εuε=Hε(x,∇uε)+χinΩ,A(x)∇uε⋅→n=0on∂Ω, | (2.9) |
and assume that there exists a function H(x,p) such that
Hε(x,p)→H(x,p)foreveryp∈RN,andalmosteveryx∈Ω. |
Then there exist a constant λ∈R and a function u∈H1(Ω) such that, up to a subsequence,
εuε→λanduε−∫Ω−uε→u |
where the limits are in the (strong) topology of H1(Ω). Moreover, (λ,u) solve problem (1.3).
Proof. We first claim that
∃K>0(independent of ε):‖∇uε‖L2(Ω)≤K∀ε>0. | (2.10) |
We proceed by contradiction and suppose that (2.10) is not true. This implies the existence of a subsequence of uε, not relabeled, such that ‖∇uε‖L2(Ω)→∞. We set
wε:=uε−∫Ω−uε‖∇uε‖L2(Ω). |
Since wε has zero average and ‖∇wε‖L2(Ω)=1, we deduce that wε is weakly relatively compact in H1(Ω) and strongly in L2(Ω) by Rellich's theorem. We observe that wε satisfies
−div(A(x)∇wε)+εwε=1‖∇uε‖L2(Ω)[Hε(x,∇uε)+χ]−ε|Ω|∫Ωuε‖∇uε‖L2(Ω)in Ω. | (2.11) |
Last term is a sequence of real numbers that we estimate, integrating (2.9), as
ε∫Ωuε‖∇uε‖L2(Ω)=1‖∇uε‖L2(Ω)∫ΩHε(x,∇uε)+1‖∇uε‖L2(Ω)⟨χ,1⟩≤1‖∇uε‖L2(Ω)∫Ω[b(x)|∇uε|+f(x)]+‖χ‖(H1(Ω))′|Ω|12‖∇uε‖L2(Ω)≤1‖∇uε‖L2(Ω)[‖b‖L2(Ω)‖∇uε‖L2(Ω)+‖f‖L1(Ω)+‖χ‖(H1(Ω))′|Ω|12], |
where we used (2.8) as well. Since b∈LN(Ω) (N≥2) and f∈L1(Ω), and since ‖∇uε‖L2(Ω) diverges, the right-hand side is bounded. Hence ε|Ω|∫Ωuε‖∇uε‖L2(Ω) is a bounded sequence and there exists a real value λ∈R such that, up to subsequences (not relabeled here), we have
ε|Ω|∫Ωuε‖∇uε‖L2(Ω)→λ. | (2.12) |
Now we use once more (2.8) to estimate
1‖∇uε‖L2(Ω)|Hε(x,∇uε)|≤b(x)|∇wε|+f(x)‖∇uε‖L2(Ω). | (2.13) |
Since b∈LN(Ω) and wε is bounded in H1(Ω), the product b(x)|∇wε| is bounded in L2NN+2(Ω). In addition, it is also equi-integrable: indeed, for any set E, one has
∫E|b(x)∇wε|2NN+2≤‖∇wε‖2NN+2L2(Ω)(∫Eb(x)N)2N+2≤C(∫Eb(x)N)2N+2 |
and last term (independent of ε) goes to zero as |E|→0. Thus, in view of (2.13), the term 1‖∇uε‖L2(Ω)Hε(x,∇uε) is bounded and equi-integrable in L2NN+2(Ω). Therefore, we have
{wε is bounded in H1(Ω),−div(A(x)∇wε)+εwε=hε+χ‖∇uε‖L2(Ω) | (2.14) |
for some hε bounded and equi-integrable in L2NN+2(Ω).
