Professional Documents
Culture Documents
IP
CR
LEAST ACTION NODAL SOLUTIONS FOR A QUASILINEAR DEFOCUSING
SCHRÖDINGER EQUATION WITH SUPERCRITICAL NONLINEARITY
Abstract. In this paper we consider the existence of least action nodal solutions for the quasilinear
US
defocusing Schrödinger equation in H 1 (RN ):
k
−∆u + u∆u2 + V (x)u = g(u) + λ|u|p−2 u,
2
where N ≥ 3, V (x) is a positive continuous potential, g(u) is of subcritical growth, p ≥ 2∗ = 2N /(N − 2)
and λ, k are two non-negative parameters. By considering a minimizing problem restricted on a partial
Nehari manifold, we prove the existence of least action nodal solution via deformation flow arguments
and L∞ -estimates.
by UNIVERSITY OF NEW ENGLAND on 04/14/18. For personal use only.
Commun. Contemp. Math. Downloaded from www.worldscientific.com
AN
1. Introduction and main results
In this paper we are interested in the existence of a least action nodal solutions in H 1 (RN ) for the
modified quasilinear Schrödinger equation
k
(1.1) − ∆u + u∆u2 + V (x)u = g(u) + λ|u|p−2 u in RN ,
2
where V : RN → (0, ∞) is a continuous and positive potential function, g : R → R is a continuous and
DM
subcritical function, λ, k ≥ 0 are two parameters and p ≥ 2∗ = 2N /(N − 2) is the critical exponent in
the common sense. The existence of solutions for (1.1) is closely related to the study of the standing
waves ω(x, t) = u(x)e−(iEt)/~ for the superfluid film equation arising in the plasma physics (see [5]),
k
(1.2) i~∂t ω = −∆ω + W (x)ω − e h(|ω|2 )ω + ω∆ω 2 ,
2
where W (x) is a given potential and e h(u2 )u = g(u) + λ|u|p−2 u is a real function. So, ω(x, t) will be a
such solution of (1.2) if and only if u(x) solves equation (1.1) with V (x) = W (x) − E.
For the case k = 0 and λ = 0, equation (1.1) becomes a semilinear Schrödinger equation. The existence
of positive ground states or least action nodal solutions for the semilinear Schrödinger equation has been
TE
studied widely, we may refer the readers to [2, 4, 18, 20] and the references therein for the literature of
nodal solutions of the semilinear Schrödinger equation.
For k = −1, the modified quasilinear Schrödinger equation has received a lot of attention. The
appearance of the quasilinear part u∆u2 makes the problem much more complicated, it is quite difficult
to study the associated energy functional directly in the Sobolev space H 1 (RN ) and requires one to
develop new techniques to apply the variational methods. The existence of a positive ground state
solution of equation (1.1) has been proved in [12] and [19] by introducing parameter λ in front of the
nonlinear term. In [13], by changing of variables, the authors studied the quasilinear problem was
EP
transformed to a semilinear one and the existence of a positive solution was proved by the Mountain-Pass
lemma in an Orlicz working space. Different from the changing variable methods, in [16] the authors
introduced a new perturbation techniques to study a class of subcritical quasilinear problems including
Carlos Alberto Santos was partially supported by CAPES/Brazil Proc. N o 2788/2015 − 02;
Jiazheng Zhou was partially supported by CNPq/Brazil Proc. N o 232373/2014 − 0.
1
AC
T
Accepted manuscript to appear in CCM
IP
2 M. YANG, CARLOS. A. SANTOS, AND J. ZHOU
the Modified Schrödinger equation 1.1 and they also proved the existence of solutions for the critical
case in [17].
CR
The existence of sign-changing solution is an interesting topic i.e. looking for solutions u with
6 0, where u+ (x) = max{u(x), 0} ≥ 0, and u− (x) = min{u(x), 0} ≤ 0, x ∈ RN . In [14]
u+ , u− =
the authors proved the existence of sign-changing ground state solution for (1.1) with k = −1 and
g(s) = |s|p−2 s, s ∈ R with 3 ≤ p < 22∗ −1, that is, g having subcritical growth (22∗ plays the role of critical
exponent here), and V is a continuous function such that 0 < V0 = inf RN V (x) ≤ lim|x|→∞ V (x) = V∞
with V (x) ≤ V∞ − A/(1 + |x|m ), for |x| ≥ M , for some real constants A, M, m > 0. The perturbation
arguments in [17] was also successfully applied to study the existence of multiple nodal solutions for a
US
general class of sub-critical quasilinear Schrödinger equation in [15].
Also, we would also like to mention [6, 7, 9, 11, 14] and references therein for some recent progress
of the study of the quasilinear Schrödinger equation for k < 0. However, in [8, 10], the nonlinearity g
is permitted to behave in a critical way, under the more restrictive assumption that V is symmetric
radially positive and differentiable continuous function with V 0 (r) ≥ 0 for r ≥ 0. Their approach was
based on Mountain Pass Theorem on Nehari manifolds.
However, for the case k > 0, it seems that there are few existence results of solutions, and the existence
by UNIVERSITY OF NEW ENGLAND on 04/14/18. For personal use only.
of sign-changing solutions to equation (1.1) has not been considered yet, specially with supercritical
Commun. Contemp. Math. Downloaded from www.worldscientific.com
AN
nonlinearity. Thus the aim of the present paper is to study the existence of sign-changing solution for a
quasilinear defocusing Schrödinger equation with supercritical nonlinearity.
To state the main results, we may assume that the potential function V is continuous such that
V (x) ≥ V0 > 0 for all x ∈ RN , and:
(V1 ) V is radially symmetric, i.e., V (x) = V (|x|) and V ∈ L∞ (RN ), or
(V2 ) lim|x|→∞ V (x) = ∞.
