Skip to content
BY 4.0 license Open Access Published by De Gruyter Open Access February 17, 2020

Non-occurrence of the Lavrentiev phenomenon for a class of convex nonautonomous Lagrangians

Carlo Mariconda ORCID logo and Giulia Treu
From the journal Open Mathematics


We consider the classical functional of the Calculus of Variations of the form


where Ω is a bounded open subset of ℝn and F : Ω × ℝ × ℝn → ℝ is a Carathéodory convex function; the admissible functions u coincide with a prescribed Lipschitz function ϕ on ∂Ω. We formulate some conditions under which a given function in ϕ + W01,p (Ω) with I(u) < +∞ can be approximated in the W1,p norm and in energy by a sequence of smooth functions that coincide with ϕ on ∂Ω. As a particular case we obtain that the Lavrentiev phenomenon does not occur when F(x, u, ξ) = f(x, u) + h(x, ξ) is convex and xF(x, 0, 0) is sufficiently smooth.

1 Introduction

In 1927, Lavrentiev [1] provided an example of the fact, later called the Lavrentiev phenomenon, that the infimum, over the set of absolutely continuous functions, of a one-dimensional functional of the calculus of variations may be strictly lower than the infimum of the same functional over the set of Lipschitz functions satisfying the same boundary conditions: the example was refined by Manià in [2] and, more recently, by Ball-Mizel in [3]. Finding the conditions that ensure the non-occurrence of the Lavrentiev phenomenon has some interest, if just for ensuring to catch the infimum of the functional via standard numerical methods. We point out that some authors refer to as the Lavrentiev phenomenon just the fact that the infima among the two aforementioned classes of functions differ without taking care the boundary datum. If one allows the boundary datum to vary, things change dramatically: in Lavrentiev’s celebrated example itself the infima among Lipschitz/absolutely functions are the same if one allows one boundary datum to be just arbitrarily close to the initial one. Some recent results concerning the study of this kind of “local” Lavrentiev phenomenon have been recently obtained in [4, Theorem 4].

Alberti and Serra Cassano proved in [5] that, when the integration set is an interval in ℝ, the phenomenon does not occur for autonomous Lagrangians. For scalar problems, where the domain is multi-dimensional, few results appeared in the literature. Of course, the problem becomes much easier if one imposes growth conditions of the Lagrangian from above, since approximations are facilitated by Lebesgue’s dominated convergence. A two-dimensional analogue of Manià’s example was provided in [6]. There are examples in [7] of functionals of the form F(x, ∇u) depending on the independent variable x of the space and on the gradient ∇u of the admissible functions that exhibit the Lavrentiev phenomenon; there are also cases in which the Lavrentiev phenomenon does not occur (see [8]). No example is known to the authors for scalar problems when the Lagrangian is autonomous.

It was conjectured by Buttazzo-Belloni in [9] that the phenomenon should not occur when the Lagrangian F(u, ∇u) is autonomous and convex in both variables, a fact that they proved in the case of a (strongly) star-shaped domain under the hidden growth assumption that F(u, 0) is summable and zero as a boundary datum. Other results that appeared aimed to prove the conjecture: we mention Ekeland-Temam who proved in [10] its validity for functionals of the gradient on a Lipschitz domain for a zero boundary datum; Bonfanti and Cellina in [6, 11] considered autonomous Lagrangians that are sum of a radial function of the gradient ∇u and a function of the variable u, under some smoothness assumptions on the boundary and on the boundary datum. A complete answer to the conjecture was given by Bousquet-Mariconda-Treu in [12], where they showed that whenever F(u, ∇u) is convex, given u W01,1 (Ω) with a Lipschitz boundary datum and finite energy (i.e., F(u, ∇u) ∈ L1(Ω)), there is no Lavrentiev gap at u: there exists a sequence (uk)k of Lipschitz functions that share the same boundary datum and converge to u both in W1,1 and in energy, no matter if u is a minimizer.