This implies that wε is actually strongly compact in H1(Ω). We recall the argument for the reader's convenience. The key point is that, if w is a weak limit of wε, then
∫Ωhε(wε−w)→ε→00. | (2.15) |
In fact, we have
∫Ωhε(wε−w)=∫ΩhεTk(wε−w)+∫ΩhεGk(wε−w) |
where Tk(s)=max(min(s,k),−k) is the truncation function and Gk(s) is the difference s−Tk(s). By Hölder inequality, since wε−w is bounded in L2∗(Ω) one has, by definition of Gk(⋅):
|∫ΩhεGk(wε−w)|≤C(∫{|wε−w|>k}|hε|2NN+2)N+22N |
where last term vanishes as k→∞, uniformly with respect to ε, since hε is equi-integrable in L2NN+2(Ω). Therefore one has
∫Ωhε(wε−w)=∫ΩhεTk(wε−w)+rk |
where rkk→∞→0 uniformly with respect to ε. But hε is bounded in L2NN+2(Ω) while Tk(wε−w)→0 strongly in Lp(Ω) for any p<∞ due to Rellich theorem; hence the first integral in the right-hand side converges to zero as ε→0. Therefore, by letting ε→0 and then k→∞, we deduce that (2.15) holds true. With (2.15) in hands, it is now easy to deduce from (2.14) that
‖∇wε−∇w‖2L2(Ω)≤1α∫ΩA(x)∇(wε−w)⋅∇(wε−w)→ε→00. |
We have obtained so far that there exists a function w∈H1(Ω) such that, up to subsequences,
wε→wstrongly in H1(Ω). |
In particular, this implies, up to subsequences, that ∇wε→∇w almost everywhere in Ω. Therefore, as a consequence of (2.13), the term 1‖∇uε‖L2(Ω)Hε(x,∇uε) weakly converges in L2NN+2(Ω) towards some function B(x) which satisfies |B(x)|≤b(x)|∇w(x)| a.e. in Ω.
Finally, due to (2.11), (2.12), (2.13), w satisfies (in weak sense)
{−div(A(x)∇w)≤b(x)|∇w|+λin Ω,A(x)∇w⋅→n=0on ∂Ω. | (2.16) |
By Lemma 2.1 we deduce that w is constant, and since ∫Ωw=0 this means w(x)≡0. But this is a contradiction with the fact that wε strongly converges to w in H1(Ω) and ‖∇wε‖L2(Ω)=1 for every ε. The contradiction proves that the assertion (2.10) is true.
Thanks to (2.10), now we integrate (2.9) and we get
ε|∫Ωuε|=|∫ΩHε(x,∇uε)+⟨χ,1⟩|≤∫Ωb(x)|∇uε|+f(x)+|⟨χ,1⟩|≤C, |
due to the growth condition (2.8). Therefore, ε∫Ωuε is a bounded sequence, and there exists a real value λ∈R such that, up to subsequences,
ε∫Ω−uε→λ. |
We now set ^uε:uε−∫Ω−uε. Then ^uε is bounded in H1(Ω) and solves
−div(A(x)∇^uε)+ε^uε=Hε(x,∇^uε)+χ−ε∫Ω−uεin Ω. | (2.17) |
With the same arguments used before for wε, we can show that ^uε is strongly compact in H1(Ω), so there exists a function u∈H1(Ω) such that ^uε→u in H1(Ω). We may also assume that ∇^uε→∇u almost everywhere in Ω, and therefore almost uniformly as well; since Hε(x,∇^uε) is equi-integrable and Hε(x,p)→H(x,p), we can deduce that Hε(x,∇^uε) converges to H(x,∇u) (which belongs to L2NN+2(Ω) due to (2.8)). Finally, passing to the limit in (2.17), we obtain that (u,λ) is a solution to problem (1.3).
The proof of Theorem 1.1 immediately follows from the above two results.
Proof of Theorem 1.1. For ε>0, let uε be the solution of the elliptic problem (1.7). Applying Lemma 2.3, we have that, up to subsequences, εuε→λ and uε−∫Ω−uε→u, where (λ,u) gives a solution to problem (1.3).
Now we assume further the condition (1.6) and we prove uniqueness of λ and u (up to a constant). This is a straightforward consequence of Lemma 2.1. Indeed, let (u1,λ1) and (u2,λ2) be solutions to problem (1.3). Then w:=u1−u2 is a solution to
{−div(A(x)∇w)=H(x,∇u1)−H(x,∇u2)+λ1−λ2in Ω,A(x)∇w⋅→n=0on ∂Ω, |
where we have, due to (1.6),
|H(x,∇u1)−H(x,∇u2)|≤b(x)|∇w|. |
From Lemma 2.1 we deduce that λ1=λ2 and that u1−u2 is a constant.
Given the solution uε of (1.7), we now investigate the behavior of uε−λε−ˉu where (λ,ˉu) is a solution of the ergodic problem (1.3). In order to fix a reference solution, we normalize ˉu so that ∫Ωˉu=0. So (λ,ˉu) is uniquely defined from Theorem 1.1, provided (1.6) holds. The proof of Theorem 1.2 now follows as a Corollary of the previous results.