The nonlinearity g is supposed to satisfy:
(G1 ) g ∈ C(R, R) is such that g(0) = 0 and odd;
DM
x∈RN k ρ ρ ρ
The nodal solution found in Theorem 1.1 minimizes the energy functional among all possible sign-
changing solutions for (1.1), and so we call it the least action nodal solution. As a corollary, Theorem 1.1
shows the existence of least energy nodal solution for equation (1.1) with just subcritical growth (λ = 0)
which also improves or complements works [6–9, 11, 14] in some sense. However, our main objetive in
this paper is improve Theorem 1.1 in [1] that considered g(u) = |u|q−2 u with q ∈ (2, 2∗ ) and, by using
Mountain Pass Theorem, the authors showed existence of solutions for (1.1) without giving information
EP
The paper is organized as follows. In Sect.2, we introduce the variational framework for the quasilinear
defocusing Schrödinger equation. In Sect.3, by establishing some auxiliary lemmas, we are able to prove
AC
T
Accepted manuscript to appear in CCM
IP
SUPERCRITICAL QUASILINEAR SCHRÖDINGER EQUATION 3
the existence of nodal solution for (1.1) with subcritical growth. Finally, we prove the main result by
penalizing the supercritical nonlinearity and obtaining a L∞ -estimate.
CR
Notation. We will use the following notations frequently:
• C, C0 , C1 , C2 , . . . denote positive (possibly different) constants.
• BR denotes the open ball centered at the origin with radius R > 0.
• C0∞ (RN ) denotes functions infinitely differentiable with compact support in RN .
• For 1 ≤ s ≤ ∞, Ls (RN ) denotes the usual Lebesgue space with the norms
ˆ 1/s
|u|s := |u|s , 1 ≤ s < ∞;
US
RN
|u|∞ := inf{C > 0 : |u(x)| ≤ C almost everywhere in RN }.
• H 1 (RN ) denotes the Sobolev spaces with usual norm
1/2
kuk1,2 := |∇u|22 + |u|22 .
• The weak convergence in H 1 (RN ) is denoted by *, and the strong convergence by →.
by UNIVERSITY OF NEW ENGLAND on 04/14/18. For personal use only.
Commun. Contemp. Math. Downloaded from www.worldscientific.com
I(u) =
1
ˆ
2 RN
AN
In this section, let us consider equation (1.1) with λ = 0. Formally, this equation has a variational
structure, that is, by considering
(1 − k|u|2 )|∇u|2 +
1
ˆ
2 RN
V (x)|u|2 −
ˆ
G(u),
RN
a function u ∈ H 1 (RN ) is said to be a weak solution of equation (1.1) if it satisfies
ˆ ˆ ˆ ˆ
2 2
(1 − k|u| )∇u∇ϕ − k |∇u| uϕ + V (x)uϕ = g(u)ϕ
DM
RN RN RN RN
1 N 0 1 N
for all ϕ ∈ H (R ), which means hI (u), ϕi = 0 for all ϕ ∈ H (R ).
First, we point out that, under (V1 ) and (V2 ), the subset
ˆ
E = {u ∈ H 1 (RN ) / V (x)u2 (x) < ∞}
RN
1 N
is a closed subspace of H (R ). Moreover,
ˆ ˆ
kuk2E = |∇u|2 + V (x)u2 (x)
RN RN
define an norm on E which is equivalent to usual norm on H 1 (RN ). However, the presence of the second
order nonhomogeneous term u∆u2 prevents us to work directly with the functional I, because it is not
TE
was chosen in (1.3). Clearly, when l(t) = 1 − kt2 , we derive that (2.1) turns into (1.1). Then, by
using Morse type L∞ -estimate, we will prove that there exist k0 , λ√ ∗ > 0 such that for all k ∈ [0, k0 ) and
λ ∈ [0, λ∗ ) the solution found verifies the estimate maxRN |u| < σ/ k. After that, we conclude that the
solutions obtained are solutions of the original equation (1.1).
For the equation (2.1), we will consider l : R → R defined by
p σ
2
1 − kt , if 0 ≤ t < √ ,
√ r k
l(t) =
C
σ3 k 1 σ
√ + , if t ≥ √ ,
kt 1 − σ 2 ρ k
AC
T
Accepted manuscript to appear in CCM
IP
4 M. YANG, CARLOS. A. SANTOS, AND J. ZHOU
and l(t) = p l(−t) for all t ≤ 0. So, it follows from the choice of σ = σ(ρ) > 0 for ρ > 1 in (1.3) that
l ∈ C 1 (R, ( 1/ρ, 1)) is an even function and it increases in (−∞, 0) and decreases in [0, +∞).
CR
Note that (2.1) is the Euler-Lagrange equation associated to the energy functional
1 1
ˆ ˆ ˆ
(2.2) Ik (u) = l2 (u)|∇u|2 + V (x)|u|2 − G(u)
2 RN 2 RN RN
√
for |u| < σ/ k.
In the sequel, we will prove √ the existence of nontrivial sign changing critical points u of (2.2)
satisfying sup N |u(x)| ≤ σ/ k. This means that it is a nontrivial sign changing solution of (2.1) with
√ x∈R
l(u) = 1 − ku2 , and so, a nontrivial sign changing solution of (1.1) can be got from the function l.
US
In what follows, we set ˆ t
L(t) = l(s)ds, t ∈ R.
0
By a simple computation, we see that the inverse function L−1 (t) exists and it is an odd function.