We consider here a convex nonautonomous Lagrangian F(x, u, ∇u), and establish a sufficient condition under which no Lavrentiev gap occurs at any admissible function. As a byproduct it turns out that the Lavrentiev phenomenon does not occur if the Lagrangian is of the form


with f(⋅, 0) of class C1(Ω) and h(⋅, 0) of class C2(Ω). The methods used here are mainly based on [12, 13, 14]: we show that we can approximate a function (both in W1,1 and in energy) with a sequence of bounded functions that are Lipschitz in a neighbourhood of the boundary of the domain.

A partial motivation for studying these kind of functionals comes from minimization problems in the Heisenberg group where one wants to consider functionals that generalize those studied in [15] and in references therein.

We do not consider here the vectorial case for which, when the Lagrangian depends only on the gradient, there are both examples of the occurrence of the Lavrentiev phenomenon and cases where it does not occur [7].

The authors are grateful to Pierre Bousquet for his useful comments and to both the referees for having carefully read the manuscript.

2 Notation and assumptions


  1. The scalar product of x, y in ℝn is denoted by 〈x, y〉.

  2. The pointwise maximum (resp. minimum) of two functions u, v is denoted by uv (resp. uv), u+ = u ∨ 0 (resp u = (–u)∨ 0) is the positive (resp. negative) part of u.

  3. The convex subgradient of a function g : ℝm → ℝ at ξ0 ∈ ℝm is the set

  4. The partial convex subgradient of F(x, s, ξ) with respect to x at (x0, s0, ξ0) is the convex subgradient of xF(x, s0, ξ0) at x = x0, it will denoted by xF(x0, s0, ξ0). Analogously we will denote by sF(x0, s0, ξ0) (resp. ξF(x0, s0, ξ0)) the partial convex subgradients of F(x, s, ξ) with respect to s (resp. ξ) at (x0, s0, ξ0). Also, the convex subgradient of (s, ξ) ↦ F(x, s, ξ) is denoted by s,ξF(x, s, ξ).

  5. For E ⊆ ℝn, |E| is the n-dimensional Lebesgue measure of E.

  6. 1E is the indicator function of a set E.

2.1 Assumptions

  1. Ω ⊂ ℝn is an open and bounded set.

  2. F : Ω × ℝ × ℝn → ℝ, (x, s, ξ) ↦ F(x, s, ξ) is a Carathéodory function, bounded below by 〈α(x), ξ〉 + β(x) for some αL1(Ω; ℝn), βL1(Ω).

  3. ϕ is a Lipschitz function on Ω.

  4. p ≥ 1, and for vϕ + W01,p (Ω) we define I(v) := Ω F(x, v, ∇ v) dx (the “energy”).

The following structure condition will be used, in alternative to the boundedness of the reference function u, in our main result.

2.2 Hypothesis (H)

Hypothesis (H)

  1. There are positive sequences (τk)k and (σk)k such that:

    1. limk+ τk = +∞, limk+ σk = +∞;

    2. For each k ∈ ℕ, there are selections (qτk(x), ζτk(x)) of s,ξF(x, τk, 0) and (qσk(x), ζσk(x)) of s,ξF(x, –σk, 0), and C ≥ 0 satisfying

      divζτk(x)C,divζσk(x)C. (2.1)

  2. There is a measurable and bounded selection q(x) of the subgradient sF(x, 0, 0) of F at (x, 0, 0) and, for all k ∈ ℕ,

    qσk(x)q(x)qτk(x)a.e. xΩ. (2.2)

Remark 2.1

  1. Condition (2.2) is satisfied if sF(x, s, ξ) is convex. Indeed the monotonicity of the subdifferential implies that, for a.e. xΩ,

  2. When F is of class C2(Ω) and (s, ξ) ↦ F(x, s, ξ) is convex for a.e. x, Hypothesis (H) reduces to Condition 1, namely that there are increasing, divergent sequences (τk)k and (σk)k such that


    for a suitable C ≥ 0.

Here are some Lagrangians that satisfy Hypothesis (H).