Proof of Theorem 1.2. Let us define
vε:=1ε(uε−λε−ˉu). |
One can check that vε solves
{−div(A(x)∇vε)+εvε=1ε[H(x,∇uε)−H(x,∇ˉu)]−ˉuin Ω,A(x)∇vε⋅→n=0on ∂Ω. | (2.18) |
We notice that, thanks to (1.6), the function
˜Hε(x,p):=1ε[H(x,εp+∇ˉu)−H(x,∇ˉu)] |
satisfies
|˜Hε(x,p)|≤b(x)|p|,|˜Hε(x,p)−˜Hε(x,q)|≤b(x)|p−q| |
and, from the differentiability of H, we have
˜Hε(x,p)→Hp(x,∇ˉu)⋅pfor every p∈RN and a.e. x∈Ω. |
Therefore Lemma 2.3 applies and we deduce that there exists a constant θ∈R and a function w∈H1(Ω) such that
εvε→θ,vε−∫Ω−vε→w |
where the limits are in the strong topology of H1(Ω) and (θ,w) is the unique couple which satisfies the linear elliptic problem (1.8) with the normalized condition ∫Ωw=0. We stress that the uniqueness of the limit couple implies that the whole sequence vε converges.
Coming back from vε to uε, this means that
limε→0(uε−λε−ˉu)=θ |
where the limit is meant in H1(Ω), and in addition
1ε(uε−∫Ω−uε−ˉu)→w. |
This latter convergence yields the first order expansion for the gradient in L2(Ω):
∇uε=∇ˉu+ε∇w+o(ε)as ε→0. |
Remark 2.4. The case N=2 can be dealt with in exactly the same way as before, except that the threshold summability of the drift term should be adapted to the Sobolev embedding of dimension two. Since for N=2 the space H1(Ω) is embedded in Lp(Ω) for every p<∞, but not in L∞(Ω), one needs here b(x)∈Lq(Ω) for some q>2 in order that the product b(x)|∇u| belongs to the dual space (H1(Ω))′. Therefore, conditions (1.5) and (1.6) should be changed into
|H(x,p)|≤b(x)|p|+f(x),for some b(x)∈Lq(Ω), q>2, and f(x)∈Lm(Ω), m>1, | (2.19) |
and respectively,
|H(x,p)−H(x,q)|≤b(x)|p−q|,b(x)∈Lq(Ω), q>2, | (2.20) |
for almost every x∈Ω and for every p,q∈RN.
Replacing assumptions (1.5) and (1.6) with, respectively, (2.19) and (2.20), the results stated in Theorem 1.1 and Theorem 1.2 are true for the dimension N=2, and the proof remains the same up to the obvious modifications in the Lebesgue spaces which are involved.
A similar remark holds true for the case of dimension N=1; in that case it is enough to assume b(x)∈L2(Ω) and f(x)∈L1(Ω).
In this subsection we give a short extension of the result of Theorem 1.1 to the case of nonlinear operators. To this purpose, we introduce a function a(x,p):Ω×RN→RN which is assumed to be measurable with respect to x, for every p∈RN, and continuous with respect to p, for a.e. x∈Ω. We assume that a(x,p) satisfies the following monotonicity and growth conditions:
(a(x,p)−a(x,q))⋅(p−q)≥α|p−q|2,for some α∈R, α>0, | (2.21) |
and
|a(x,p)|≤β(|p|+k(x))for some β∈R, k(x)∈L2(Ω), | (2.22) |
for almost every x∈Ω and every p,q∈RN.
Notice that (2.21)–(2.22) imply also the coercivity condition
a(x,p)⋅p≥α|p|2−˜k(x),where α∈R, α>0, ˜k(x)∈L1(Ω). | (2.23) |
We also assume that a(x,p) satisfies the following asymptotic condition for |p|→∞:
for a.e. x∈Ω, every p∈RN, ∃limt→∞a(x,tp)t. | (2.24) |
Of course, the case a(x,p)=A(x)⋅p+K(x), for a matrix A(x) satisfying (1.4) and a vector field K∈L2(Ω), is the simplest example where conditions (2.21)–(2.24) are satisfied.
Thanks to (2.21)–(2.24), we can extend Theorem 1.1 to a nonlinear setting. Notice that additional terms in the equation, belonging to (H1(Ω))′, can be included here in the vector field a(x,p) or in the local function H(x,p).