Moreover, it is very important to note that L, L−1 ∈ C 2 (R). The lemma below shows some important
properties of the functions l and L−1 that will be used in the later part of the paper.
by UNIVERSITY OF NEW ENGLAND on 04/14/18. For personal use only.
Commun. Contemp. Math. Downloaded from www.worldscientific.com
(2.3)
1
4+ +
ρ
AN
Remark 2.1. From assumption (G4 ), if ρ2 > ρ1 > 1 and g(t)/tρ2 is non-decreasing, then g(t)/tρ1 is
non-decreasing as well. Thus, if g(t)/tρ is non-decreasing for some ρ > 1, we can assume that ρ is
sufficiently close to 1, satisfying
r
1
ρ2
8
ρ
8 √
+ > √ and 2 < 2 ρ < θ.
ρ
Lemma 2.2. The functions l and L−1 satisfy:
L−1 (t)
(1) limt→0 t = 1;
L−1 (t) √
DM
(2) lim = ρ;
q t→∞ t √
1
(3) ρ t ≤ l(t)t ≤ L(t) ≤ t and t ≤ L−1 (t) ≤ ρt, for all t ≥ 0;
σ2 t 0
(4) − 1−σ 2 ≤ l(t) l (t) ≤ 0, for all t ≥ 0;
−1 δ
[L (t)]
(5) l(L−1 (t))t , t > 0 is increasing for δ > 1 and non-decreasing for δ = 1,
L−1 (t) L−1 (t)
(6) l(L−1 (t))tρ , t > 0 is decreasing for ρ > 1 close to 1 and t , t > 0 is non-decreasing.
Proof. Conclusions (1)-(4) can be proved by adapting the strategies in [1]. Here, let us prove conclusions
(5) and (6).
(5). Since,
h (L−1 (t))δ i0 δ[L−1 (t)]δ−1 [L−1 (t)]0 l(L−1 (t))t − [L−1 (t)]δ [l0 (L−1 (t))(L−1 (t))0 t + l(L−1 (t))]
TE
= , t>0
l(L−1 (t))t [l(L−1 (t)t)]2
and
(2.4) [L−1 (t)]0 l(L−1 (t)) = 1
hold, the claim follows from (3) and (4) of Lemma 2.2 due to the fact that
δ[L−1 (t)]δ−1 [L−1 (t)]0 l(L−1 (t))t −[L−1 (t)]δ [l0 (L−1 (t))(L−1 (t))0 t + l(L−1 (t))]
EP
l0 (L−1 (t))
≥ δ[L−1 (t)]δ−1 t − l(L−1 (t)) [L
−1
(t)]δ t − [L−1 (t)]δ l(L−1 (t))
[L−1 (t)]0 l(L−1 (t))tρ − L−1 (t)[l0 (L−1 (t))(L−1 (t))0 tρ + l(L−1 (t))ρtρ−1 ]
= , t > 0.
l(L−1 (t))tρ [l(L−1 (t))tρ ]2
AC
T
Accepted manuscript to appear in CCM
IP
SUPERCRITICAL QUASILINEAR SCHRÖDINGER EQUATION 5
CR
[L−1 (t)]0 l(L−1 (t))tρ −L−1 (t)[l0 (L−1 (t))(L−1 (t))0 tρ + l(L−1 (t))ρtρ−1 ]
US
t
[L−1 (t)]0 t − L−1 (t) = l(L−1 (t)) − L−1 (t)
h i
t
= L−1 (t) l(L−1 (t))L −1 (t) − 1 , t > 0,
which means
h L−1 (t) i0 [L−1 (t)]0 t − L−1 (t)
by UNIVERSITY OF NEW ENGLAND on 04/14/18. For personal use only.
= ≥ 0, t > 0.
Commun. Contemp. Math. Downloaded from www.worldscientific.com
t t2
0
we can observe that the functional Ik can be rewritten in the form
1 1
ˆ ˆ ˆ
−1
Jk (v) = 2
|∇v| + V (x)|L (v)| −2
G(L−1 (v)).
2 RN 2 RN RN
DM
ˆ h
L−1 (v) g(L−1 (v)) i
∇v∇φ + V (x) φ − φ = 0,
RN l(L−1 (v)) l(L−1 (v))
since v is a critical point of Jk . Thus, by calculating, we know
ˆ h i
(1 − k|u|2 )∇u∇ϕ − k|∇u|2 uϕ + V (x)uϕ − g(u)ϕ = 0
RN
√
EP
The following embedding result plays an important role in showing that the positive and negative
parts of a minimizing function on the partial Nehari manifold are non-trivial functions.
Proposition 2.4. The function L−1 is such that:
(1) the map v 7−→ L−1 (v) from E, || · ||E to Ls (RN ), | · |s is continuous for 2 ≤ s ≤ 2∗ ,
C
(2) under (V2 ) the above map is compact for 2 ≤ s < 2∗ , and under (V1 ) with N ≥ 2, this map is
compact for 2 < s < 2∗ .
AC
T
Accepted manuscript to appear in CCM
IP
6 M. YANG, CARLOS. A. SANTOS, AND J. ZHOU
√ √
Proof. (1) For v ∈ E, we have |L−1 (v)| ≤ ρ|v| and |∇L−1 (v)| ≤ ρ|∇v|. And so L−1 (v) ∈ E, by
classical embedding, we know L−1 (v) ∈ Ls (RN ) for 2 ≤ s ≤ 2∗ . Now, take (vn ) a sequence in E such
CR
that vn → v in E. Thus, we have vn → v in Ls (RN ) for 2 ≤ s ≤ 2∗ , and
√
|L−1 vn − L−1 (v)| ≤ ρ|vn − v|.