Proposition 2.2

(Validity of Hypothesis (H)). Assume that the map (s, ξ) ↦ F(x, s, ξ) is convex for a.e. x and that

  1. Either F(x, s, ξ) = F(s, ξ), i.e., F is autonomous, or

  2. F(x, s, ξ) = f(x, s) + h(x, ξ) for some Carathéodory functions f : Ω × ℝ → ℝ and h : Ω × ℝn → ℝ with xxf(x, 0) bounded and xh(x, 0) of class C2(Ω).

Then F fulfills Hypothesis (H).


Let (τk)k, (σk)k be arbitrary positive divergent sequences.

  1. Since ζτk, ζσk do not depend on x it turns out that their divergence is zero. Similarly, Point 2 of Hypothesis (H) is fulfilled since any qsF(0, 0) does not depend on x and F is convex.

  2. Assume now that F(x, s, ξ) = f(x, s) + h(x, ξ). Then, for each k,


    and div ∇ξh(x, 0) is continuous, thus bounded on Ω. Moreover, any element of sF(x, 0, 0) is an element of xf(x, 0) and is thus bounded. Condition (2.2) follows from the convexity of sF(x, s, ξ), proving the validity of Hypothesis (H).□

3 Approximation lemmas

In this section, we establish two preliminary results that will be used in the proof of Theorem 4.2.

As a first step, we give a sufficient condition under which there is no Lavrentiev gap between Wϕ1,p (Ω) and Wϕ1,p (Ω) ∩ L(Ω). We have defined the space Wϕ1,p (Ω) as the set of those functions uW1,p(Ω) such that the extension of u by ϕ on ℝnΩ belongs to Wloc1,p (ℝn). We still denote by u this extension. In particular, (uϕ) belongs to W1,p(ℝn) and has compact support.

Lemma 3.1

Let F : Ω × ℝ × ℝn → ℝ satisfy Hypothesis (H). Then for every u in Wϕ1,p (Ω) such that F(x, u, ∇u) ∈ L1(Ω), there exists a sequence (uk)k in Wϕ1,p (Ω) ∩ L(Ω) such that (uk)k converges to u in Wϕ1,p (Ω) and

limk+I(uk)=I(u). (3.1)


Let (τk)k and (σk)k satisfy the conditions formulated in Hypothesis (H). For k large enough such that both τk > |ϕ|L(Ω) and σk > |ϕ|L(Ω), we define uk by

uk(x)=(u+τk)(x)(uσk)(x)=u(x)if σku(x)τk,τkif u(x)τk,σkif u(x)σk.

It is clear that uk Wϕ1,p (Ω) ∩ L(Ω) and that uk converges to u in Wϕ1,p (Ω). Moreover,

I(uk)={σkuτk}F(x,u,u)dx+{uτk}F(x,τk,0)dx+{uσk}F(x,σk,0)dx. (3.2)

Let q(x) and (qτk(x), ζτk(x)) be as in Hypothesis (H). We have

F(x,u,u)F(x,τk,0)+qτk(x)(uτk)+ζτk(x)(uτk) a.e..

Moreover, from Point 2 of Hypothesis (H) we get

F(x,u,u)F(x,τk,0)+q(x)(uτk)+ζτk(x)(uk) a.e. on {uτk}.

Since (uτk)+ W01,1 (Ω), integration on {uτk} then gives


Therefore, Hypothesis (H) yields

{uτk}F(x,u,u)dx{uτk}F(x,τk,0)(q+C)udx. (3.3)

Analogously we get

{uσk}F(x,u,u)dx{uσk}F(x,σk,0)+(q+C)udx. (3.4)

It follows from (3.2), (3.3) and (3.4) that

I(uk){σkuτk}F(x,u,u)dx+{uτk}F(x,u,u)dx+{uσk}F(x,u,u)dx+(q+C){uσk}{uτk}|u|dxI(u)+(q+C){uσk}{uτk}|u|dx. (3.5)

Since uL1(Ω), Lebesgue’s Theorem implies that

lim supk+I(uk)I(u).

By Fatou lemma,

lim infk+I(uk)I(u)

and (3.1) follows.□

We now prove that there is no Lavrentiev gap at u Wϕ1,p (Ω) if u is Lipschitz continuous on a neighborhood of ∂Ω.