Theorem 2.5. Assume that a(x,p) satisfies (2.21)–(2.24), and that H(x,p) satisfies (1.5) (or it satisfies (2.19) if N=2). Then there exist a constant λ∈R and a function u∈H1(Ω) which solve, in the weak sense, the elliptic problem
{−div(a(x,∇u))+λ=H(x,∇u)inΩ,a(x,∇u)⋅→n=0on∂Ω. | (2.25) |
In addition, if (1.6) holds true (respectively, (2.20) if N=2), then λ is unique and u is unique up to addition of a constant.
Proof. For ε>0, we consider the approximating problem
{−div(a(x,∇uε))+εuε=H(x,∇uε)in Ω,a(x,∇uε)⋅→n=0on ∂Ω. | (2.26) |
Then we aim at showing that the a priori estimate (2.10) holds true and we proceed, as in Lemma 2.3, by contradiction. This allows us to build a sequence wε such that ‖∇wε‖L2(Ω)=1, wε has zero average and satisfies
−div(a(x,‖∇uε‖L2(Ω)∇wε)‖∇uε‖L2(Ω))+εwε=H(x,∇uε)‖∇uε‖L2(Ω)−ε|Ω|∫Ωuε‖∇uε‖L2(Ω)in Ω, | (2.27) |
where ‖∇uε‖L2(Ω)→∞. As in Lemma 2.3, last term is a relatively compact sequence of real numbers, and we observe that the right-hand side is bounded and equi-integrable in L2NN+2(Ω), so that wε satisfies
{wε is bounded in H1(Ω),−div(ˆaε(x,∇wε))+εwε=hε | (2.28) |
for some hε bounded and equi-integrable in L2NN+2(Ω), where
ˆaε(x,p):=a(x,‖∇uε‖L2(Ω)p)‖∇uε‖L2(Ω). |
Now, let w be a weak limit of wε in H1(Ω), and a strong limit in L2(Ω). Notice that we have that
∫Ωˆaε(x,∇w)∇(wε−w)→0 |
because ˆaε(x,∇w) is strongly convergent in L2(Ω) due to (2.24) and (2.22), by Lebesgue's theorem. Therefore, using (2.28) and (2.15) as well, we deduce that
limε→0∫Ω(ˆaε(x,∇wε)−ˆaε(x,∇w))∇(wε−w)=0. | (2.29) |
Notice that ˆaε satisfies (2.21) for the same α>0, and for every ε>0. Then (2.29) implies that wε→w strongly in H1(Ω) (and, up to a subsequence, ∇wε→∇w almost everywhere in Ω). Now, for every φ∈H1(Ω), we set
ˆa(x,∇φ(x)):=limε→0ˆaε(x,∇φ(x)) | (2.30) |
which exists after (2.24) and belongs to L2(Ω) due to (2.22). It is a consequence of (2.29) and a standard monotonicity argument (sometimes known as Minty's argument) that ˆaε(x,∇wε(x)) converges to ˆa(x,∇w(x)) weakly in L2(Ω). Finally, passing to the limit in (2.27), we obtain that w is a weak solution of
{−div(ˆa(x,∇w(x)))+λ=B(x)in Ω,ˆa(x,∇w(x))⋅→n=0on ∂Ω |
for some λ∈R and some B(x)∈L2NN+2(Ω) satisfying condition (2.2). Since ˆa(x,p) satisfies (2.7) (because ˆa(x,0)=0 and (2.21) holds), we deduce from Lemma 2.1, Remark 2.2, that λ=0 and that w is constant (hence w=0 because it has zero average). We get a contradiction with ‖∇w‖L2(Ω)=1. This proves that the a priori estimate (2.10) holds true, that is ‖∇uε‖L2(Ω) is uniformly bounded. Then the compactness of uε in H1(Ω) follows in a similar way as before, using (2.21), and we conclude as in Theorem 1.1 the existence of λ∈R, u∈H1(Ω) which solve problem (2.25).
By using (1.6) and (2.21), the uniqueness of λ and the uniqueness of u, up to a constant, are proved exactly as in Theorem 1.1. Even if the problem is nonlinear, and Lemma 2.1 cannot be literally applied, the arguments are exactly the same as used in Lemma 2.1, now applied to w:=u1−u2; the nonlinearity of the operator is readily handled with assumption (2.21).
