Nonex Ade
Nonex Ade
Nonex Ade
1. Introduction
In this paper we consider the following nonlinear partial differential in-
equality of elliptic type:
−∆A u ≥ Φ(u), u ≥ 0 a.e., (1.1)
where ∆A u = div(A(∇u)) is the A-Laplacian. Under certain assumptions
on A, Φ, u we prove nonexistence of nonnegative solutions of (1.1). The
assumptions can be sketched as follows.
(A): A : Rn → Rn is a given function such that A(λ) = Ā(|λ|)
|λ|2
λ for λ 6= 0,
where Ā(s) : [0, ∞) → [0, ∞) is a convex function with Ā(0) = 0, satisfying
the ∆0 -condition:
Ā(s1 s2 ) ≤ C Ā(s1 )Ā(s2 ),
with a universal constant C. The typical example of such an A-Laplacian
is the usual p-Laplacian, ∆p u = div(|∇u|p−2 ∇u), corresponding to Ā(λ) =
|λ|p . Another example is the logarithmic Laplacian, where we take Ā(s) =
|s|p (ln(2 + s))α , α ≥ 0.
with some parameter s ∈ (0, 1). The function Ψ̃ is chosen in such a way
that Φ(·)Ψ̃(·) grows slower than Φ(·). Inequality (1.6) is then improved in
Section 5 (as a third step of our reasoning), where we obtain the estimate
Z
Φ(u)φ(x) dx ≤ L(φ)dx,
Rn
Nonexistence results for differential inequalities 311
Note that the authors of papers [40, 49, 53] deal with inequality (1.7),
which is the special case of (1.1), restricted to Φ(s) = sq .
In another paper [23], Section 3, the author applies the techniques of
Mitidieri and Pohozaev to the equation in the form
−div (h(x)g(u)A(|Du|Du)) ≥ f (x, u, Du) in RN , u ≥ 0,
where one assumes that f (x, z, ξ) ≥ a(x)z q |ξ|θ in RN × [0, ∞) × RN , with
nonnegative a. We do not require any special behavior on the function Φ
involved in (1.1).
Let us remark that existence/nonexistence results on Rn differ signifi-
cantly from those on bounded regular domains. For example, according to
the result [49], there are no nonnegative solutions of (1.2) on the entire space
Rn when p − 1 < q < n(p−1) n−p , p < n. On the other hand, in this range of
parameters, they do exist on bounded starshaped domains, even with zero
boundary data [54].
For the related existence results we refer for example to [26] and their ref-
erences. For results concerning parabolic problems, inequalities for systems,
higher-order operators, or inequalities on the Heisenberg group, we refer e.g.
to [8, 10, 11, 14, 16, 21, 22, 24, 28, 29, 47, 48, 50, 51, 52, 56, 57, 58, 60, 64, 65,
66, 68, 69]. Motivations to consider PDE’s in the Orlicz setting that come
from mathematical physics can be found for example in [1, 4, 15, 32, 46].
Logarithmic Orlicz functions and logarithmic Sobolev spaces are an object
of particular interest, see e.g. [5, 19, 33, 35]. Finally, let us also mention
that the derivation of nonexistence results in the Orlicz setting was stated
as an open problem by the authors of the influential paper [49], see remarks
on page 132 in their later paper [53].
0
Fact 2.2. If an N -function F satisfies the ∆ -condition, then H = F −1
satisfies the inequality
H(s1 )H(s2 ) ≤ B̄H(s1 s2 ),
1
with constant B̄ = H( 1
)
, independent of s1 , s2 .
D2
F
and Z
Ā (|∇w (x) − ∇w(x)|) dx → 0, as → 0.
Rn
We explain the second convergence only. We have |∇w (x) − ∇w(x)| → 0
almost everywhere and so Ā (|∇w (x) − ∇w(x)|) → 0 almost everywhere.
Moreover, by the monotonicity of Ā and the ∆2 −condition we deduce that
Ā(|a − b|) ≤ Ā(2|a|) + Ā(2|b|) ≤ C Ā(|a|) + Ā(|b|) ,
with some constant C depending only on Ā. Hence,
Ā (|∇w (x) − ∇w(x)|) ≤ C Ā (|∇w (x)|) + C Ā (|∇w(x)|) ,
with some general constant C. Moreover, we have
Z
Ā |∇w (x)| ≤ Ā ((|∇w| ∗ φ )(x)|) = Ā |∇w(x − y)|φ (y)dy
Rn
Z
≤ Ā (|∇w(x − y)|) φ (y)dy = Ā (|∇w|) ∗ φ (2.6)
Rn
because Ā is convex and φ (x)dx is a probability measure. The function in
(2.6) is of the form k = k∗φ , where k := Ā (|∇w|) belongs to L1 . Therefore,
k → k strongly in L1 and we deduce that Ā (|∇w (x)|) + C Ā (|∇w(x)|) is
dominated by functions Ck + Ck, which converge in L1 . We complete the
argument by using the Lebesgue Dominated Convergence Theorem.
Using the fact that Ā satisfies the ∆2 −condition and Theorem 9.4 in [44],
we deduce that w converges in the norm to w in W 1,Ā (Rn ). Moreover, all
functions w are smooth, nonnegative and compactly supported. Therefore,
inequality (2.5) can be tested as well with nonnegative smooth and compactly
supported functions w.
