Professional Documents
Culture Documents
DEPARTAMENTO DE MATEMÁTICAS
TRUJILLO - PERÚ
2019
(joint work with Claudianor O. Alves, Universidad Federal de Campina Grande, Brasil
Paper published in Glasgow Mathematical Journal )
1 Motivation
2 Problem
4 Main Result
1 Motivation
2 Problem
4 Main Result
and other suitable conditions, Rabinowitz showed that there is 0 > 0 such that, for all
∈ (0, 0 ), problem (1) has at least one nontrivial weak solution u , by using mountain pass
and comparison arguments .
Wang 1993:
N +2
He considered f (x, u) = up where p ∈ (1, N −2
), then by assuming (b1 ), Wang showed the
existence of a sequence k → 0 and a local maximum xk ∈ Rn of uk = uk such that
and M
uk (x) ≤ Ce−
|x−xk |
,
(f3 ) There are f ∈ C(R) and a ∈ C(Rn ) such that f satisfies (AR), (f2 ) and
and
m{x ∈ Rn / f (x, ξ) > f (ξ) for all ξ > 0} > 0.
Then, by using mountain pass and comparison argument, Felmer et all. showed the existence
of a positive ground state of (2).
César Torres (UNT) NRO July 19, 2019 5 / 40
Fractional Schrödinger Equation with Competing Potential Functions
Shang and Zhang 2016:
then for > 0 small enough, equation (3) has a nontrivial nonnegative ground state solution,
where for any ξ ∈ Rn , G(ξ) be the ground state level associated to the problem
where Ω ⊂ Rn be a bounded domain, f ∈ C(Ω) and g ∈ C(∂Ω) and M is the singular integral
Z
|u(x + z) − u(x)|p−2 (u(x + z) − u(x))
M [u](x) = v.p. (p − σ) dz,
B(0,ρ(x))
|z|n+σ
where 1 < p < ∞, σ < p and ρ ∈ C(Ω) satisfies the following condition
By using the theory of viscosity solution the authors showed the existence of solution uσ for
problem (4). Furthermore, they showed that, uσ converge uniformly in Ω to the solution of
the Dirichlet problem
2α (−∆)α
ρ u + u = f (u) in Rn , u ∈ H α (Rn ). (6)
The operator (−∆)α ρ is a non-local regional laplacian, with range of scope determined by the
positive function ρ ∈ C(Rn , R+ ) and defined by
Z Z Z
[u(x + z) − u(x)][v(x + z) − v(x)]
(−∆)α
ρ uvdx = dzdx.
Rn Rn B(0,ρ(x))
|z|n+2α
Moreover ρ satisfies
(ρ1 ) There are numbers 0 < ρ0 < ρ∞ ≤ ∞ such that
(ρ2 ) In case ρ∞ = ∞ we further assume that there exists a ∈ (0, 1) such that
ρ(x)
lim sup ≤ a.
|x|→∞ |x|
n+2α
(f4 ) ∃ C > 0 such that; |f (t)| ≤ C(1 + |t|p ), 1 < p < n−2α
,
and ρ satisfies (ρ1 ) − (ρ2 ). Then (6) possesses at least one nontrivial weak solution.
Moreover this solution satisfies u(x) ≥ 0 a.e. for all x ∈ Rn .
We are interested in the concentration behavior of weak solutions for the equation
2α (−∆)α p
ρu+u = u , in Rn , u ∈ H α (Rn ), (7)
when → 0.
Remark
The scope function ρ, that describes the size of the ball of the influential region of the
non-local operator, plays a key role in deciding the concentration point of ground states of
the equation. Even though, at a first sight, the minimum point of ρ seems to be the
concentration point, there is a non-local effect that needs to be taken in account. This
non-local effect is governed by the concentration function H defined by
Z Z
|S n−1 | 1 1 1 dy 1 dy
H(x) = − − 2α + − ,
2α ρ(x)2α ρ∞ 2 C + (x)
|y|n+2α 2 C − (x)
|y|n+2α
and
C + (x) = {y ∈ Rn : ρ(x) < |y| < ρ(x + y)}.