Remark 2.6. Condition (2.24) is assumed here in order that we can follow the same approach used before in the proof of Theorem 1.1. We believe this condition to be unnecessary for a similar result to hold for general nonlinear operators, however removing this condition would need a substantial change in the method of proof (e.g. using symmetrization methods), which is beyond the scope of this note.
Remark 2.7. The result remains true for the case of dimension N=1, up to requiring b∈L2(Ω) and f∈L1(Ω) in assumptions (1.5) and (1.6).
The same approach as before can be used in case of more singular dependence with respect to x. We have in mind here that the assumption (1.5) is replaced by
|H(x,p)|≤b(x)|p|+f(x),for some b(x)∈LN(Ω), f(x)∈Lm(Ω), | (2.31) |
for almost every x∈Ω and for every p∈RN, where m≥1, so that data can belong to Lebesgue spaces of any order.
Of course, if m≥2NN+2, assumption (2.31) implies (1.5), so there is nothing new to be proved. By contrast, if m<2NN+2, we cannot expect the solutions to belong to H1(Ω) anymore; however it is still possible to obtain similar results in a setting of generalized solutions. We recall below the notion of renormalized solutions. The truncation function is denoted, as before, by Tk(s):=max(min(s,k),−k). In what follows we suppose that N>1.
Definition 2.8. A function u, belonging to W1,q(Ω) for every q<NN−1, is a renormalized solution of problem (2.25) if:
(i) Tk(u)∈H1(Ω) for every k>0, and H(x,∇u)∈L1(Ω).
(ii) For every function h:R→R which is C1 with compact support, it holds
∫Ωa(x,∇u)∇(h(u)φ)+λ∫Ωh(u)φ=∫ΩH(x,∇u)h(u)φ∀φ∈H1(Ω)∩L∞(Ω). | (2.32) |
(iii)
limn→∞1n∫{n<|u|<2n}|∇u|2=0. | (2.33) |
We recall that this notion of solution, introduced by P. L. Lions and F. Murat (in a joint paper unpublished, whose content can be found in [16,18]) is nowadays currently used as a formulation of elliptic problems with L1-data, or even Radon measures. For similar problems with first order terms having linear growth, we refer e.g., to [5].
It is to be noted that, in the renormalized formulation, the truncations of the solution belong to the energy space H1(Ω), so that (2.32) makes sense for any h with compact support. Condition (2.33), in turn, implies that renormalized solutions are also distributional solutions. To this purpose one can take h=S(rn) where S(t) is a piecewise linear function, supported in [−2,2], such that S(t)≡1 for t∈[−1,1]; then S(rn)→1 as n→∞ and the distributional formulation (with test functions φ∈C1(¯Ω)) is recovered thanks to (2.33).
Up to replacing the (H1(Ω))′ formulation with the renormalized setting, Theorem 2.5 admits a natural extension which is the following one. We recall (see [7]) that elliptic equations (including possibly nonlinear divergence form operators with discontinuous coefficients) with source terms in Lm(Ω), m<2NN+2, admit solutions which belong to W1,m∗(Ω), where m∗ is the Sobolev exponent (Nm)/(N−m).
Theorem 2.9. Let N≥3. Assume that a(x,p) satisfies (2.21)–(2.24) and that H(x,p) satisfies (2.31) for some 1<m<2NN+2. Then there exists a constant λ∈R and a function u∈W1,m∗(Ω) which is a renormalized solution of (2.25). In addition, if (1.6) holds true, λ is unique and u is unique up to addition of a constant.