Lemma 3.2

Assume that F : Ω × ℝ × ℝn → ℝ is convex with respect to its three variables. Let u in Wϕ1,p (Ω) be such that F(x, u, ∇u) ∈ L1(Ω). Assume, moreover, that either u is bounded or that F fulfils Hypothesis (H). If u is Lipschitz continuous on a neighborhood of ∂Ω, then there exists a sequence (uk)k in Lipϕ(Ω) such that (uk)k converges to u in Wϕ1,p (Ω) and


Moreover, if u is bounded in L(Ω), then the sequence (uk)k may be taken to be bounded in L(Ω).


From Lemma 3.1 it is not restrictive to assume that u is bounded. We may consider u as extended by ϕ out of Ω. By assumption, there exists an open set V ⊂ ℝn such that ∂ΩV and u is Lipschitz continuous on VΩ. In particular u and ∇u are in L(VΩ).

Let ρ Cc (B1, ℝ+) be even, Rn ρ dx = 1 and for k = 1, 2, ..., (ρk)k be the sequence of mollifiers defined by ρk(x) := knρ(kx). Let also θ Cc (Ω, [0, 1]) be such that θ = 1 on a neighborhood of ΩV. We then define


Notice first that uk ∈ Lipϕ(Ω). Indeed, if θ = 1 then uk = u * ρk, otherwise {0 ≤ θ < 1} ⊂ VΩ where ∇u is bounded. Clearly, (uk)k converges to u in Wϕ1,p (Ω). This implies

lim infk+I(uk)I(u).

It remains to show that

lim supk+I(uk)I(u). (3.6)

For this purpose, we decompose I(uk) = Ω F(x, uk, ∇uk)dx as the sum

I(uk)={θ=1}F(x,uk,uk)dx+{0θ<1}F(x,uk,uk)dx. (3.7)

On the set {0 ≤ θ < 1} ⊂ VΩ, ∇uL(VΩ) and


Let k be such that


Then, for kk and x ∈ {0 ≤ θ < 1} we have


which in turn means that under the above assumptions both uk and ∇uk are bounded by a constant that does not depend on k. Since (uk)k converges to u in W1,p(Ω) we may assume, by taking a subsequence, that (uk, ∇uk)k converges a.e. to (u, ∇u). Now, since F is bounded on bounded sets, by Lebesgue’s Theorem we have

limk+{0θ<1}F(x,uk,uk)dx={0θ<1}F(x,u,u)dx. (3.8)

On the set {θ = 1} we have


It remains to show that

lim supk0{θ=1}F(x,uρk,uρk)dx{θ=1}F(x,u,u)dx; (3.9)

afterwards, in view of (3.7) and (3.8), we get (3.6). Notice that, since ρ is even,


By Jensen’s inequality,




Since F(x, u, ∇u) ∈ L1(Ω), we get (3.9).□

4 Main result

We consider here domains Ω that are locally strongly star-shaped in the sense of [12, Definition 2.9]. These include Lipschitz ones and allow even some cusps at some boundary points.

Definition 4.1

An open and bounded set Ω is called locally strongly star-shaped if for every p∂Ω, there exists an open set H ⊂ ℝn such that pH and HΩ is strongly star-shaped, i.e., there is zHHΩ such that zH + λ(ΩzH) is relatively compact in Ω for every λ ∈ [0, 1[.

Theorem 4.2

(Non-occurrence of the Lavrentiev gap). Assume that Ω is locally strongly star-shaped and that F : Ω × ℝ × ℝn → ℝ is convex. Let u Wϕ1,p (Ω) be such that F(x, u, ∇u) ∈ L1(Ω). Assume, moreover, that either u is bounded or that F satisfies the structure Hypothesis (H). Then the Lavrentiev gap for I does not occur at u, i.e. there exists a sequence (uk)k in Lipϕ(Ω) converging to u in W1,p(Ω) and such that

limk+I(uk)=I(u). (4.1)

Moreover, if u is bounded in L(Ω), the sequence (uk)k may be taken to be bounded in L(Ω).