Nikodym ACL Characterization Theorem. We will need the following
variant of the Nikodym ACL Characterization Theorem. Its proof is an
easy modification of the classical Nikodym ACL Characterization Theorem,
which can be found e.g. in [46], Section 1.1.3.
Theorem 2.1 (Nikodym ACL Characterization Theorem).
1,1
i) Let u ∈ Wloc (Rn ). Then for every i ∈ {1, . . . , n} and for almost every
a ∈ Rn−i−1 × {0} × Ri the function
R 3 t 7→ u(a + tei ) (2.7)
is locally absolutely continuous on R; in particular, for almost every point x ∈
Rn the distributional derivative Di u(x) is the same as the usual derivative
at x.
Nonexistence results for differential inequalities 317
ii) If for every i ∈ {1, . . . , n} and for almost every a ∈ Rn−i−1 × {0} × Ri
the function (2.7) is locally absolutely continuous on R, and all the deriva-
tives Di u computed almost everywhere, together with the function u, are
1,1
locally integrable on Rn , then u belongs to Wloc (Rn ).
1,1
iii) Let u ∈ Wloc (Rn ) and Ω ⊂ Rn be an open subset. Then u belongs
to W 1,Ā (Ω) if and only if every derivative Di u computed almost everywhere,
together with the function u, belongs to the space LĀ (Ω).
Remark 2.2. As a direct consequence of the above theorem we obtain that
1,Ā 0
if u belongs to Wloc (Rn ) and Ψ is locally Lipschitz, and if moreover Ψ is
1,Ā
finite or u is locally finite, then the composition Ψ ◦ u belongs to Wloc (Rn )
as well.
Remark 2.3. Note that inequality (2.5) implies that Φ(u)w ∈ L1 (Rn ) for
every nonnegative and compactly supported w, whenever u satisfies (2.4). As
for every w ∈ W 1,Ā (Rn ) we have |w| ∈ W 1,Ā (Rn ) (this follows from Theorem
2.1), inequalities (2.4), (2.5) and Fact 2.3 imply that Φ(u)w ∈ L1 (Rn ) for
every compactly supported w ∈ W 1,Ā (Rn ).
Having an arbitrary u ∈ Wloc1.1 (Rn ) it is possible to define its value at every
(2) the mapping t 7→ Φ(t)Φs,A (t) is bounded on each interval [0, R],
R > 0;
(Fs ) there exist a positive constant D and an N -function F such that F ∗
0
satisfies the ∆ -condition and
Ā( Φ t (t) ) Φ(t) 1/(1−s)
s,A
F t ≤ D for every t > 0, (2.11)
Φ (t)
tB(t)
s,A
3. Caccioppoli-type estimates
The proof of our main result will be based on the following proposition,
which seems to be of separate interest. Inequality (3.1) involves ∇u on the
left-hand side and only u on the right-hand side. Because of that property
we call it a Caccioppoli-type inequality (see e.g. [13, 37]).
Proposition 3.1. Assume that
(i) Ā is an N -function which satisfies the ∆0 -condition,
(ii) Ψ̃ : [0, ∞) → [0, ∞) is decreasing, locally Lipschitz on (0, ∞), and
0
tΨ̃ (t) ≤ −C Ψ̃(t) almost everywhere, with C > 0 independent of t,
Nonexistence results for differential inequalities 319
Before we prove the proposition let us state the following lemma. Applying
it with t = 0, we deduce that the second integral in (3.1) is well defined. For
readers’ convenience we submit its proof in the Appendix.
1,1
Lemma 3.1. Let u ∈ Wloc (Rn ) be defined everywhere by the formula (2.8)
and let t ∈ R. Then
{x ∈ Rn : u(x) = t} ⊆ {x ∈ Rn : ∇u(x) = 0} ∪ N (3.2)
where |N | = 0.
Now we are in a position to prove Proposition 3.1.
Proof of Proposition 3.1. For u, φ as in the assumptions of the proposition
we fix 0 < δ < R and denote
uδ,R (x) := min (u(x) + δ, R) , G(x) := Ψ̃(uδ,R (x))φ(x). (3.3)
The function G is nonnegative and compactly supported. Applying Theorem
1,Ā
2.1, we verify that uδ,R ∈ Wloc (Rn ). Because δ ≤ uδ,R ≤ R and Ψ̃ is locally
1,Ā
Lipschitz, we also have Ψ̃(uδ,R ) ∈ Wloc (Rn ). Since Ψ̃ is not necessarily
1,Ā
globally Lipschitz, it might happen that Ψ̃(u) does not belong to Wloc (Rn ),
and this is why we had to introduce truncated functions uδ,R . It follows that
G ∈ W01,Ā (Rn ) ⊂ W 1,Ā (Rn ). In particular we see that its weak derivative
∇G can be computed almost everywhere, and that for almost every x we
have
0
∇G(x) = Ψ̃ (u(x) + δ)∇u(x) · χ{u(x)≤R−δ} · φ(x) + Ψ̃(uδ,R (x))∇φ(x)
320 A. Kalamajska, K. Pietruska-Paluba, and I. Skrzypczak
0
(to this goal, use Theorem 2.1 again). Since Ψ̃ is locally bounded, both
functions on the right-hand side above belong to LĀ (Rn ).