Proof. We only sketch the main steps, and the main differences with the arguments of Theorem 2.5. We start with the solutions uε of the approximating problem
{−div(a(x,∇uε))+εuε=Hε(x,∇uε)in Ω,a(x,∇uε)⋅→n=0on ∂Ω, | (2.34) |
where Hε(x,p)=T1ε(H(x,p)). Here the truncation of H is only needed if one wants to work, at fixed ε, with the more comfortable setting of finite energy solutions. Then we claim that there exists K>0 independent of ε, such that
‖∇uε‖Lm∗(Ω)≤K∀ε>0. | (2.35) |
The proof of (2.35) is done, as before, by contradiction. In this case we build a subsequence, not relabeled, such that ‖∇uε‖Lm∗(Ω)→∞ and we define
wε=uε−∫Ω−uε‖∇uε‖Lm∗(Ω) |
which solves
−div(ˆa(x,∇wε))+εwε=Hε(x,∇uε)‖∇uε‖Lm∗(Ω)−ε|Ω|∫Ωuε‖∇uε‖Lm∗(Ω)in Ω, | (2.36) |
where ˆaε(x,p) is defined now as
ˆaε(x,p):=a(x,‖∇uε‖Lm∗(Ω)p)‖∇uε‖Lm∗(Ω). |
Compared to Theorem 2.5, we now have that the right-hand side is bounded and equi-integrable in Lm(Ω), hence wε satisfies
{wε is bounded in W1,m∗(Ω),−div(ˆaε(x,∇wε))+εwε=hε | (2.37) |
for some hε which is bounded and equi-integrable in Lm(Ω). One can prove now that wε is actually strongly compact in W1,m∗(Ω). The argument needs a modification of what is done in Lemma 2.3: first of all, by only using that hε is weakly converging in L1(Ω), one can prove (see e.g., [18,14]) that
limε→0∫Ω(ˆaε(x,∇Tk(wε))−ˆaε(x,∇Tk(w)))∇(Tk(wε)−Tk(w))=0 |
which yields, thanks to (2.21),
Tk(wε)→Tk(w)in H1(Ω), for every k>0. | (2.38) |
To estimate Gk(wε)=wε−Tk(wε), we take
[(σ+|Gk(wε)|)γ−σγ]sign(wε) |
as test function in (2.37), for σ,γ>0. We get
γ∫Ωˆaε(x,∇wε)∇Gk(wε)(σ+|Gk(wε)|)γ−1≤∫Ωhε[(σ+|Gk(wε)|)γ−σγ]sign(wε) |
which implies, using (2.23) and the definition of ˆa,
γα∫Ω|∇Gk(wε)|2(σ+|Gk(wε)|)γ−1≤∫Ωhε|(σ+|Gk(wε)|)γ−σγ|+γ‖∇uε‖2Lm∗(Ω)∫Ω˜k(x)(σ+|Gk(wε)|)γ−1. | (2.39) |
We choose γ=N(m−1)N−2m, so that |wε|γ=|wε|m∗∗m′ is bounded in Lm′(Ω) due to the bound of wε in W1,m∗(Ω) and Sobolev embedding. Moreover, we have γ<1 since m<2NN+2. Taking for instance σ=1, the previous inequality implies
γα∫Ω|∇Gk(wε)|2(1+|Gk(wε)|)γ−1≤2∫{|wε|>k}|hε|(1+|wε|)γ+γ‖∇uε‖2Lm∗(Ω)‖˜k‖L1(Ω)≤C(∫{|wε|>k}|hε|m)1m+C‖∇uε‖2Lm∗(Ω). |
Due to the equi-integrability of hε in Lm(Ω) we deduce
∫Ω|∇Gk(wε)|2(1+|Gk(wε)|)γ−1≤δk+C‖∇uε‖2Lm∗(Ω) | (2.40) |
where δk denotes a quantity which vanishes as k→∞ uniformly with respect to ε. From Hölder inequality, since m∗<2 we have
∫Ω|∇Gk(wε)|m∗≤(∫Ω|∇Gk(wε)|2(1+|Gk(wε)|)γ−1)m∗2(∫Ω(1+|Gk(wε)|)(1−γ)m∗2−m∗)1−m∗2. |
The precise value of γ yields (1−γ)m∗2−m∗=m∗∗, so last term is bounded and from (2.40) we deduce
∫Ω|∇Gk(wε)|m∗≤Cδm∗2k+C‖∇uε‖m∗Lm∗(Ω). |
Putting together this information with (2.38), it follows that wε→w strongly in W1,m∗(Ω). In particular, up to a subsequence, ∇wε→∇w almost everywhere in Ω. This implies that Hε(x,∇uε)‖∇uε‖Lm∗(Ω) weakly converges to some function B(x) in Lm(Ω), and, because of (2.31) and the a.e. convergence of ∇wε, it follows that B satisfies (2.2). If the vector filed ˆa(x,p) is defined as in (2.30), then by passing to the limit we obtain a function w such that
{w∈W1,m∗(Ω),−div(ˆa(x,∇w))+λ=B(x)with B(x)∈Lm(Ω) satisfying (2.2). |
Now we get a contradiction by showing that w=0. This needs a slight refinement of the argument of Lemma 2.1; indeed, w does not belong a priori to H1(Ω). However, from (2.39) one obtains, letting first ε→0 and then σ→0, the inequality