Then ˆ ˆ
|L−1 (vn ) − L−1 (v)|s ≤ ρs/2 |vn − v|s → 0.
RN RN
That is, L−1 (vn ) → L−1 (v) in Ls (RN ).
(2) In the case (V1 ), by the compact embedding from E → Ls (RN ) for 2 ≤ s < 2∗ , we have v → L−1 (v)
US
is compact from E to Ls (RN ) for 2 ≤ s < 2∗ . Similarly, we can prove the compactness for the other
case.
3. Auxiliary Results
Before stating the auxiliary results, let us point out some consequences of our hypotheses.
by UNIVERSITY OF NEW ENGLAND on 04/14/18. For personal use only.
Remark 3.1. From assumption (G2 ), there exists c > 0 such that
Commun. Contemp. Math. Downloaded from www.worldscientific.com
Remark 3.2. From assumption (G3 ), there exists a constant K > 0 such that
G(t) ≥ K|t|θ for all |t| > δ
for each δ > 0 given.
After these, let us associate to the functional Jk the Nehari manifold
DM
2 RN 2 RN RN
for each v ∈ E with v ± 6= 0.
Lemma 3.3. Assume that (G1 ) − (G3 ) hold. If v ∈ E with v ± 6= 0, then there exist α, β > 0 such that
hJk0 (αv + + βv − ), v + i = 0 and hJk0 (αv + + βv − ), v − i = 0,
that is, αv + + βv − ∈ M , and (α, β) ∈ (0, ∞)2 is a critical point of hvk .
EP
IP
SUPERCRITICAL QUASILINEAR SCHRÖDINGER EQUATION 7
CR
g(L−1 (tv + )) +
ˆ ˆ
0 + − + + − +
hJk (tv + sv ), tv i ≥ (t∇v + s∇v )(t∇v ) − tv
RN RN l(L−1 (tv + ))
−1
(tv )| + c |L−1 (tv + )|q +
|L + 2
ˆ ˆ
≥ t2 |∇v + |2 − p tv
RN RN 1/ρL−1 (tv + )
√ √
= t2 |∇v + |22 − ρt2 |v + |22 − ρc tq |v + |qq ,
US
which means there exists tm > 0 sufficiently small such that
hJk0 (tm v + + sv − ), tm v + i > 0 for all s > 0,
since q > 2. Taking advantage of g, L−1 being odd functions and l being even, it follows from similar
arguments as above that
g(L−1 (−|sv|− )) −
ˆ ˆ
hJk0 (tv + + sv − ), sv − i ≥ (t∇v + + s∇v − )(s∇v − ) − −1 (−|sv|− ))
sv
by UNIVERSITY OF NEW ENGLAND on 04/14/18. For personal use only.
RN RN l(L
Commun. Contemp. Math. Downloaded from www.worldscientific.com
AN
≥ s2
ˆ
RN
|∇v − |2 −
√
ˆ
RN
L−1 (−|sv − |)2 + c L−1 (−|sv − |)q −
√
p
1/ρL−1 (−|sv − |)
= s2 |∇v − |22 − ρs2 |v − |22 − ρc sq |v − |qq .
So, we get sm > 0, sufficiently small, such that
|sv |
and
hJk0 (Rv + + sv − ), Rv + i < 0, hJk0 (tv + + Rv − ), Rv − i < 0 for all t, s ∈ [r, R].
Hence, the Lemma follows from Miranda’s theorem.
Lemma 3.4. Assume that (V1 ), or (V2 ) and (G1 ) − (G3 ) hold. Then:
(i) for all v ∈ N , we have
√ ˆ
θ − 2 ρ
C
ˆ
Jk (v) ≥ |∇v|2 + V (x)|L−1 (v)|2 ,
2θ RN RN
AC
T
Accepted manuscript to appear in CCM
IP
8 M. YANG, CARLOS. A. SANTOS, AND J. ZHOU
CR
ˆ ˆ
|∇v|2 + V (x)|L−1 (v)|2 ≥ γ, for all v ∈ N
RN RN
and ˆ ˆ
|∇v ± |2 + V (x)|L−1 (v ± )|2 ≥ γ, for all v ∈ M.
RN RN
US
1 1 g(L−1 (v))v
ˆ ˆ ˆ
G(L−1 (v)) ≤ g(L−1 (v))L−1 (v) ≤ .
RN θ RN θ RN l(L−1 (v))
Then, by using this inequality and (2.5) for v ∈ N , we have
1 1 L−1 (v)v
ˆ ˆ ˆ
G(L−1 (v)) ≤ |∇v|2 + V (x) ,
RN θ RN θ RN l(L−1 (v))
by UNIVERSITY OF NEW ENGLAND on 04/14/18. For personal use only.
Commun. Contemp. Math. Downloaded from www.worldscientific.com
√
and consequently, by adding and subtracting the term ρ/θ RN V (x)|L−1 (v)|2 to the expression of Jk ,
there holds
Jk (v) ≥
2θ
√ ˆ
θ − 2 ρ
RN
AN
|∇v|2 +
ˆ
RN
1ˆ
V (x)|L−1 (v)|2 +
´
θ RN
h√
V (x) ρ|L−1 (v)|2 −
L−1 (v)v i
l(L−1 (v))
.