From now on we proceed in three steps.
Step 1. We show that under our assumptions we have for every 0 < δ < R
Z Z Ψ̃(u + δ)
Φ(u)Ψ̃(u + δ)φ dx + (C − ) Ā(|∇u|) φ dx
{u≤R−δ} {∇u6=0, u≤R−δ} u+δ
|∇φ| u + δ Ψ̃(u + δ)
Z
≤ Ā( ) φ dx + C̃(δ, R), (3.4)
supp φ∩{∇u6=0, u≤R−δ} φ u+δ
where
hZ Z i
C̃(δ, R) = Ψ̃(R) B(|∇u|)h∇u, ∇φi dx− Φ(u)φ dx .
supp φ∩{∇u6=0, u>R−δ} {u>R−δ}
R
Note that, according to Remark 2.3, the integral Rn Φ(u)φ dx is finite.
Before we start the proof of (3.4), we introduce some auxiliary notation:
Z
0
Ã(δ, R) = Ā(|∇u|)Ψ̃ (u + δ)φ dx,
{∇u6=0, u≤R−δ}
Z Ψ̃(u + δ)
Ã1 (δ, R) = Ā(|∇u|) φ dx,
{∇u6=0, u≤R−δ} u+δ
Z
B̃(δ, R) = B(|∇u|)h∇u, ∇φiΨ̃(u + δ) dx,
{∇u6=0, u≤R−δ}
Z
C̃1 (δ, R) = Ψ̃(R) Φ(u)φ dx,
{u>R−δ}
Z
C̃2 (δ, R) = Ψ̃(R) B(|∇u|)h∇u, ∇φi dx,
{∇u6=0, u>R−δ}
Z |∇φ| u + δ Ψ̃(u + δ)
D̃(, δ, R) = Ā φ dx.
supp φ∩{∇u6=0, u≤R−δ} φ u+δ
Then we note that
Z Z
I := Φ(u)G dx = Φ(u)Ψ̃(uδ,R )φ dx (3.5)
n Rn
ZR Z
= Φ(u)Ψ̃(u + δ)φ dx + Ψ̃(R) Φ(u)φ dx
{u≤R−δ} {u>R−δ}
Z
= Φ(u)Ψ̃(u + δ)φ dx + C̃1 (δ, R).
{u≤R−δ}
Nonexistence results for differential inequalities 321
As B(|∇u|)|∇u| = Ā(|∇u|) |∇u| , we can apply Lemma 2.1 with s1 = |∇u| and
|∇φ| uδ,R
s2 = φ to get
Z
Ψ̃(u + δ)
B̃(δ, R) ≤ Ā(|∇u|) φ dx
{∇u6=0, u≤R−δ} u+δ
Z |∇φ| u + δ Ψ̃(u + δ)
+ Ā φ dx = Ã1 (δ, R) + D̃(, δ, R).
{∇u6=0, u≤R−δ} φ u+δ
Combining these estimates we get that
Z
I≤ Φ(u)Ψ̃(u + δ)φ dx + C̃1 (δ, R) ≤ − C Ã1 (δ, R) + B̃(δ, R) + C̃2 (δ, R)
{u≤R−δ}
Note that Ã1 (δ, R), D̃(, δ, R), and consequently C̃1 (δ, R) are finite. This
and (3.5) imply
Z
Φ(u)Ψ̃(u + δ)φ dx + (C − )Ã1 (δ, R)
{u≤R−δ}
where D(R) := supt∈[0,R] Ā(t) Ψ̃(t)t and DĀ is the constant from the ∆0 −
condition for Ā. The quantity D(R) is finite by assumption (iii), and inde-
pendent of δ. Note that
|∇φ| u + δ Ψ̃(u + δ)
χ{|∇u|6=0, u≤R−δ} · Ā
φ u+δ
|∇φ| u Ψ̃(u)
δ→0
→ χ{|∇u|6=0, u≤R} · Ā , a.e.
φ u
This follows from Lemma 3.1 (which implies that the sets {u = 0, |∇u| =
6 0}
and {u = R, |∇u| = 0} are of measure zero) and the continuity outside
Nonexistence results for differential inequalities 323
zero of the functions involved. As we have assumed that Ā( |∇φ| φ )φ is inte-
grable over the set supp φ∩{∇u 6= 0}, the Lebesgue Dominated Convergence
Theorem yields
Z |∇φ| u Ψ̃(u)
δ→0
D̃(, δ, R) → D̃(, R) := Ā φ dx.
supp φ∩{∇u6=0, u≤R} φ u
Our argument shows in particular that D̃(, R) is finite for any R > 0.
The left-hand side of (3.4) is easier to handle. Both integrands are in-
creasing when δ decreases to zero (as both Ψ̃(t), Ψ̃(t) t are decreasing – by
assumption (ii)), so we can pass to the limit as δ → 0 under the integral
signs, by applying the Monotone Convergence Theorem. We obtain (3.6).