In view of Lemma 3.2 it is enough to provide a sequence (uk)k in Wϕ1,p (Ω) satisfying the conditions of the claim with the exception that it is just Lipschitz continuous in a neighborhood of ∂Ω (instead of Lipschitz on Ω). We may consider u to be extended by ϕ out of Ω. Also, in view of Lemma 3.1, it is not restrictive to assume that u is bounded.

Without loss of generality, we can assume that F ≥ 0. Indeed, since F is convex with respect to its variables, if x0Ω and (a, q, ζ) ∈ ∂F(x0, 0, 0) then


Moreover additive affine terms do not perturb our convergence results: if a sequence (uk)k converges to u in W1,p(Ω) then (I(uk))k converges to I(u) if and only if Ω G(x, uk, ∇uk)dx converges to Ω G(x, u, ∇u)dx.

Consider first the case where Ω is strongly star-shaped with respect to the origin, i.e., for every h ∈ [0, 1[, is relatively compact in Ω. Given λ, h ∈ ]1/2, 1[, set


Notice that uhλ converges to u in W1,1 as λ, h → 1 and that uhλ = ϕ on ℝn. We then write


as a convex combination in λ, namely

x,uhλ,uhλ=λxh,uxh,λuxh+(1λ)x1λ/h1λ,ξhλ(x),ϕ(x)λϕxh1λ, (4.2)



The convexity of F yields

I(uhλ)λΩFxh,uxh,uxhdx+(1λ)ΩFx1λ/h1λ,ξhλ(x),ϕ(x)λϕxh1λdx (4.3)

Since F ≥ 0, we get


Fix λ ∈ ]1/2, 1[; we then study the second term of the right hand side of (4.3). Since h ≥ 1/2, we have





limh1x1λ/h1λ=x,limh1ξhλ(x)=ϕ(x) a.e.



The function F being bounded on bounded sets, by means of the dominated convergence theorem we get


so that

lim suph1I(uhλ)λI(u)+(1λ)I(ϕ).

The right-hand side term of the latter inequality tends to I(u) as λ tends to 1. Hence, for every i ∈ ℕ, i ≥ 1, there are sequences λi and ki ∈ ℕ with kii such that ukiλi u in W1,p(Ω) as i → +∞, ukiλi are Lipschitz in a neighbourhood of ∂Ω, and


In particular we get

lim supi+I(ukiλi)I(u).

Also, Fatou’s lemma gives

lim infi+I(ukiλi)I(u),

and thus limi+ I( ukiλi ) = I(u), proving the claim.

The case of a general locally strongly star-shaped domain follows with the obvious changes as in the proof of [12, Theorem 4.1].□

Remark 4.3

Assume that F(x, u, ∇u) = f(x, u) + h(∇u) with f convex and superlinear. In Theorem 4.2 the (alternative) assumption that u is bounded is satisfied if, for instance, for every constant boundary datum k ∈ ℤ, the minimizers of I among the functions that are equal to k in the boundary of Ω are bounded. Indeed, the fact that ϕ is bounded and the comparison principles of [16, 17] show that u is bounded too.


This research is partially supported by the Padua University grant SID 2018 “Controllability, stabilizability and infimum gaps for control systems”, prot. BIRD 187147.


[1] Mikhail Alekseevich Lavrentiev, Sur quelques problèmes du calcul des variations, Ann. Mat. Pura Appl. 4 (1927), 9–28.10.1007/BF02409983Search in Google Scholar

[2] Basilio Manià, Sopra un esempio di Lavrentieff, Boll. Unione Mat. Ital. 13 (1934), 147–153.Search in Google Scholar

[3] John MacLeod Ball and Victor J. Mizel, One-dimensional variational problems whose minimizers do not satisfy the Euler-Lagrange equation, Arch. Rational Mech. Anal. 90 (1985), no. 4, 325–388.10.1007/978-3-642-61598-6_16Search in Google Scholar

