Q -curvature type problem on bounded domains of R n

Abstract

In this paper, we establish compactness and existence results to a Branson–Paneitz type problem on a bounded domain of $R^{n}$ with Navier boundary condition.

Keywords

Nonlinear elliptic PDE critical exponent lack of compactness critical points at infinity

1. Introduction

In this work we are concerned with positive solutions of a nonlinear fourth order equation under the Navier boundary condition. Let K be a given function on a smooth bounded domain Ω of $R^{n}, n ⩾ 5$ . We are looking for a map $u : Ω \to R$ satisfying the following critical fourth order PDE $\begin{matrix} (1.1) & \{\begin{matrix} Δ^{2} u = K (x) u^{\frac{n + 4}{n - 4}}, \\ u > 0 in Ω, \\ Δ u = u = 0 on \partial Ω . \end{matrix} \end{matrix}$

The interest of this equation comes from its resemblance to the so called Q-curvature problem on closed manifolds involving the Branson–Paneitz operator. The latter has widely studied in the last two decades. (See [1,3,8,9,12,13,16–19] and the references therein for details).

Problem (1.1) has a variational structure with challenging mathematical difficulties. Indeed, if there is general standard line of attack to solve the analogous of (1.1) in the subcritical case. These approaches do not apply to the critical case since the embedding $H ↪ L^{\frac{2 n}{n - 4}} (Ω)$ where $H : = H^{2} (Ω) \cap H_{0}^{1} (Ω)$ , is not compact.

When $K = 1$ , the problem is called the Yamabe type problem. In this case, the existence of solutions of problem (1.1) depends on the topology of Ω. More precisely, if Ω is a star-shaped bounded domain, Van der Vorst [23] proved that (1.1) has no solution. When Ω has a non trivial homology group, Ebobisse–Ould Ahmedou showed that (1.1) has a solution [20].

When $K \neq 1$ , there have been many works dealing with (1.1). In these contributions, the conditions on $K (x)$ ensuring the solvability of (1.1) have been discussed. In [8,14] and [15], some existence results were obtained under the following two hypotheses:

(A) $\begin{matrix} \frac{\partial K}{\partial ν} (x) < 0, \forall x \in \partial Ω . \end{matrix}$ Here ν is the unit outward normal vector on $\partial Ω$ .

$(n d)$ K is a $C^{2}$ -positive function having only non degenerate critical points such that $\begin{matrix} Δ K (x) \neq 0 if \nabla K (x) = 0 . \end{matrix}$

Observe that $(n d)$ -condition would excludes some interesting class of functions K. For example the $C^{1}$ -functions and smooth functions having degenerate critical points. Our main motivation in this study, is to include a wider class of functions K for which (1.1) admits a solution. Our main assumption is the following β-flatness condition:

${(f)}_{β}$ Assume that K is a $C^{1}$ -positive function on $\overline{Ω}$ such that for each critical point y of K, there exists a real number $β = β (y) > 1$ , such that $\begin{matrix} K (x) = K (y) + \sum_{k = 1}^{n} b_{k} {| {(x - y)}_{k} |}^{β} + R (x - y), \end{matrix}$ for x close to y. Here $b_{k} = b_{k} (y) \in R ∖ {0}$ , for $k = 1, \dots, n$ , and $\sum_{s = 0}^{[β]} | \nabla^{s} R (z) | | z |^{s - β} = o (1)$ , as z tends to zero. Here $[β]$ denotes the integer part of β.

Note that the above mentioned $(n d)$ -condition is a particular case of the β-flatness assumption (in a suitable coordinates system) taking $β (y) = 2$ for any critical point y of K.

In the first part of this paper, we are interested with the case $1 < β ⩽ n - 4, n ⩾ 6$ . Our aim is to provide a full description of the lack of compactness of the associated variational problem to (1.1). Indeed, we will give a characterization of all critical points at infinity of the functional J in $Σ^{+}$ and state an Euler–Hopf type of existence result.

Let G denote the Green’s function of the bilaplacian under Navier boundary condition on Ω. It is defined by $\begin{matrix} G (x, y) = | x - y |^{- (n - 4)} - H (x, y), for x \neq y \in Ω, \end{matrix}$ where H its regular part.