Step 3. We complete the proof of the proposition by letting R → ∞ in
(3.6).
Without loss of generality we can assume that the integral in the right-
hand side of (3.1) is finite, as otherwise the inequality follows trivially. There-
fore, (3.1) follows from (3.6) by the Monotone Convergence Theorem, once
we note that limR→∞ C(R) e = 0, due to the integrability of B(|∇u|)h∇u, ∇φi
and Φ(u)φ. The proof is complete.
Remark 3.1. Introduction of the parameters δ and R was necessary as
we needed to move some quantities in the estimates to opposite sides of
inequalities. To legitimately do that, we must make sure that they are
finite.
where
Φ(u) 1
1−s
Ψ1,s (u) = Φ(u)Φs,A (u) = ,
uB(u)
Φs,A (u) (uΦ(u) )s 1−s
1
Ψ2,s (u) = = .
u Ā(u)
In the next section the above inequality will be improved to a better one.
Remark 4.1. Let us note that
Z |∇φ| 1 Z |∇φ| 1
1−s 1−s
C̄ Ā φ dx ≤ C̄ Ā φ dx.
supp φ∩{∇u6=0} φ supp φ φ
In particular (4.1) implies
Z Z
Ψ1,s (u)φ(x) dx + Ā(|∇u|)Ψ2,s (u)φ dx
Rn {∇u6=0}
Z |∇φ| 1
1−s
≤ C̄ Ā φ dx,
supp φ φ
where the right-hand side does not involve u.
Proof. Without loss of generality we can assume that the right-hand side
integral in (4.1) is finite. For s ∈ (0, 1) the integral
Z |∇φ|
Ā φ dx
supp φ∩{∇u6=0} φ
will be finite as well. As
Ā(t)
Φs,A (t) = (Φ(t)Φs,A (t))s , (4.2)
t
we observe that the function t 7→ Φs,A (t) satisfies all the assumptions of
Proposition 3.1. Therefore we can apply (3.6) for Ψ̃ = Φs,A , with = C2 and
an arbitrary R > 0. We obtain
Z Z Φ (u)
C s,A
Φ(u)Φs,A (u)φ dx + Ā(|∇u|) φ dx (4.3)
{u≤R} 2 {∇u6=0, u≤R} u
C
Z 2 |∇φ| Φ (u)
s,A
≤ Ā u φ dx + C̃(R) := L,
2 supp φ∩{∇u6=0,u≤R} C φ u
Nonexistence results for differential inequalities 325
where C̃(R) is the same as in (3.6), with all the quantities involved being
finite. The ∆0 −condition for Ā implies that
2 |∇φ| 2 2 |∇φ|
Ā( u) ≤ DĀ Ā( )Ā( )Ā(u),
C φ C φ
and so the right-hand side integral in (4.3) is not bigger than
2
Z |∇φ| Ā(u)Φ (u)
2 s,A
DĀ Ā( ) Ā φ dx.
C supp φ∩{∇u6=0,u≤R} φ u
Taking into account the relation (4.2) and applying the classical Young in-
equality
1 1
ab ≤ (1 − s)a 1−s + sb s
to the numbers
1 |∇φ|
a= Ā , b = κ(Φ(u)Φs,A (u))s ,
κ φ
with arbitrary κ > 0, we obtain from (4.3) that
h 1
Z |∇φ| 1
2 1−s
L ≤ DĀ Ā C2 C2 κ1 1−s (1 − s) Ā φ dx
supp φ∩{∇u6=0, u≤R} φ
Z i
1
+κ s s Φ(u)Φs,A (u)φ dx + C(R),
e
{u≤R}
where C(R)
e was given by (3.7).
1
2 2 C 1
s
Choose now κ = κ0 in such a way R that κ0 sDĀ Ā C 2 = 2 . After rear-
ranging (which is permitted since {u≤R} Φ(u)Φs,A (u)φ dx < ∞) we obtain
Z Z
Φs,A (u)
Φ(u)Φs,A (u)φ dx + C Ā(|∇u|) φ dx
{u≤R} {∇u6=0, u≤R} u
Z |∇φ| 1
1−s
≤ C̄ Ā φ dx + 2C̃(R), (4.4)
supp φ∩{∇u6=0, u≤R} φ
1 1
where C̄ = DĀ2 Ā( 2 )C 1 ) 1−s (1 − s), κ s sD 2 Ā( 2 ) C = 1 .
C κ0 0 Ā C 2 2
It only remains to pass to the limit as R → ∞ in (4.4). We know that
C̃(R) → 0 when R → ∞ (see Step 3 in the proof of Proposition 3.1), and we
1
have assumed that the integral of (Ā( |∇φ| φ ))
1−s φ over the set supp φ ∩ {∇u 6=
1 n
Z Φ(u) 1 Z Ā( |∇φ| ) o
2 1−s ∗ φ
II ≤ DĀ Ā( )δ D φ dx + F φ dx
{∇u6=0} uB(u) supp φ δ
|∇φ|
Z
1 n 1 o 1
2
≤ DĀ Ā( ) δDa + DF ∗ δF ∗ ( ) F ∗ Ā( ) φ dx =: DĀ 2
Ā( )pδ .