[4] Paolo Baroni, Maria Colombo, and Giuseppe Mingione, Regularity for general functionals with double phase, Calc. Var. Partial Differential Equations 57 (2018), no. 2, art. 62.10.1007/s00526-018-1332-zSearch in Google Scholar

[5] Giovanni Alberti and Francesco Serra Cassano, Non-occurrence of gap for one-dimensional autonomous functionals, Calculus of variations, homogenization and continuum mechanics (Marseille, 1993), vol. 18 of Ser. Adv. Math. Appl. Sci., pp. 1–17, World Sci. Publ., River Edge, NJ, 1994.Search in Google Scholar

[6] Giovanni Bonfanti and Arrigo Cellina, On the non-occurrence of the Lavrentiev phenomenon, Adv. Calc. Var. 6 (2013), no. 1, 93–121.Search in Google Scholar

[7] Jan Malý and William P. Ziemer, Fine regularity of solutions of elliptic partial differential equations, vol. 51 of Mathematical Surveys and Monographs, American Mathematical Society, Providence, RI, 1997.10.1090/surv/051Search in Google Scholar

[8] Pierre Bousquet, Carlo Mariconda, and Giulia Treu, Non-occurrence of the Lavrentiev gap for a class of nonconvex Lagrangians, in preparation (2020).Search in Google Scholar

[9] Giuseppe Buttazzo and Marino Belloni, A survey on old and recent results about the gap phenomenon in the calculus of variations, Recent developments in well-posed variational problems, vol. 331 of Math. Appl., pp. 1–27, Kluwer Acad. Publ., Dordrecht, 1995.10.1007/978-94-015-8472-2_1Search in Google Scholar

[10] Ivar Ekeland and Roger Témam, Convex analysis and variational problems, vol. 28 of Classics in Applied Mathematics, Society for Industrial and Applied Mathematics (SIAM), Philadelphia, PA, english edition, 1999.10.1137/1.9781611971088Search in Google Scholar

[11] Giovanni Bonfanti and Arrigo Cellina, The nonoccurrence of the Lavrentiev phenomenon for a class of variational functionals, SIAM J. Control Optim. 51 (2013), no. 2, 1639–1650.10.1137/12086618XSearch in Google Scholar

[12] Pierre Bousquet, Carlo Mariconda, and Giulia Treu, On the Lavrentiev phenomenon for multiple integral scalar variational problems, J. Funct. Anal. 266 (2014), 5921–5954.10.1016/j.jfa.2013.12.020Search in Google Scholar

[13] Carlo Mariconda and Giulia Treu, Non-occurrence of a gap between bounded and Sobolev functions for a class of nonconvex Lagrangians, J. Convex Analysis 27 (2020), no. 4, to appear.Search in Google Scholar

[14] Pierre Bousquet, Carlo Mariconda, and Giulia Treu, A survey on the non occurence of the Lavrentiev gap for convex, autonomous multiple integral scalar variational problems, Set-Valued Var. Anal. 23 (2015), no. 1, 55–68.10.1007/s11228-014-0305-4Search in Google Scholar

[15] Andrea Pinamonti, Francesco Serra Cassano, Giulia Treu, and Davide Vittone, BV minimizers of the area functional in the Heisenberg group under the bounded slope condition, Ann. Sc. Norm. Super. Pisa Cl. Sci. 14 (2015), no. 3, 907–935.Search in Google Scholar

[16] Carlo Mariconda and Giulia Treu, A comparison principle and the Lipschitz continuity for minimizers, J. Convex Anal. 12 (2005), no. 1, 197–212.Search in Google Scholar

[17] Carlo Mariconda and Giulia Treu, A Haar-Rado type theorem for minimizers in Sobolev spaces, ESAIM Control Optim. Calc. Var. 17 (2011), 1133–1143.10.1051/cocv/2010038Search in Google Scholar

Received: 2020-01-02
Accepted: 2020-01-10
Published Online: 2020-02-17

© 2020 Carlo Mariconda and Giulia Treu, published by De Gruyter

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

Scroll Up Arrow