Let $K$ denote the set consisting of all critical points of $K (x)$ . For any $y \in K$ , we define $\begin{matrix} \tilde{i} (y) = ♯ {b_{k} (y), b_{k} (y) < 0} . \end{matrix}$ Let $\begin{array}{l} K_{< n - 4} = {y \in K, β (y) < n - 4}, K_{< n - 4}^{+} = {y \in K_{< n - 4}, - \sum_{k = 1}^{n} b_{k} (y) > 0}, \\ K_{n - 4} = {y \in K, β (y) = n - 4}, K_{n - 4}^{+} = {y \in K_{n - 4}, - c_{1} \sum_{k = 1}^{n} b_{k} (y) + c_{2} H (y, y) > 0}, \end{array}$ and $\begin{matrix} K_{> n - 4} = {y \in K, β (y) > n - 4}, \end{matrix}$ where $\begin{matrix} c_{1} = \int_{R^{n}} \frac{| x_{1} |^{n - 4}}{{(1 + | x |^{2})}^{n}} d x, c_{2} = \int_{R^{n}} \frac{d x}{{(1 + | z |^{2})}^{\frac{n + 4}{2}}} . \end{matrix}$ Here $x_{1}$ is the first component of x in some geodesic normal coordinates system.

To any p-tuple of distinct points $τ_{p} = (y_{ℓ_{1}}, \dots, y_{ℓ_{p}}) \in K_{n - 4}^{p}, 1 ⩽ p ⩽ ♯ K$ , we associate a $p \times p$ symmetric matrix $M (τ_{p}) = {(m_{i j})}_{1 ⩽ i, j ⩽ p}$ defined by: $\begin{array}{l} m_{i i} = - \frac{1}{K {(y_{ℓ_{i}})}^{\frac{n}{4}}} (c_{1} \sum_{k = 1}^{n} b_{k} (y_{ℓ_{i}}) - c_{2} H (y_{ℓ_{i}}, y_{ℓ_{i}})) and \\ m_{i j} = - c_{2} \frac{G (y_{ℓ_{i}}, y_{ℓ_{j}})}{{(K (y_{ℓ_{i}}) K (y_{ℓ_{j}}))}^{\frac{n - 4}{8}}}, for i \neq j . \end{array}$ Let $ρ (τ_{p})$ be the least eigenvalue of $M (τ_{p})$ .

(B) Assume that $ρ (τ_{p}) \neq 0$ for any $1 ⩽ p ⩽ ♯ K$ .

Lastly define $\begin{array}{l} C_{n - 4}^{\infty} : = {τ_{p} = (y_{l_{1}}, \dots, y_{l_{p}}) \in {(K_{n - 4}^{+})}^{p}, 1 ⩽ p ⩽ ♯ K, s.t. y_{ℓ_{i}} \neq y_{ℓ_{j}} \forall i \neq j, and ρ (τ_{p}) > 0}, \\ C_{< n - 4}^{\infty} : = {τ_{p} = (y_{l_{1}}, \dots, y_{l_{p}}) \in {(K_{< n - 4}^{+})}^{p}, 1 ⩽ p ⩽ ♯ K, s.t. y_{ℓ_{i}} \neq y_{ℓ_{j}} \forall i \neq j} \end{array}$ and $\begin{matrix} C^{\infty} : = C_{n - 4}^{\infty} \cup C_{< n - 4}^{\infty} \cup (C_{n - 4}^{\infty} \times C_{< n - 4}^{\infty}) . \end{matrix}$

The following result describes the lack of compactness of the problem (1.1).

Theorem 1.1.
Under the assumptions $(A), (B)$ and ${(f)}_{β}$ for $1 < β ⩽ n - 4$ . The critical points at infinity of the associated variational problem to ( 1.1 ) (see Definition 2.4 ) are: $\begin{matrix} {(y_{ℓ_{1}}, \dots, y_{ℓ_{p}})}_{\infty} : = \sum_{i = 1}^{p} \frac{1}{K {(y_{ℓ_{i}})}^{\frac{n - 4}{2}}} P δ_{(y_{l_{i}}, \infty)} \end{matrix}$ where $(y_{l_{1}}, \dots, y_{l_{p}}) = (τ_{p}) \in C^{\infty}$ . The index of a such critical points at infinity is $i (τ_{p}) = p - 1 - \sum_{i = 1}^{p} (n - \tilde{i} (y_{ℓ_{i}}))$ .