Here we interpret the quotient 1/ρ2α
∞ as zero, when ρ∞ = ∞.
|x| = ρ(x + x0 ), x ∈ Rn ,
Theorem
n+2α
Let 0 < α < 1, n ≥ 2. Suppose that ρ satisfies (ρ1 )-(ρ3 ) and 1 < p < n−2α . Then for each
sequence m → 0, there exists a subsequence such that for every m, there is a non-negative
solution um = um of (7) that concentrates around a global minimum point x0 of H, as
m → 0. In more precise terms, for every δ > 0 there exists R > 0 and 0 > 0 such that if
< 0 we have
Z Z
u2m (x)dx ≤ n
m δ, and u2m (x)dx ≥ n
m C, ∀m ≤ 0 ,
B c (x0 ,m R) B(x0 ,m R)
1 Motivation
2 Problem
4 Main Result
2α (−∆)α
ρ u + Q(x)u = K(x)|u|
p−1 u in Rn ,
(P )
u ∈ H α (Rn ),
n+2α
where p ∈ (1, n−2α ), the operator (−∆)α
ρ is a variational version of the non-local regional
Laplacian, with range of scope determined by the positive function ρ ∈ C(Rn , R+ ) defined as
Z Z Z
[u(x + z) − u(x)][v(x + z) − v(x)]
(−∆)α
ρ uvdx = dzdx. (8)
Rn Rn B(0,ρ(x))
|z|n+2α
1 Motivation
2 Problem
4 Main Result
Now for ρ ∈ C(Rn , R+ ), we introduce the regional fractional Sobolev space Hρα (Rn ) by
Z Z
|u(x + z) − u(x)|
Hρα (Rn ) = 2 n
u ∈ L (R )/ dzdx < +∞
Rn B(0,ρ(x))
|z|n+2α
p
endowed with the norm kukα = hu, uiρ and inner product
Z Z Z
[u(x + z) − u(x)][v(x + z) − v(x)]
hu, viρ = u(x)v(x)dx + dzdx
Rn Rn B(0,ρ(x))
|z|n+2α
kukα ≤ Ckukρ
Remark-Embedding
(1) By Proposition 1 we have the continuous embedding Hρα (Rn ) ,→ H α (Rn ) and since th
embedding H α (Rn ) ,→ Lq (Rn ) is continuous for all q ∈ [2, 2∗α ] and locally compact for all
2n
q ∈ [2, 2∗α ), where 2∗α = n−2α is the fractional critical exponent, then we obtain
◦ Hρα (Rn ) ,→ Lq (Rn ) continuously for any q ∈ [2, 2∗α ].
◦ Hρα (Rn ) ,→ Lqloc (Rn ) compactly for any q ∈ [2, 2∗α ).
(2) Since kukρ ≤ kukα and by Proposition 1, kukα ≤ Ckukρ , then k · kρ and k · kα are
equivalent norms in H α (Rn ).
Since the spaces H α (Rn ) and Hρα (Rn ) are Hilbert spaces, then by the Lax-Milgran
representation theorem we can find a unique bijective linear map Mρ : H α (Rn ) → H α (Rn )
such that
hu, viρ = hMρ u, viα
Z
= ((−∆)α Mρ u + Mρ u)(x)v(x)dx ∀u, v ∈ H α (Rn ).
Rn
(−∆)α α
ρ = (−∆) ◦ Mρ + Mρ − I, (9)
where I is the natural injection from H α (Rn ) to H −α (Rn ) and H −α (Rn ) denotes the dual
space of H α (Rn ). Moreover we have that for all u, v ∈ H α (Rn )
Z Z Z
[u(x + z) − u(x)][v(x + z) − v(x)]
(−∆)α
ρ uvdx = dzdx. (10)
Rn Rn B(0,ρ(x))
|z|n+2α
supp(ϕ) ∩ B(0, R) = φ
then
Z
(−∆)α
ρ u(x)ϕ(x)dx = 0.