δ supp φ φ
Note that pδ is finite. When pδ = 0, then the estimate follows trivially, and
when pδ 6= 0 we get the inequality
2 1
L ≤ C̄a + DĀ Ā( )pδ , (5.2)
valid for every > 0, δ > 0. We choose = (δ) satisfying
1 C̄a
= Ā−1 ,
(DĀ )2 pδ
so that both terms in the right-hand side in (5.2) are the same. This gives
2C̄a
L ≤ I + II ≤ C̄a
, (5.3)
Ā−1 2p
DĀ δ
2C̄a 2C̄ ab
L≤ = · b. (5.4)
C̄(F ∗ )−1 ( DDa∗ b ) C̄(F ∗ )−1 ( DD ∗ · ab )
Ā−1 2D
2DĀ
F
Ā−1 2
2DĀ
F
Note that according to Fact 2.2 both the functions (F ∗ )−1 and Ā−1 satisfy
the inequality B̄H(λ1 λ2 ) ≥ H(λ1 )H(λ2 ), with a suitably chosen constant
B̄. The result follows.
328 A. Kalamajska, K. Pietruska-Paluba, and I. Skrzypczak
We have the following observations, whose proofs are given in the Appendix.
0
Lemma 6.1. Let M be a N -function satisfying the ∆ -condition. Then there
exists λ ≥ 0 such that
R1
(1) 0 M ( 1s )sλ ds < ∞,
|∇φR,λ |
φR,λ dx ≤ C(λ)Rn M R1 ,
R
(2) R≤|x|≤2R M φR,λ
where
Z 1
n−1 2
C(λ) = θn−1 2 DM M (λ) · M ( 1s )sλ ds ,
0
θn−1 is the surface of the unit sphere S n−1 in Rn .
Lemma 6.2. Let M be an arbitrary strictly convex function. Then the
a a
mappings R+ 3 a 7→ M −1b ( a ) · b and R+ 3 b 7→ M −1b ( a ) · b are increasing.
b b
The basic tool for our final consideration will be the following function:
a (R) t
λ
Hλ (R) = R · bλ (R), where R(t) = (6.2)
bλ (R) (F ◦ Ā)−1 (t)
∗
where φR,λ was defined by (6.1), with some constants Ē, L̄ independent of R.
R For the proof of part ii) we note that, as limR→∞ F(R) = 0, we get
Rn Φ(u)dx = 0. This is impossible for a nonzero u. The theorem is proven.
q
p− +1
We choose s ∈ ( q+1p−1
, p−1
q ). It is immediate to see that, when q > p−1,
then this interval in nonempty and contained in (0, 1).
Condition (A) is obvious. The function Φs,A in the present case is equal
to Mκ,γ (t), where
qs − (p − 1) βs − α
κ = κ(s) = , γ = γ(s) = . (7.1)
1−s 1−s
Inequality (2.10) from condition (Φs,A ) is tantamount to the fact that
Φs,A (t)tC is decreasing. Finding such C > 0 is possible if and only if κ <
0, i.e., when s < p−1 q . Since Φ(t)Φs,A (t) is also of the logarithmic form
considered, condition 2) of (Φs,A ) will be satisfied as well, since the argument
t is raised to the power 1+q−p
1−s > 0 there.
To verify condition (Fs ), observe that (2.11) is of the form F (Mr1 ,β1 ) ≤
DMr2 ,β2 , with some parameters r1 , r2 , β1 , β2 (they are listed below). Because
of the nature of this relation, we seek F in logarithmic form also. Fact 7.1
implies that
r2 r1 β2 − r2 β1
F (t) = tµ log(2 + t)η , with µ = , η = (7.2)
r1 r1
does the trick. We have two additional requirements: F is supposed to be
convex, and F ∗ should satisfy the ∆0 −condition. They will hold when µ > 1,
η ≤ 0, which is exactly
r2 > r1 , β2 r1 ≤ β1 r2 . (7.3)
Direct computation gives
r1 = r1 (s) = (p − 1)(1 − κ), β1 = β1 (s) = −(p − 1)γ + α,
q − (p − 1) β−α
r2 = r2 (s) = , β2 = β2 (s) = . (7.4)
1−s 1−s
q
p−
p−1
+1
The first condition of (7.3) is satisfied when s > q+1 , and the second
condition reads
(β − α) (q − (p − 1))
(p − 1)(1 − κ) ≤ (α − (p − 1)γ) . (7.5)
(1 − s) (1 − s)
Substituting the values of κ and γ from (7.1) and rearranging we get
(p − 1)(β − α)[p − s(q + 1)] ≤ [(1 − s)α − (p − 1)(βs − α)],
αq
which boils down to q(β − α) ≤ p−1 + β(q − p). This is nothing but the
q
condition β ≤ α p−1 , which we have assumed.