γα∫Ω|∇Gk(w)|2|Gk(w)|γ−1≤∫Ω[b(x)|∇w|+|λ|]|Gk(w)|γ. | (2.41) |
This replaces now the starting inequality (2.3) of Lemma 2.1; using that
∫Ω|∇Gk(w)|m∗≤(∫Ω|∇Gk(w)|2|Gk(w)|γ−1)m∗2(∫Ω|Gk(w)|(1−γ)m∗2−m∗)1−m∗2≤C(∫Ω[b(x)|∇Gk(w)|+|λ|]|Gk(w)|γ)m∗2(∫Ω|Gk(w)|(1−γ)m∗2−m∗)1−m∗2 |
and the equalities γm′=m∗∗ and (1−γ)m∗2−m∗=m∗∗, one can prove with similar steps as in Lemma 2.1 that w∈L∞(Ω). Once w is proved to be bounded, then it has also finite energy and the conclusion of Lemma 2.1 applies. This completes the contradiction argument and concludes the proof of the a priori estimate (2.35). Once uε is proved to be bounded in W1,m∗(Ω), we repeat the same argument to deduce that uε−∫Ω−uε is relatively compact and converges, up to subsequences, to some u∈W1,m∗(Ω). Due to (2.31), this implies that H(x,∇uε)→H(x,∇u) in Lm(Ω). By well-known results (see e.g., [5,18]), the limit function u is a weak and a renormalized solution of the limiting problem (2.25).
Let us now sketch the uniqueness part. Assume that λ1, u1 and λ2,u2 are (renormalized) solutions of the problem. Then it is possible to prove, using renormalization arguments (e.g., the arguments used for the obtention of inequality (3.14) in [19, Theorem 3.1]), that u1,u2 satisfy
γ∫Ω(a(x,∇u1)−a(x,∇u2))∇Gk(u1−u2)(σ+|Gk(w)|)γ−1≤∫Ω|H(x,∇u1)−H(x,∇u2)||(σ+|Gk(w)|)γ−σγ|+|λ1−λ2|∫Ω|(σ+|Gk(w)|)γ−σγ| |
where w=u1−u2 and γ=N(m−1)N−2m as before. The reader may check that all terms here are well defined because of the precise value of γ and since u1,u2∈W1,m∗(Ω). Using (2.21) and (1.6) it follows that
γα∫Ω|∇Gk(w)|2(σ+|Gk(w)|)γ−1≤∫Ωb(x)|∇w||(σ+|Gk(w)|)γ−σγ|+|λ1−λ2|∫Ω|(σ+|Gk(w)|)γ−σγ|. |
By letting σ→0 one gets inequality (2.41) with λ=λ1−λ2. This inequality allows us to show that w∈L∞(Ω) as we said before. Now, let us assume for instance that λ1>λ2; since u1−u2∈L∞(Ω), then ε(u1−u2)<λ1−λ2 for ε sufficiently small. Therefore, we get that u1, u2 are, respectively, a renormalized sub and super solution of the same equation
−div(a(x,∇u))+εu=H(x,∇u)+εu2−λ2. |
This implies (e.g., as in [19, Theorem 3.1], or in [6]) that u1≤u2, which yields a contradiction, because u1,u2 are solutions up to addition of any constant. We notice here that, if u is a renormalized solution of (2.25), then it can be easily proved that u+c is also a renormalized solution, whenever c∈R. The contradiction obtained proves that λ1=λ2. Then, we replace u1 with u1−med(u1−u2); now again from inequality (2.41) (obtained before for w=u1−u2−med(u1−u2) and λ=λ1−λ2=0) we can deduce that w=0, i.e., u1−u2=med(u1−u2). Hence the two solutions differ by a constant.
Finally, we conclude by observing that a similar result can be proved in the limiting case m=1, up to requiring that b(x)∈Lq(Ω) for some q>N. More precisely, we have the following result, whose proof can be done with similar arguments as indicated above.
Theorem 2.10. Let N≥2. Assume that a(x,p) satisfies (2.21)–(2.24) and that H(x,p) satisfies (2.31) with m=1 and b(x)∈Lq(Ω) for some q>N. Then there exists a constant λ∈R and a renormalized solution u of (2.25). In addition, if (1.6) holds true with b(x)∈Lq(Ω) for some q>N, then λ is unique and u is unique up to addition of a constant.