Proof of (ii). By using Remark 3.1, the definition of N , the facts of l is even while L is odd, and (3) of
Lemma 2.2, we obtain
L−1 (v)v g(L−1 (v))v g(L−1 (v))L−1 (v)v
ˆ ˆ ˆ ˆ
|∇v|2 + V (x) = =
RN RN l(L−1 (v)) RNˆ l(L
−1 (v))
RN l(L
−1 (v))L−1 (v)
√
≤ ρ g(L−1 (v))L−1 (v)
ˆ hRN i
≤ |L−1 (v)|2 + c |L−1 (v)|q ,
RN
TE
for each > 0 given. Now, it follows from (3) of Lemma 2.2 again that
1h i L−1 (v)v
ˆ ˆ ˆ ˆ
2 −1 2 2
|∇v| + V (x)|L (v)| ≤ |∇v| + V (x)
(3.2) 2 RN RN ˆRN h RN l(L−1 (v))
i
≤ |L (v)| + c |L−1 (v)|q ,
−1 2
RN
EP
for all v ∈ E.
C
Besides this, by taking 0 < τ < 1 such that q = τ 2 + (1 − τ )2∗ , it follows from (3) of Lemma 2.2,
0 < V0 ≤ V (x), x ∈ RN , inequality (3.3), the Hölder inequality, and from the Sobolev embedding
AC
T
Accepted manuscript to appear in CCM
IP
SUPERCRITICAL QUASILINEAR SCHRÖDINGER EQUATION 9
theorem, that
ˆ τ ˆ
CR
∗ 1−τ
ˆ
|L−1 (v)|q ≤ |L−1 (v)|2 |L−1 (v)|2
RN RN ˆ RN ˆ
2∗ (1−τ ) τ ∗ 1−τ
−1
≤ρ 2 |L (v)|2 v2
RN RN
(3.4) 2∗ (1−τ ) ∗
ˆ τ ˆ 2∗ (1−τ
2
)
2 (1−τ ) −1
≤ρ 2 S |L (v)|2 |∇v|2
RN RN
hˆ i 2τ +2∗2(1−τ )
2 −1 2
≤ C2 |∇v| + V (x)|L (v)| ,
RN
US
where C1 , C2 are positive constants.
Now, by using (3.2), (3.3) and (3.4), we obtain
1 h ˆ i hˆ i 2τ +2∗2(1−τ )
2 −1 2 2 −1 2
− C1 |∇v| + V (x)|L (v)| ≤ c C2 |∇v| + V (x)|L (v)| ,
2 RN RN
h 1/2 − C i 2(1−τ
ˆ
RN AN
|∇v|2 + V (x)|L−1 (v)|2 ≥
Lemma 3.5. Assume the same hypotheses of Lemma 3.4, and (vn ) being a sequence in M . Then
DM
ˆ
lim inf |L−1 (vn± )|q dx > 0.
n→∞ RN
for all n ∈ N.
TE
Remark 3.6. By Lemma 3.5 and (3) of Lemma 2.2, there exists a constant γ1 > 0 such that
ˆ
|vn± |q ≥ γ1 > 0.
RN
∂hvk (t, s) ρ
=t − −1 (tv + ))(tv + )ρ
− (v ) ,
∂t RN t
ρ−1
RN l(L l(L−1 (tv + ))(tv + )ρ
AC
T
Accepted manuscript to appear in CCM
IP
10 M. YANG, CARLOS. A. SANTOS, AND J. ZHOU
for all s ≥ 0 given. Now, by using (G4 ), (5), (6) of Lemma 2.2, and the monotonicity of l, L−1 , we obtain
CR
g(L−1 (tv + )) V (x)L−1 (tv + )
−1
−
+
l(L (tv ))(tv ) + ρ l(L−1 (tv + ))(tv + )ρ
US
g(L−1 (v + )) L−1 (v + )
= −1 + + ρ
− V (x) −1
l(L (v ))(v ) l(L (v + ))(v + )ρ
for 0 < t < 1, and in a similar way, we obtain
g(L−1 (tv + )) V (x)L−1 (tv + )
−
by UNIVERSITY OF NEW ENGLAND on 04/14/18. For personal use only.
for t > 1.
AN >
g(L−1 (v + ))
−1 +
l(L (v ))(v ) + ρ
−
V (x)L−1 (v + )
l(L−1 (v + ))(v + )ρ
So, it follows from above informations, (5) of Lemma 2.2, and the hypothesis v ∈ M , that
∂hvk ∂hvk
(3.5) (t, s) > 0 for 0 < t < 1, and (t, s) < 0 for t > 1
∂t ∂t
for all s ≥ 0, and in a similar way together with the facts of l being even and L odd, we get
DM
∂hvk ∂hvk
(3.6) (t, s) > 0 for 0 < s < 1, and (t, s) < 0 for s > 1
∂s ∂s
hold for all t ≥ 0, as well.
In particular, given (t, s) ∈ [0, ∞)2 , it follows from (3.5) that hvk (t, s) < hvk (1, s). And by (3.6), we
have hvk (1, s) < hvk (1, 1). That is, hvk (t, s) < hvk (1, 1) = Jk (v). So, the Lemma is proved.
(ii) ∂s (t, s)
k
α2 β2
(α1 v + ) + (β1 v − ) = α2 v + + β2 v − .
α1 β1
Since α1 v + +β1 v − ∈ M , then by the Remark 3.9, we have α2 /α1 = β2 /β1 = 1, and so (α1 , β1 ) = (α2 , β2 ).
4. Proof of Theorem
C
Now, we are ready to prove Theorem 1.1 by applying the auxiliary results in Section 3 together with
Deformation arguments.