Nonexistence results for differential inequalities 331
Our next observation is that all the functions involved in the definition
of F are of logarithmic form Mp,α , and that their argument is r = R1 . The
condition F(R) → 0 when R → ∞ can be written as
(Ā(r))1/1−s
lim = 0. (7.6)
r→0 r n (F ∗ ◦ Ā)−1 Ā1/1−s (r)
F ∗ ◦Ā
Since Mp,α (t) ∼ Mp,0 (t) when t < 1, the behavior of all the functions in (7.6)
when r → 0 is exactly the same as in the homogeneous case, and so we can
suppress the logarithmic part of the functions Ā and F, without changing
the asymptotics we examine. After some elementary algebra we obtain that,
for R large, F(R) ∼ Rσ , where
1 r1
µ0 −p (p − 1) + r2 (q + 1)(p − 1)
σ = n−1+ = (n − 1) − = n−1− . (7.7)
1−s 1−s q+1−p
n(p−1)
We see that, when q < n−p , then σ is negative, and so F(R) → 0 when
R → ∞. We are done.
Remark 7.1. Taking α = β = 0 in Proposition 7.1, we retrieve the result
of Mitidieri and Pohozaev from [49], asserting the nonexistence of nontrivial
nonnegative solutions of the inequality −∆p u ≥ uq (except for the critical
case q = n(p−1)
n−p ).
set of Lebesgue points of PA (a) in [−R, R], i.e., all points x0 ∈ PA (a), for
which we have Z
1
lim χPA (a) (x) dx = 1. (8.1)
→0 2 (x −,x +)
0 0
0
As |PA (a)| = |PA∗ (a)|, we deduce that the set Q∗A := {(a, t) : a ∈ Q̃ , t ∈
PA∗ (a)} ⊆ QA is of full measure in QA . Moreover, no point in PA∗ (a) is an
isolated point, as (8.1) applies to PA∗ (a) as well. For every (a, t) ∈ Q∗A , we
have
∂u u(a, t + h) − u(a, t) s̄ − s̄
(a, t) = lim ∗ = lim ∗ = 0.
∂xn h→0,t+h∈PA (a) h h→0,t+h∈PA (a) h
References
[1] H.-D. Alber, “Materials with Memory. Initial-Boundary Value Problems for Constitu-
tive Equations with Internal Variables,” Lecture Notes in Mathematics 1682. Springer,
Berlin 1998.
[2] A.D. Alexandrov, A characteristic property of the spheres, Ann. Mat. Pura Appl., 58
(1962), 303–315.
[3] A. Ambrosetti, J. Garcia Azorero, and I. Peral, Perturbation of ∆u + u(N +2)/(N −2) =
0, the scalar curvature problem in RN , and related topics, J. Funct. Anal., 165 (1999),
117–149.
[4] J.M. Ball, Convexity conditions and existence theorems in nonlinear elasticity, Arch.
Rat. Mech. Anal., 63 (1977), 337–403.
[5] C. Bennett and K. Rudnick, On Lorentz-Zygmund spaces, Dissertationes Math.
(Rozprawy Mat.), 175 (1980), 1–72.
[6] H. Berestycki, Qualitative properties of semilinear elliptic equations in unbounded
domains, in: “Progress in Elliptic and Parabolic Partial Differential Equations (Capri,
1994),” Pitman Res. Notes Math. Ser., Vol. 350, pp. 19-42, Longman, Harlow, 1996.
[7] H. Berestycki, L. Caffarelli, and L. Nirenberg, Further qualitative properties for ellip-
tic equations in unbounded Lipschitz domains, Comm. Pure Appl. Math., 50 (1997),
1089–1111.
[8] M.-F. Bidaut-Veron and S. Pohozaev, Nonexistence results and estimates for some
nonlinear elliptic problems, J. Anal. Math., 84 (2001), 1–49.
334 A. Kalamajska, K. Pietruska-Paluba, and I. Skrzypczak
[9] B. Bojarski and P. Hajlasz, Pointwise inequalities for Sobolev functions and some
applications, Studia Math., 106 (1993), 77–92.
[10] H. Brézis and X. Cabré, Some simple nonlinear PDE’s without solutions, Boll. Unione
Mat. Ital. Sez. B Artic. Ric. Mat., 8 (1998), 223–262.
[11] H. Brézis and L. Nirenberg, Positive solutions of nonlinear elliptic equations involving
critical Sobolev exponents, Comm. Pure Appl. Math., 36 (1983), 437–477.
[12] H. Brézis and J.L. Vasquez, Blow-up solutions of some nonlinear elliptic problems,
Rev. Mat. Complut., 10 (1997), 443–469.
[13] R. Caccioppoli, Limitazioni integrali per le soluzioni di un’equazione lineare ellitica a
derivate parziali, Giorn. Mat. Battaglini, 80 (1951), 186–212.
[14] G. Caristi, Existence and nonexistence of global solutions of degenerate and singular
parabolic systems, Abstr. Appl. Anal., 5 (2001), 373–383.
[15] A. Cianchi, Some results in the theory of Orlicz spaces and applications to variational
problems. Nonlinear analysis, function spaces and applications, Vol. 6 (Prague, 1998),
Acad. Sci. Czech Rep., Prague 1999, pp. 50–92.
[16] P. Clement, R. Manasevich, and E. Mitidieri, Positive solutions for a quasilinear
system via blow up, Comm. Partial Differential Equations, 18 (1993), 2071–2106.
[17] E.N. Dancer, Some notes on the method of moving planes, Bull. Austral. Math. Soc.,
46 (1992), 425–434.