Rn
1 Motivation
2 Problem
4 Main Result
problem (P ) has a ground state solution u ∈ Hρα (Rn ) for small enough. Moreover, for
each sequence m → 0, there is a subsequence such that for each m ∈ N, the solution um
concentrates around a minimum point ξ ∗ of the function C(ξ), in the following sense: given
δ > 0, there are 0 , R > 0 such that
Z Z
|um |2 dx ≤ n
mδ and |um |2 dx ≥ n
m C, ∀m ≤ 0 ,
B c (ξ∗ ,m R) B(ξ∗ ,m R)
Remark
We would like to point out that the condition (C) is not empty, because it holds by
supposing that there is ξ0 ∈ Rn such that
p+1 p+1 n
n
− 2α p−1
− 2α
Q(ξ0 ) p−1 Q∞
2
< 2
.
K(ξ0 ) p−1 p−1
K∞
where
Z Z Z
[u(x + z) − u(x)][v(x + z) − v(x)]
hu, viξ = dzdx + Q(ξ)uvdx.
Rn Rn
|z|n+2α Rn
where
Γξ = {γ ∈ C([0, 1], H α (Rn )) : γ(0) = 0, Jξ (γ(1)) < 0}.
By problem (2), we know that (13) has a nontrivial nonnegative ground state solution, that
is, C(ξ) is the least critical value of Jξ .
Lemma 1
The function ξ → C(ξ) is continuous.
In the next lemma, D denotes the ground state level of the functional J : H α (Rn ) → R given
by
Z Z Z Z
1 |u(x) − u(z)|2 1
J(u) = dzdx + |u(x)|2 dx − |u|p+1 dx
2 Rn Rn
|x − z|n+2α Rn
p+1 Rn
Lemma 2
The function C(ξ) verifies the equality
p+1 n
− 2α
Q(ξ) p−1
C(ξ) = 2
D, ∀ξ ∈ Rn . (14)
K(ξ) p−1
Corollary 1
By Lemma 2, if there is ξ0 ∈ Rn such that
p+1 p+1 n
n
− 2α p−1
− 2α
Q(ξ0 ) p−1 Q∞
2
< 2
,
K(ξ0 ) p−1 p−1
K∞
we have
inf C(ξ) < lim inf C(ξ) = C(∞),
ξ∈Rn |ξ|→+∞
where C(∞) is the mountain pass level of the functionals J∞ : H α (Rn ) → R given by
Z Z Z Z
1 |u(x) − u(z)|2 1
J∞ (u) = dzdx + Q∞ |u| dx 2
− K∞ |u|p+1 dx.
2 Rn Rn
|x − z|n+2α Rn
p+1 Rn
The function u that verifies the above equality is called a ground state solution of (P ).
From now on, we are considering in H α (Rn ) the following norm
Z Z Z 12
|v(x + z) − v(x)|2
kvkρ = dzdx + Q(x)|v(x)|2 dx ,
Rn B(0,ρ (x))
|z|n+2α Rn
which is equivalent the usual norm of H α (Rn ). Moreover, we note that Iρ ∈ C 1 (H α (Rn ), R)
with Z
Iρ0 (u)v = hu, viρ − K(x)|u(x)|p−1 u(x)v(x)dx, ∀ v ∈ H α (Rn )
Rn
where
Z Z Z
[u(x + z) − u(x)][v(x + z) − v(x)]
hu, viρ = dzdx + Q(x)uvdx.
Rn B(0,ρ (x))
|z|n+2α Rn
with
Γρ = {γ ∈ C([0, 1], H α (Rn )) : γ(0) = 0, Iρ (γ(1)) < 0}.
In the sequel, Nρ denotes the Nehari manifold associated to the functional Iρ , that is,
It is easy to see that all nontrivial solutions of (P ) belongs to Nρ . Moreover, by applying
standard arguments, it is possible to prove the equality below
From (16), if Cρ is a critical value of Iρ then it is the least energy critical value of Iρ .
Hereafter, we say that Cρ is the ground state level of Iρ .