AC
T
Accepted manuscript to appear in CCM
IP
SUPERCRITICAL QUASILINEAR SCHRÖDINGER EQUATION 11
Proof of Theorem for λ = 0. It follows from Lemma 3.4 that there exists c0 > 0 such that
CR
c0 = inf Jk (w).
w∈M
US
ˆ 1/2
|∇wn |2 → ∞,
RN
it would follow from Lemma 3.4 that
√ ˆ
θ−2 ρ
Jk (wn ) ≥ |∇wn |2 → ∞,
2θ RN
by UNIVERSITY OF NEW ENGLAND on 04/14/18. For personal use only.
AN ˆ
RN
V (x)wn2 → ∞,
then it would follow from Lemma 3.4 again and (3) of Lemma 2.2, that
θ−2 ρ
√ ˆ
Jk (wn ) ≥ V (x)|L−1 (wn )|2
2θ√ ˆRN
θ−2 ρ
≥ V (x)wn2 → ∞,
2θ RN
DM
which is a contradiction again.
Hence, without loss of generality, we can assume up to a subsequence that there exist w, w1 , w2 ∈
H 1 (RN ) such that
wn * w wn+ * w1 wn− * w2 in H 1 (RN ),
wn * w wn+ * w1 wn− * w2 in Lγ (RN ) for all γ ∈ [2, 2∗ ]
and
wn −→ w wn+ −→ w1 wn− −→ w2 in Lγloc (RN ) for all γ ∈ (1, 2∗ ).
Since the mappings w → w+ and w → w− are continuous from Lγloc (RN ) in Lγloc (RN ), we obtain
w1 = w+ ≥ 0 and w2 = w− ≤ 0. Besides this, it follows from Proposition 2.4, that
L−1 (wn± ) → L−1 (w± ) in Lγ (RN ) for all γ ∈ (2, 2∗ ).
TE
ˆ ˆ
−→ .
RN l(L−1 (wn± )) RN l(L−1 (w± ))
So, together with Fatou’s Lemma, we obtain that Jk is weak lower semi-continuous.
Hence, by wn ∈ M , it follows from Lemma 3.7 that
− + − wn
c0 ≤ hw +
k (α, β) = Jk (αw + βw ) ≤ lim inf Jk (αwn + βwn ) = lim inf hk (α, β)
≤ lim inf hw + −
k (1, 1) = lim inf Jk (wn + wn ) = lim inf Jk (wn ) = c0 ,
n
C
IP
12 M. YANG, CARLOS. A. SANTOS, AND J. ZHOU
CR
are going to show Jk0 (v) = 0.
Suppose, by contradiction, that Jk0 (v) 6= 0. Then there exist δ > 0 and ν such that
||Jk0 (w)|| ≥ ν for every w ∈ E with ||w − v|| ≤ 2δ.
Since v + 6= 0 and v − = 6 0, we can take L = min{||v + ||, ||v − ||} > 0 and, without loss of generality, we
may assume 6δ < L.
Let D = [ 12 , 32 ] × [ 12 , 32 ]. Since, hJk0 (v ± ), v ± i = 0 and by Lemma 3.7,
US
Jk (tv + + sv − ) < Jk (v) = c0 ,
holds for (t, s) ∈ D with t 6= 1 or s 6= 1, we obtain that
c̃ = max Jk (tv + + sv − ) < c0 .
∂D
Applying [ [21] Lemma 2.3] with = min{(c0 −c̃)/2, νδ/8} and S = B(v, δ), there exists η ∈ C([0, 1]×E, E)
such that
by UNIVERSITY OF NEW ENGLAND on 04/14/18. For personal use only.
(i) η(θ, u) = u if θ = 0 or if u ∈
c0 +
(ii) η(1, Jk ) ∩ B(v, δ) ⊂ Jk ; c0 −
AN
(iii) Jk (η(1, w)) ≤ Jk (w) for every w ∈ E,
where Jka = {w ∈ E : Jk (w) ≤ a}. Consequently, we have
(4.1) max Jk (η(1, tv + + sv − )) < c0 .
(t,s)∈D
On the other hand, we claim that there exists (t0 , s0 ) ∈ D such that
η(1, t0 v + + s0 v − ) ∈ M.
DM
1
it follows from property (i) for η and inequality (4.2) that ϕ(t, s) = tv + + sv − if t = 2 for s ∈ [ 12 , 32 ].
Thus,
Ψ( 21 , s) = hJk0 (ϕ+ ( 21 , s)), ϕ+ ( 12 , s)i, hJk0 (ϕ− ( 12 , s)), ϕ− ( 12 , s)i
= hJk0 ( 12 v + ), 12 v + i, hJk0 (sv − ), sv − )i ,
and it follows from (3.5) that
1 1 1 ∂hvk 1
(4.3) hJk0 ( v + ), v + i = ( , s) > 0.
EP
2 2 2 ∂t 2
On the other hand, using property (i) for η again and t = 23 in (4.2), ϕ(t, s) = tv + + sv − for s ∈ [ 12 , 23 ],
Ψ( 32 , s) = hJk0 (ϕ+ ( 32 , s)), ϕ+ ( 32 , s)i, hJk0 (ϕ− ( 23 , s))ϕ− ( 32 , s)i
= hJk0 ( 32 v + ), 23 v + i, hJk0 (sv − ), sv − i .
Again, it follows from (3.5) that
C
3 3 3 ∂hvk 3
(4.4) hJk0 ( v + ), v + i = ( , s) < 0.
2 2 2 ∂t 2
AC
T
Accepted manuscript to appear in CCM
IP
SUPERCRITICAL QUASILINEAR SCHRÖDINGER EQUATION 13
CR
1 1 1 ∂hvk 1
(4.5) hJk0 ( v − ), v − i = ( , s) > 0
2 2 2 ∂s 2
and
3 3 3 ∂hvk 3
(4.6) hJk0 ( v − ), v − i = ( , s) < 0.