[18] G.H. Derrick, Comments on nonlinear wave equations as models for elementary par-
ticles, J. Mathematical Phys., 5 (1964), 1252–1254.
[19] D.E. Edmunds and H. Triebel, Logarithmic Sobolev spaces and their applications to
spectral theory, Proc. London Math. Soc., 71 (1995), 333–371.
[20] M.J. Esteban and P.-L. Lions, Existence and non-existence results for semilinear el-
liptic problems in unbounded domains, Proc. Roy. Soc. Edinburgh Sect. A, 93 (1982),
1–14.
[21] A. Farina, Liouville-type theorems for elliptic problems, in: M. Chipot (Ed.), Hand-
book of Differential Equations -vol.4; Stationary Partial Differential Equations, Else-
vier, 2007, 60–116.
[22] A. Farina and J. Serrin, Entire solutions of completely coercive quasilinear elliptic
equations. J. Differential Equations, 250 (2011), 4367–4408.
[23] R. Filippucci, Nonexistence of positive weak solutions of elliptic inequalities, Nonlinear
Anal., 70 (2009), 2903–2916.
[24] R. Filippucci, Nonexistence of nonnegative solutions of elliptic systems of divergence
type, J. Differential Equations, 250 (2011), 572–595.
[25] A. Fiorenza and M. Krbec, Indices of Orlicz spaces and some applications, Comment.
Math. Univ. Carolin., 38 (1997), 433–451.
[26] B. Franchi, E. Lanconelli, and J. Serrin, Existence and uniqueness of nonnegative
solutions of quasilinear equations in RN , Adv. Math., 118 (1996), 177–243.
[27] H. Fujita, On the blowing up of solutions of Cauchy problem for ut = ∆u + u1+α , J.
Fac. Sci. Univ. Tokyo Sect. I, 13 (1966) 109–124.
[28] M. Garcia-Huidobro, R. Manasevich, E. Mitidieri, and C. Yarur, Existence and nonex-
istence of positive singular solutions for a class of semilinear elliptic systems, Arch.
Ration. Mech. Anal., 140 (1997), 253–284.
[29] Y. Giga and R.V. Kohn, Characterizing blowup using similarity variables, Indiana
Univ. Math. J., 36 (1987), 1–40.
Nonexistence results for differential inequalities 335
[30] B. Gidas and J. Spruck, Global and local behavior of positive solutions of nonlinear
elliptic equations, Comm. Pure Appl. Math., 34 (1981), 525–598.
[31] B. Gidas and J. Spruck, A priori bounds for positive solutions of nonlinear elliptic
equations, Comm. Partial Differential Equations, 6 (1981), 883–901.
[32] J.-P. Gossez, Nonlinear elliptic boundary value problems for equations with rapidly
(or slowly) increasing coefficients, Trans. Amer. Math. Soc., 190 (1974), 163–205.
[33] L. Gross, Logarithmic Sobolev Inequalities and Contractivity Properties of Semigroups,
in: Dirichlet Forms, Varenna 1992, E. Fabes et al. (eds), Lecture Notes in Mathematics
1563, Springer-Verlag, Berlin-Heidelberg, 1993, pp. 54–88.
[34] J. Gustavsson and J. Peetre, Interpolation of Orlicz spaces, Studia Math., 60 (1977),
33–59.
[35] D. Haroske, Some logarithmic function spaces, entropy numbers, applications to the
spectral theory, Dissertationes Math. (Rozprawy Mat.) 373 (1998), 1–59.
[36] H. Hudzik, Density of C0∞ (Rn ) in generalized Orlicz-Sobolev space WM k
(Rn ), Funct.
Approx. Comment. Math. 7 (1979), 15–21.
[37] T. Iwaniec and C. Sbordone, Caccioppoli estimates and very weak solutions of elliptic
equations. Renato Caccioppoli and modern analysis. Atti Accad. Naz. Lincei Cl. Sci.
Fis. Mat. Natur. Rend. Lincei (9) Mat. Appl. 14 (2003), no. 3 (2004), 189–205.
[38] A. Kalamajska and M. Krbec, Gagliardo–Nirenberg inequalities in regular Orlicz
spaces involving nonlinear expressions, J. Math. Anal. Appl., 362 (2010), 460–470.
[39] A. Kalamajska and K. Pietruska–Paluba, Interpolation inequalities for derivatives in
logarithmic Orlicz spaces, Coll. Math., 106 (2006), 93–107.
[40] A.G. Kartsatos and V.V. Kurta, Nonexistence theorems for weak solutions of quasi-
linear elliptic equations, Abstr. Appl. Anal., 6 (2001), 163–189.
[41] N. Kawano, E. Yanagida, and S. Yotsutani, Structure theorems for positive radial
solutions to div(|Du|m−2 Du) + K(|x|)uq = 0 in Rn J. Math. Soc. Japan, 45 (1993),
719–742.
[42] B. Kawohl, Symmetrization-or how to prove symmetry of solutions to PDE, in: Partial
Differential Equations, Theory and Numerical Solution, Eds. W. Jäger, O. John, J.
Nečas, K. Najzar, J. Stara, Chapman and Hall CRS Research Notes in Math., 406,
214–229.