The characterization of the critical points at infinity allows us to prove the following existence result.
Theorem 1.2.
Suppose that $(A), (B)$ and ${(f)}_{β}$ for $1 < β ⩽ n - 4$ hold. If additionally $\begin{matrix} \sum_{τ_{p} \in C^{\infty}} {(- 1)}^{i (τ_{p})} \neq 1, \end{matrix}$ then ( 1.1 ) has a solution.

In the second part of this paper, we are interested to the case of any $β > 1$ . We prove a partial description of the lack of compactness of the problem in that case and we provide a perturbation result.
Theorem 1.3.
Assume that K satisfies $(A)$ and ${(f)}_{β}$ for $β > 1$ . The critical points at infinity in $V (1, ε)$ are $\begin{matrix} P δ_{(y, \infty)}, y \in K_{< n - 4}^{+} \cup K_{n - 4}^{+} \cup K_{> n - 4} . \end{matrix}$ Such critical points at infinity has an index equal to $n - \tilde{i} (y)$

Now we state our perturbation result.
Theorem 1.4.
Under the assumptions $(A)$ and ${(f)}_{β}$ for $β > 1$ , if $\begin{matrix} \sum_{y \in K_{< n - 4}^{+} \cup K_{n - 4}^{+} \cup K_{> n - 4}} {(- 1)}^{n - \tilde{i} (y)} \neq χ (Ω) \end{matrix}$ then ( 1.1 ) has a solution provided K close to 1. Here $χ (Ω)$ is the Euler–Poincaré characteristic of Ω.

Our method hinges on the critical points at infinity theory of A. Bahri [4]. In Section 2, we state the variational structure associated to problem (1.1). In Section 3, we provide an asymptotic expansion of the gradient of J, without assuming any upper bound condition on the β-flatness condition. In Section 4, we characterize the critical points at infinity and we prove Theorems 1.1 and 1.3. Lastly in Section 5, we prove Theorems 1.2 and 1.4.
2. Preliminaries tools

Let $H : = H^{2} (Ω) \cap H_{0}^{1} (Ω)$ with the norm $\begin{matrix} ‖ u ‖_{H} = {(\int_{Ω} {(Δ u (x))}^{2} d x)}^{\frac{1}{2}} . \end{matrix}$ Define $\begin{matrix} Σ = {u \in H, s.t. ‖ u ‖_{H} = 1} and Σ^{+} = {u \in \sum, u > 0} . \end{matrix}$ Let $\begin{matrix} J (u) = \frac{{(\int_{Ω} {(Δ u)}^{2})}^{\frac{n}{n - 4}}}{\int_{Ω} K (x) u^{\frac{2 n}{n - 4}} d x} . \end{matrix}$

Observe that if u is a critical point of J on $Σ^{+}$ , then $J {(u)}^{\frac{n - 4}{8}} \cdot u$ is a solution of (1.1).

J does not satisfy the Palais–Smale condition on $Σ^{+}$ (P.S for short). This is due to the loss of compactness of the embedding $H ↪ L^{\frac{2 n}{n - 4}} (Ω)$ . Next, we describe the sequences failing P.S condition. For $a \in Ω$ and $λ > 0$ , let $\begin{matrix} (2.1) & δ_{a, λ} (x) = c_{n} {(\frac{λ}{1 + λ^{2} | x - a |^{2}})}^{\frac{n - 4}{2}}, \end{matrix}$ where $c_{n}$ is a positive constant chosen such that $δ_{a, λ}$ is the family of solutions of the following problem (see [21]): $\begin{matrix} (2.2) & Δ^{2} u = | u |^{\frac{8}{n - 4}} u, u > 0 in R^{n} . \end{matrix}$

Let $P δ_{a, λ}$ the unique solution of $\begin{matrix} \{\begin{matrix} Δ^{2} P δ_{a, λ} = {δ_{a, λ}}^{\frac{n + 4}{n - 4}} & in Ω \\ P δ_{a, λ} = Δ P δ_{a, λ} = 0 & on \partial Ω . \end{matrix} \end{matrix}$