Alessio Porretta aknowledges the support of Fondation Sciences Matématiques de Paris and Istituto Nazionale di Alta Matematica (GNAMPA).
The authors declare no conflict of interest.
[1] | Arisawa M, Lions PL (1998) On ergodic stochastic control. Commun Part Diff Eq 23: 2187-2217. |
[2] | Bensoussan A, Frehse J (1987) On Bellman equation of ergodic type with quadratic growth Hamiltonian, In: Contributions to Modern Calculus of Variations, Wiley, 13-26. |
[3] | Bardi M, Capuzzo-Dolcetta I (1997) Optimal Control and Viscosity Solutions of Hamilton-JacobiBellman Equations, Boston: Birkhäuser. |
[4] | Bensoussan A, Lions JL, Papanicolaou G (1978) Asymptotic Analysis for Periodic Structures, Amsterdam: North-Holland. |
[5] | Betta F, Mercaldo A, Murat F, et al. (2003) Existence of renormalized solutions to nonlinear elliptic equations with a lower-order term and right-hand side a measure. J Math Pure Appl 82: 90-124. |
[6] | Betta F, Mercaldo A, Murat F, et al. (2005) Uniqueness results for nonlinear elliptic equations with a lower order term. Nonlinear Anal 63: 153-170. |
[7] | Boccardo L, Gallouët T (1992) Nonlinear elliptic equations with right hand side measures. Commun Part Diff Eq 17: 641-655. |
[8] | Capuzzo Dolcetta I, Menaldi JL (1988) Asymptotic behavior of the first order obstacle problem. J Differ Equations 75: 303-328. |
[9] | Davini A, Fathi A, Iturriaga R, et al. (2016) Convergence of the solutions of the discounted equation. Invent Math 206: 29-55. |
[10] | Droniou J, Vazquez JL (2009) Noncoercive convection-diffusion elliptic problems with Neumann boundary conditions. Calc Var 34: 413-434. |
[11] | Gilbarg D, Trudinger N (1983) Partial Differential Equations of Second Order, 2 Eds., BerlinNew-York: Springer-Verlag. |
[12] | Gimbert F (1985) Problèmes de Neumann quasilineaires. J Funct Anal 68: 65-72. |
[13] | Ishii H, Mitake H, Tran HV (2017) The vanishing discount problem and viscosity Mather measures. Part 1: the problem on a torus. J Math Pure Appl 108: 125-149. |
[14] | Leone C, Porretta A (1998) Entropy solutions for nonlinear elliptic equations in L1. Nonlinear Anal Theor 32: 325-334. |
[15] | Lions PL (1985) Quelques remarques sur les problèmes elliptiques quasilinéaires du second ordre. J Anal Math 45: 234-254. |
[16] | Lions PL (1996) Mathematical Topics in Fluid Mechanics. Vol. 1. Incompressible Models, New York: The Clarendon Press. |
[17] | Mitake H, Tran HV (2017) Selection problems for a discounted degenerate viscous HamiltonJacobi equation. Adv Math 306: 684-703. |
[18] | Murat F, Soluciones renormalizadas de EDP elipticas no lineales, Preprint 93023 of Laboratoire d'Analyse Numérique, Université Paris VI, 1993. Available from: http://archive.schools.cimpa.info/anciensite/NotesCours/PDF/2009/Alexandrie_Murat_2.pdf. |
[19] | Porretta A (2008) On the comparison principle for p-Laplace type operators with first order terms, In: On the Notion of Solutions to Nonlinear Elliptic Problems: Results and Developments, Quaderni di Matematica 23, Dept. Math. Seconda Univ. Napoli, Caserta, 459-497. |
[20] | Porretta A, Elliptic equations with first-order terms, notes of the CIMPA School, Alexandria, 2009. Available from: http://archive.schools.cimpa.info/anciensite/NotesCours/PDF/2009/Alexandrie_Porretta.pdf. |
[21] | Stampacchia G (1965) Le problème de Dirichlet pour les equations élliptiques du second ordre à coefficients discontinus, Ann I Fourier 15: 189-258. |
[22] | Ziemer WP (1989) Weakly Differentiable Functions, New York: Springer-Verlag. |
1. | Son N.T. Tu, Vanishing discount problem and the additive eigenvalues on changing domains, 2022, 317, 00220396, 32, 10.1016/j.jde.2022.01.055 |