[43] M.C. Knaap and L.A. Peletier, Quasilinear elliptic equations with nearly critical
growth, Comm. Part. Diff. Eq., 14 (1989), 1351–1383.
[44] M.A. Krasnoselskii and Ya.B. Rutickii, “Convex Functions and Orlicz Spaces,” P.
Noordhoff Ltd. Groningen 1961.
[45] H. Levine, The role of critical exponents in blowup theorems, SIAM Rev., 32 (1990),
262–288.
[46] V.G. Maz’ya, “Sobolev Spaces,” Springer–Verlag, Berlin, 1985.
[47] E. Mitidieri, Nonexistence of positive solutions of semilinear elliptic systems in Rn ,
Differential Integral Equations, 9 (1996), 465–479.
[48] E. Mitidieri and S. Pohozaev, Absence of global positive solutions to quasilinear elliptic
inequalities, Dokl. Akad. Nauk, 359 (1998), 456–460.
[49] E. Mitidieri and S. Pohozaev, Nonexistence of positive solutions to quasilinear elliptic
problems in Rn , Proc. Steklov. Inst. Math., 227 (1999), 186–216, (translated from Tr.
Mat. Inst. Steklova, 227 (1999), 192–222).
336 A. Kalamajska, K. Pietruska-Paluba, and I. Skrzypczak
[50] E. Mitidieri and S. Pohozaev, Nonexistence of positive solutions for a system of quasi-
linear elliptic equations and inequalities in Rn , Dokl. Math., 59 (1999), 351–355 (trans-
lated from: Dokl. Akad. Nauk., 366 (1999), 13–17.)
[51] E. Mitidieri and S. Pohozaev, A priori estimates and blow-up of solutions to nonlinear
partial differential equations and inequalities. Transl. from the Russian, Proc. Steklov
Inst. Math., 234 (2001), 1–362. (translated from Tr. Mat. Inst. Steklova, 234 (2001),
1–383.
[52] E. Mitidieri and S. Pohozaev, Nonexistence of weak solutions for some degenerate
elliptic and parabolic problems on Rn , J. Evol. Equ., 1 (2001), 189–220.
[53] E. Mitidieri and S. Pohozaev, Towards a unified approach to nonexistence of solutions
for a class of differential inequalities, Milan J. Math., 72 (2004), 129–162.
[54] M. Otanu, Existence and nonexistence of nontrivial solutions of some nonlinear de-
generate elliptic equations, J. Funct. Anal., 76 (1988), 140–159.
[55] W.-M. Ni and J. Serrin, Nonexistence theorems for quasilinear partial differential
equations, Proceedings of the conference commemorating the 1st centennial of the
Circolo Matematico di Palermo (Palermo, 1984). Rend. Circ. Mat. Palermo (2) Suppl.
No. 8 (1985), 171–185.
[56] A.M. Piccirillo, L. Toscano, and S. Toscano, Nonexistence theorems for some nonlinear
systems of elliptic inequalities, NoDEA Nonlinear Differential Equations Appl., 12
(2005), 323–350.
[57] S. Pohozaev and A. Tesei, Nonexistence of local solutions to semilinear partial differ-
ential inequalities, Ann. Inst. H. Poincaré Anal. Non Linéaire, 21 (2004), 487–502.
[58] S. Pohozaev, Essentially nonlinear capacities induced by differential operators, Dokl.
Akad. Nauk, 357 (1997), 592–594.
[59] P. Pucci and J. Serrin, A general variational identity, Indiana Univ. Math. J., 35
(1986), 681–703.
[60] P. Rabinowitz, Some global results for nonlinear eigenvalue problems, J. Funct. Anal.,
7 (1971), 487–513.
[61] M.M. Rao and Z.D. Ren, “Theory of Orlicz Spaces,” M. Dekker, Inc. New York, 1991.
[62] W. Reichel and H. Zou, Non-existence results for semilinear cooperative elliptic sys-
tems via moving spheres, J. Differential Equations, 161 (2000), 219–243.
[63] J. Serrin, A symmetry theorem in potential theory, Arch. Rational Mech. Anal., 43
(1971), 304–318.
[64] J. Serrin, “Positive Solutions of Prescribed Mean Curvature Problem,” Lecture Notes
in Math., 1340, Springer, New York, 1988.
[65] J. Serrin and H. Zou, Non-existence of positive solutions of Lane-Emden systems,
Differential Integral Equations., 9 (1996), 635–653.
[66] J. Serrin and H. Zou, Cauchy-Liouville and universal boundedness theorems for quasi-
linear elliptic equations and inequalities, Acta Math., 18 (2002), 79–142.
[67] I.B. Simonenko, Interpolation and extrapolation of linear operators in Orlicz spaces,
(in Russian) Mat. Sb. (N.S.) 105 (1964), 536–553.
[68] L. Veron and S. Pohozaev, Nonexistence results of solutions of semilinear differential
inequalities on the Heisenberg group, Manuscripta Math., 102 (2000), 85–99.
[69] L. Veron and S. Pohozaev, Blow-up results for nonlinear hyperbolic inequalities, Ann.
Sc. Norm. Super. Pisa Cl. Sci., 29 (2001), 393–420.