We have the following estimates where originally introduced by Bahri [4]. $\begin{array}{l} (2.3) & P δ_{a, λ} = δ_{a, λ} - \frac{c}{λ^{\frac{n - 4}{2}}} H (a, \cdot) + O (\frac{1}{λ^{\frac{n}{2}} d^{n - 2}}), \\ (2.4) & λ \frac{\partial P δ_{a, λ}}{\partial λ} = λ \frac{\partial δ_{a, λ}}{\partial λ} + \frac{n - 4}{2} \frac{c}{λ^{\frac{n - 4}{2}}} H (a, \cdot) + O (\frac{1}{λ^{\frac{n}{2}} d^{n - 2}}), \\ (2.5) & \frac{1}{λ} \frac{\partial P δ_{a, λ}}{\partial a_{k}} = \frac{1}{λ} \frac{\partial δ_{a, λ}}{\partial a_{k}} - \frac{c}{λ^{\frac{n - 4}{2}}} \frac{\partial H (a, \cdot)}{\partial a_{k}} + O (\frac{1}{λ^{\frac{n + 2}{2}} d^{n - 1}}), \end{array}$ where c is a fixed positive constant, $d = d (a, \partial Ω)$ , $a_{k}$ is the kth coordinate of a.

We define now the set of potential critical points at infinity associated to $J_{ε_{0}}$ . Let for $ε > 0$ and $p \in N^{*}$ , $\begin{matrix} V (p, ε) = \{\begin{matrix} u \in Σ^{+}, s.t., \exists a_{1}, \dots, a_{p} \in Ω, \exists λ_{1}, \dots, λ_{p} > ε^{- 1} and \\ α_{1}, \dots, α_{p} > 0 with ‖ u - \sum_{i = 1}^{p} α_{i} P δ_{a_{i}, λ_{i}} ‖ < ε, ε_{i j} < ε \forall i \neq j, \\ λ_{i} d_{i} > ε^{- 1} and | J^{\frac{n}{n - 4}} (u) α_{i}^{\frac{8}{n - 4}} K (a_{i}) - 1 | < ε \forall i = 1, \dots, p . \end{matrix} \end{matrix}$ Here, $d_{i} = d (a_{i}, \partial Ω)$ and $ε_{i j} = {(\frac{λ_{i}}{λ_{j}} + \frac{λ_{j}}{λ_{i}} + λ_{i} λ_{j} | a_{i} - a_{j} |^{2})}^{\frac{4 - n}{2}}$ .

Proposition 2.1 ([5,22]).

Assume that J has no critical points in $Σ^{+}$ . Let ${(u_{k})}_{k}$ be a sequence in $Σ^{+}$ such that $J (u_{k})$ is bounded and $\partial J (u_{k})$ goes to zero. Then there exists a positive integer p, a sequence $(ε_{k})$ with $ε_{k} \to 0$ as $k \to + \infty$ and an extracted subsequence of ${(u_{k})}_{k}$ ’s, again denoted ${(u_{k})}_{k}$ , such that $u_{k} \in V (p, ε_{k}), \forall k$ .

The following Proposition gives a parametrization of $V (p, ε)$ .

Proposition 2.2 ([5]).

For all $p \in N^{*}$ , there exists $ε_{p} > 0$ such that for any $ε ⩽ ε_{p}$ and any u in $V (p, ε)$ , the problem $\begin{matrix} min {‖ u - \sum_{i = 1}^{p} α_{i} P δ_{a_{i}, λ_{i}} ‖, α_{i} > 0, λ_{i} > 0, a_{i} \in Ω} \end{matrix}$ has a unique solution (up to a permutation). Thus, we can uniquely write u as follows $\begin{matrix} u = \sum_{i = 1}^{p} α_{i} P δ_{a_{i}, λ_{i}} + v, \end{matrix}$ where $v \in H^{2} (Ω) \cap H_{0}^{1} (Ω) \cap T_{w} W_{s} (w)$ and satisfies $\begin{matrix} (V_{0}) : ⟨ v, ψ ⟩ = 0 for ψ \in {P δ_{i}, \frac{\partial P δ_{i}}{\partial λ_{i}}, \frac{\partial P δ_{i}}{\partial a_{i}}, i = 1, \dots, p} . \end{matrix}$ Here, $P δ_{i} = P δ_{a_{i}, λ_{i}}$ and $⟨ \cdot, \cdot ⟩$ denotes the inner product on $H^{2} (Ω) \cap H_{0}^{1} (Ω)$ defined by $\begin{matrix} ⟨ u, v ⟩ = \int_{Ω} Δ u Δ v . \end{matrix}$

The following Proposition deals with the v-part of u and shows that is negligible with respect to the concentration phenomenon.