ξ0
Then, we take w (x) = w(x − ) and t > 0 satisfying Cρ ≤ Iρ (t w ) = maxt≥0 Iρ (tw ). The
ξ0
change of variable x̃ = x − gives
Z Z Z !
t2 |w(x̃ + z) − w(x̃)|2 2
Iρ (t w ) = dzdx̃ + Q(x̃ + ξ0 )w (x̃)dx̃
2 |z|n+2α
Rn B(0, 1 ρ(x̃+ξ0 )) Rn
Z
tp+1
p+1
− K(x̃ + ξ0 )|w| (x̃)dx̃.
p+1
Rn
Therefore, for any sequence n → 0, the equality Iρ0 (tn wn )(tn wn ) = 0 yields {tn } is
n
bounded. Thus, we can assume that tn → t∗ > 0, for some t∗ > 0. Thereby, taking the limit of
n → +∞, we can infer that
0
Jξ0 (t∗ w)(t∗ w) = 0.
On the other hand, we know that Jξ0 0 (w)(w) = 0, then we must have t∗ = 1. From this,
δ ρ(x) δ
I,∞ (τk uk ) ≥ C( , Q∞ − δ, K∞ + δ) = inf sup I,∞ (tv).
v∈H α (Rn )\{0} t≥0
Thus,
Z
ρ(x) τ2
Cρ ≥ C( , Q∞ − δ, K∞ + δ) + k [Q(x) − Q∞ + δ]|uk (x)|2 dx
2 B(0, R )
τkp+1
Z
+ [K∞ + δ − K(x)]|uk (x)|p+1 dx.
p+1 B(0, R )
ρ(x)
Cρ ≥ C( , Q∞ , K ∞ ) (20)
ρ(x)
where C(
, Q∞ , K ∞ ) designates the mountain pass level of the functional
Z Z Z !
1 |u(x + z) − u(x)|2
0
I∞,ξ (u) = dzdx + Q∞ |u|2 dx −
2 Rn B(0, 1 ρ(x))
|z|n+2α Rn
Z
1
K∞ |u|p+1 dx.
p+1 Rn
ρ(x)
lim inf C( , Q∞ , K∞ ) ≥ C(∞).
→0
Therefore, if there is sequence n → 0 such that the (P S)Cρ sequence has weak limit equal
n
to zero, we must have
ρ(n x)
Cρn ≥ C( , Q∞ , K∞ ), ∀n ∈ N,
n
leading to
lim inf Cρn ≥ C(∞),
n→+∞
which contradicts Lemma 3. This proves that the weak limit is nontrivial for > 0 small
enough and standard arguments show that its energy is equal to Cρ , showing the desired
result.
1 Motivation
2 Problem
4 Main Result
1
ρ̄ (x) = ρ(x + y ).
Using this function, it follows that w is a solution of the equation
(−∆)ρα w (x) + Q(x + y )w (x) = K(x + y )|w (x)|p−1 w (x), in Rn . (25)
Lemma C2
The sequence {y } is bounded. Moreover, if m ym → ξ ∗ , then
which is a contradiction, showing that {y } is bounded. Hence, there exists a subsequence of
{y } such that m ym → ξ ∗ .
César Torres (UNT) NRO July 19, 2019 34 / 40
Repeating the above arguments for the function
we have that this function satisfies the equation (25), and again {wm } is bounded in
H α (Rn ). Then wm * w in H α (Rn ) and w satisfies the equation below
in the weak sense. Furthermore, associated to (27) we have the energy functional
Z Z Z
1 |u(x + z) − u(x)|2
Jξ∗ (u) = dzdx + Q(ξ ∗ )|u(x)|2 dx
2 Rn Rn
|z|n+2α Rn
Z
1
− K(ξ ∗ )|u(x)|p+1 dx.
p+1 Rn
Using w as a test function in (25) and taking the limit of m → +∞, we get
Z Z Z Z
|w(x + z) − w(x)|2
dzdx + Q(ξ ∗ )|w(x)|2 dx ≤ K(ξ ∗ )|w|p+1 dx,
Rn Rn
|z|n+2α Rn Rn
From (28) and (29), wm → w in L2 (Rn ). From this, given δ > 0 there are 0 , R > 0 such that
Z Z
|um |2 dx ≤ n
mδ and |um |2 dx ≥ n
m C, ∀m ≤ 0 ,
B c (x∗ ,m R) B(x∗ ,m R)
MUITO OBRIGADO