2 2 2 ∂s 2
Noting that the function Ψ is continuous on D and taking (4.3)-(4.6) into account, we can apply Miranda’s
theorem to conclude that there exists (t0 , s0 ) ∈ D such that Ψ(t0 , s0 ) = (0, 0). This and (4.1) lead to a
US
contradiction. Hence, we conclude√that v is a critical point of Jk . So, by Lemma 2.3, we just need to
show that |u|∞ = |L−1 (v)|∞ ≤ σ/ k holds to conclude that u is a solution of problem (1.1) with λ = 0.
Now, set ϕ = L−1 (v)l(L−1 (v)). It follows from Lemma 2.2, that
L−1 (v)l0 (L−1 (v))
|ϕ| = |L−1 (v)l(L−1 (v))| ≤ |v|, and |∇ϕ| = |1 + ||∇v| ≤ |∇v|,
l(L−1 (v))
that is, ϕ ∈ H 1 (RN ). So, by taking ϕ as a test function in (2.5), we obtain
by UNIVERSITY OF NEW ENGLAND on 04/14/18. For personal use only.
Commun. Contemp. Math. Downloaded from www.worldscientific.com
ˆ h
L−1 (v)l0 (L−1 (v)) i
RN
1+
AN
l(L−1 (v))
As a consequence of (4) of Lemma 2.2, we have
ˆ
|∇v|2 + V (x)|L−1 (v)|2 − g(L−1 (v))L−1 (v) = 0.
m2 v in Bm .
We know vm ∈ H 1 (RN ), vm ≤ vm+1 , vm ≤ |v|2β−1 , and
(2β − 1)|v|2(β−1) ∇v in Am ,
∇vm =
m2 ∇v in Bm ,
that is, vm can be used as a test function. Besides this, we have
ˆ ˆ ˆ
EP
Letting
v|v|β−1 in Am ,
wm =
mv in Bm ,
2
we obtain wm = vvm ≤ |v|2β , wm ≤ wm+1 , and
C
β|v|β−1 ∇v in Am ,
∇wm =
m∇v in Bm .
AC
T
Accepted manuscript to appear in CCM
IP
14 M. YANG, CARLOS. A. SANTOS, AND J. ZHOU
So,
CR
ˆ ˆ ˆ
(4.9) |∇wm |2 = β 2 |v|2(β−1) |∇v|2 + m2 |∇v|2 .
RN Am Bm
US
ˆ ˆ
2 2
|∇wm | +β V (x) −1
vm
RN RN ˆ l(L (v))
L−1 (v)
ˆ ˆ
2 2(β−1) 2 2
= (β − 1) |v| |∇v| + ∇v∇vm + β V (x) vm
RN ˆ RN l(L−1 (v))
h (β − 1)2 Am i ˆ −1
L (v)
≤ +1 ∇v∇vm + +β 2 V (x) v
−1 (v)) m
2β − 1 R N R N l(L
−1
L (v)
ˆ
by UNIVERSITY OF NEW ENGLAND on 04/14/18. For personal use only.
≤ β2 [∇v∇vm + V (x) v ]
−1 (v)) m
Commun. Contemp. Math. Downloaded from www.worldscientific.com
ˆR l(L
ˆ
2 2
ˆ
= β2
R
N
2
ˆ
vm .
|L−1 (v)|2 2
ˆ
c |L−1 (v)|q
|∇wm | +β V (x) −1
vm ≤ β −1 −1
|vm |+β |v |,
−1 (v))L−1 (v)| m
RN RN l(L (v)) RN |l(L (v))L (v)| RN |l(L
that is,
L−1 (v) |L−1 (v)|2
ˆ ˆ
2 2
β V (x) v ≥ β |v |,
DM
m −1 (v))L−1 (v)| m
RN l(L−1 (v)) RN |l(L
for > 0 sufficiently small. So, we have
c |L−1 (v)|q
ˆ ˆ
|∇wm |2 ≤ β 2 −1 −1
|vm |
RN ˆRN |l(L (v))L (v)|
q
≤ β2 c ρ 2 |v|q−2 wm2
.
RN
Then, it follows from the Sobolev inequality that
ˆ ∗
NN−2 ˆ
|wm |2 ≤S |∇wm |2
Am RNˆ
q
TE
2
≤ Sβ c ρ 2 |v|q−2 wm
2
.
RN
The Hölder inequality implies that
ˆ NN−2 ˆ 1/r1
2∗ q
2 q−2
|wm | ≤ c ρ Sβ |v|2∗
2 |wm |2r1 ,
Am RN
ˆ ∗
NN−2 q
ˆ 1/r1
|v|β2 ≤ c ρ 2 Sβ 2 |v|2q−2
∗ |v|2βr1 ,
Am RN
Pi q Pi
j/σ j ) 1/σ j )
|v|σi 2∗ ≤ σ ( j=1 (c ρ 2 S|v|2q−2
∗ )
(1/2 j=1 |v|2∗ ,
AC
T
Accepted manuscript to appear in CCM
IP
SUPERCRITICAL QUASILINEAR SCHRÖDINGER EQUATION 15
that is, by doing i → ∞ and using the limitation of ||v||E , given by (4.7), together with the Sobolev
inequality, we get |v|∞ ≤ C0 , where C0 > 0 is a real constant independent of k > 0.
CR
Now, it follows from Lemma 2.2-(3) that
√ √ √
|u|∞ = |L−1 (v)|∞ ≤ ρ|v|∞ ≤ ρC0 ≤ σ/ k
√ √
holds for all k ∈ (0, k0 ), where k0 > 0 is such that ρC0 ≤ σ/ k0 . Thus, Lemma 2.3 implies that
problem (1.1), with λ = 0, admits a least energy nodal solution.