Proposition 2.3 ([4,5]).

There is a $C^{1}$ -map which to each $(α_{i}, a_{i}, λ_{i})$ such that $\sum_{i = 1}^{p} α_{i} P δ_{a_{i}, λ_{i}}$ belongs to $V (p, ε)$ associates $\overline{v} = \overline{v} (α_{i}, a_{i}, λ_{i})$ such that $\overline{v}$ is the unique solution of the following minimization problem $\begin{matrix} min {J (\sum_{i = 1}^{p} α_{i} P δ_{a_{i}, λ_{i}} + v), v \in H_{0}^{1} (Ω) and satisfies (V_{0})} . \end{matrix}$ Moreover, there exists a change of variables $v - \overline{v} \to V$ such that $\begin{matrix} J (\sum_{i = 1}^{p} α_{i} P δ_{a_{i}, λ_{i}} + v) = J (\sum_{i = 1}^{p} α_{i} P δ_{a_{i}, λ_{i}} + \overline{v}) + ‖ V ‖^{2} . \end{matrix}$

We now state the definition of critical point at infinity.

Definition 2.4 ([4]).

A critical point at infinity of J is a limit of a non-compact flow line $u (s)$ of the gradient vector field $(- \partial J)$ . By Propositions 2.1 and 2.2 , $u (s)$ can be written as: $\begin{matrix} u (s) = \sum_{i = 1}^{p} α_{i} (s) P δ_{a_{i} (s), λ_{i} (s)} + v (s) . \end{matrix}$ Denoting by $y_{i} = {lim}_{s ⟶ + \infty} a_{i} (s) and α_{i} = {lim}_{s ⟶ + \infty} α_{i} (s)$ , we then denote by $\begin{matrix} \sum_{i = 1}^{p} α_{i} P δ_{y_{i}, \infty} or {(y_{1}, \dots, y_{p})}_{\infty} \end{matrix}$ such a critical point at infinity.

3. Expansion of the gradient of J

Let ρ be a positive small constant such that for any $y \in K$ , the expansion ${(f)}_{β}$ holds in $B (y, ρ)$ . Let $\begin{matrix} \tilde{V} (p, ε) : = {u = \sum_{i = 1}^{p} α_{i} P δ_{(a_{i}, λ_{i})} \in V (p, ε), a_{i} \in B (y_{ℓ_{i}}, ρ), y_{ℓ_{i}} \in K, \forall i = 1, \dots, p} . \end{matrix}$ Under the condition (A), there is no critical points at infinity when there is a concentration point $a_{i}$ outside $⋃_{y \in K} B (y, ρ)$ , see [14]. Therefore, to characterize the critical points at infinity, we will consider only the neighborhoods $\tilde{V} (p, ε)$ .

The following proposition gives the variation of J in $\tilde{V} (p, ε)$ with respect to $λ_{i}, i = 1, \dots, p$ .

Proposition 3.1.
Assume that K satisfies ${(f)}_{β}$ condition for $β > 1$ . For $u = \sum_{i = 1}^{p} α_{i} P δ_{a_{i}, λ_{i}} \in \tilde{V} (p, ε)$ , we have the following two estimates:
$\begin{array}{l} ⟨ \partial J (u), α_{i} λ_{i} \frac{\partial P δ_{a_{i}, λ_{i}}}{\partial λ_{i}} ⟩ \\ = - 2 c_{2} \frac{J (u)}{K (a_{i})} \sum_{j \neq i} α_{i} α_{j} (λ_{i} \frac{\partial ε_{i j}}{\partial λ_{i}} + \frac{H (a_{i}, a_{j})}{{(λ_{i} λ_{j})}^{\frac{n - 4}{2}}}) \\ + O (\sum_{j = 2}^{[β]} \frac{| a_{i} - y_{ℓ_{i}} |^{β - j}}{λ_{i}^{j}}) + O (\frac{1}{λ_{i}^{β}}) \\ + o (\sum_{j \neq i} (ε_{i j} + \frac{H (a_{i}, a_{j})}{{(λ_{i} λ_{j})}^{\frac{n - 4}{2}}})) . \end{array}$