To finish the proof, let λ > 0 be a parameter and define gλ,τ (t) = λρ + g(t), t ∈ RN by
p−2
|t| t, 0 ≤ |t| ≤ τ,
US
ρτ (t) = τ p−1
g(t), |t| > τ,
g(τ )
for each τ > 0 given. So, gλ,τ fulfills assumptions (G1 )-(G4 ). Thus, by applying the prior case (problem
(1.1) with λ = 0), there exists an sign-changing solution u = uλ,τ ∈ H 1 (RN ) of the equation
−∆u + k2 u∆u2 + V (x)u = gλ,τ (u) in RN .
√
by UNIVERSITY OF NEW ENGLAND on 04/14/18. For personal use only.
of the equation AN
Hence, by defining v = vλ,τ ∈ H 1 (RN ) as u = L−1 (v), we know that v is a least energy nodal solution
,
Am g(τ ) RN
by using the Monotone Convergence Theorem, we obtain
n q
h τ p−1 io1/2β
|v|β2∗ ≤ β 1/β CSρ 2 η q−2 1 + λ ||v||2βr1 .
g(τ )
Take σ = 2∗ /(2r1 ) and let β = σ i , i = 1, 2, . . . , we get
n h Pi
τ p−1 io(1/2 j=1 1/σ )
C
P j
q
( ij=1 j/σ j ) q−2
|v|σi 2∗ ≤ σ CSρ η 2 1+λ |v|2∗ ,
g(τ )
AC
T
Accepted manuscript to appear in CCM
IP
16 M. YANG, CARLOS. A. SANTOS, AND J. ZHOU
that is, by taking the limit of i → ∞ and using (4.12) together with the Sobolev inequality, we get
CR
h τ p−1 i 2∗1−q
|v|∞ ≤ C0 1 + λ ,
g(τ )
where C0 > 0 is a real constant independent of λ, τ > 0.
Finally, choose τ > C0 sufficiently large. So, there is a λ∗ = λ∗ (τ ) > 0 such that for each fixed
0 ≤ λ ∈ (0, λ∗ ], the inequality
h τ p−1 i 2∗1−q
C0 1 + λ ≤τ
g(τ )
US
holds. So, it follows from the definition of gλ,τ , Lemma 2.2, and above inequality, that u = uλ,τ =
L−1 (vλ,τ ) is a solution of equation (1.1). This ends our proof.
Acknowledgements. The authors would like to thank the anonymous referee for his/her useful
by UNIVERSITY OF NEW ENGLAND on 04/14/18. For personal use only.
Commun. Contemp. Math. Downloaded from www.worldscientific.com
comments and suggestions which help to improve the presentation of the paper greatly.
AN References
[1] C. O. Alves, Y. Wang and Y. Shen, Soliton solutions for a class of quasilinear Schrödinger equations with a parameter,
J.Differential Equations 259 1 (2015) 318-343. 2, 4
[2] T. Bartsch and T. Weth, A note on additional properties of sign changing solutions to superlinear Schrödinger
equations, Topolog. Meth. Nonlin. Analysis 22 (2003), 1-14. 1
[3] T. Bartsch, T. Weth and M. Willem, Partial symmetry of least energy nodal solution to some variacional problems, J.
Anal. Math. 1, (2005) 1–18. 12
DM
[4] J. Chabrowski, On nodal radial solutions of an elliptic problem involving critical Sobolev exponent, Commentat. Math.
Univ. Carolinae 37 (1996), 1-16. 1
[5] S. Kurihura, Large-amplitude quasi-solitons in superfluids films. J. Phys. Soc. Japan 50, (1981), 3262–3267. 1
[6] Y. Deng, S. Peng and J. Wang, Infinitely many sign-changing solutions for quasilinear Schrödinger equations in RN,
Commn. Math. Sci. 9, No. 3 (2011), 859-878. 2
[7] Y. Deng, S. Peng and J. Wang, Nodal soliton solutions for generalized quasilinear Schrödinger equations, Journal of
Mathematical Physics 55, 051501 (2014); doi: 10.1063/1.4874108. 2
[8] Y. Deng, S. Peng and J. Wang, Nodal soliton solutions for quasilinear Schrödinger equations with critical exponent, J.
Math. Phys. 54 1 (2013) 011504. 2
[9] Y. Deng and W. Shuai, Existence and concentration behavior of sign-changing solutions for quasilinear Schrödinger
equations, Science China Mathematics 59 No. 6 (2016), 1095-1112. 2
[10] Y. Deng, Y. Li and X. Yan, Nodal solutions for a Quasilinear Schrödinger equation with critical nonlinearity and
non- square diffusion, Communications on Pure and Applied Analysis 14 No 6 ( 2015), 2487-2508. 2
[11] F. Li, X. Zhu and Z. Liang, Multiple solutions to a class of generalized quasilinear Schrödinger equations with a
TE
[20] TM. Struwe, Superlinear elliptic boundary value problems with rotational symmetry, Arch. Math. 39 (1982), 233-240. 1
[21] M. Willem, Minimax theorems, Volume 24 (Birkhäuser, 1996). 12
AC
T
Accepted manuscript to appear in CCM
IP
SUPERCRITICAL QUASILINEAR SCHRÖDINGER EQUATION 17
CR
E-mail address: mbyang@zjnu.edu.cn
US
by UNIVERSITY OF NEW ENGLAND on 04/14/18. For personal use only.
Commun. Contemp. Math. Downloaded from www.worldscientific.com
AN
DM
TE
C EP
AC