$\begin{array}{l} ⟨ \partial J (u), α_{i} λ_{i} \frac{\partial P δ_{a_{i}, λ_{i}}}{\partial λ_{i}} ⟩ \\ = - 2 c_{2} \frac{J (u)}{K (a_{i})} \sum_{j \neq i} α_{i} α_{j} (λ_{i} \frac{\partial ε_{i j}}{\partial λ_{i}} + \frac{H (a_{i}, a_{j})}{{(λ_{i} λ_{j})}^{\frac{n - 4}{2}}}) \\ + 2 α_{i}^{2} \frac{J (u)}{K (a_{i})} \{\begin{matrix} \frac{n - 4}{2} c_{1} \frac{\sum_{k = 1}^{n} b_{k} (y_{ℓ_{i}})}{λ_{i}^{β (y_{ℓ_{i}})}}, & if β (y_{ℓ_{i}}) < n - 4 \\ \frac{n - 4}{2} c_{1} \frac{\sum_{k = 1}^{n} b_{k} (y_{ℓ_{i}})}{λ_{i}^{β (y_{ℓ_{i}})}} - c_{2} \frac{H (y_{ℓ_{i}}, y_{ℓ_{i}})}{λ_{i}^{n - 4}}, & if β (y_{ℓ_{i}}) = n - 4 \\ - c_{2} \frac{H (y_{ℓ_{i}}, y_{ℓ_{i}})}{λ_{i}^{n - 4}}, & if β (y_{ℓ_{i}}) > n - 4 \end{matrix} \\ + O (| a_{i} - y_{ℓ_{i}} |^{β}) + o (\sum_{j \neq i} (ε_{i j} + \frac{H (a_{i}, a_{j})}{{(λ_{i} λ_{j})}^{\frac{n - 4}{2}}}) + \frac{1}{λ_{i}^{n - 4}}) . \end{array}$
Here $c_{1} = \int_{R^{n}} | z_{1} |^{β} \frac{| z |^{2} - 1}{{(1 + | z |^{2})}^{n + 1}} d z$ and $c_{2} = \int_{R^{n}} \frac{| z |^{2} - 1}{{(1 + | z |^{2})}^{n}} d z$ .
Proof.
For $u = \sum_{i = 1}^{p} α_{i} P δ_{a_{i}, λ_{i}} \in \tilde{V} (p, ε)$ , we have: $\begin{matrix} \partial J (u) = 2 J (u) (u + J (u) Δ^{- 1} (K u^{\frac{n + 4}{n - 4}})) . \end{matrix}$ Thus, $\begin{array}{l} ⟨ \partial J (u), α_{i} λ_{i} \frac{\partial P δ_{a_{i}, λ_{i}}}{\partial λ_{i}} ⟩ = & 2 J (u) [\sum_{j = 1}^{p} α_{i} α_{j} ⟨ P δ_{a_{j}, λ_{j}}, λ_{i} \frac{\partial P δ_{a_{i}, λ_{i}}}{\partial λ_{i}} ⟩ \\ - J {(u)}^{\frac{n}{n - 4}} \int_{Ω} K {(\sum_{j = 1}^{p} α_{j} P δ_{a_{j}, λ_{j}})}^{\frac{n + 4}{n - 4}} α_{i} λ_{i} \frac{\partial P δ_{a_{i}, λ_{i}}}{\partial λ_{i}}] . \end{array}$

Using (2.3) and (2.4) and the fact that $J {(u)}^{\frac{n}{n - 4}} α_{i}^{\frac{4}{n - 4}} K (a_{i}) = 1 + o (1)$ , we get $\begin{array}{l} ⟨ \partial J (u), α_{i} λ_{i} \frac{\partial P δ_{a_{i}, λ_{i}}}{\partial λ_{i}} ⟩ \\ = - 2 c_{2} \frac{J (u)}{K (a_{i})} \sum_{j \neq i} α_{i} α_{j} (λ_{i} \frac{\partial ε_{i j}}{\partial λ_{i}} + \frac{H (a_{i}, a_{j})}{{(λ_{i} λ_{j})}^{\frac{n - 4}{2}}}) \\ - 2 α_{i}^{2} \frac{J (u)}{K (a_{i})} (\int_{Ω} K δ_{a_{j}, λ_{j}}^{\frac{n + 4}{n - 4}} λ_{i} \frac{\partial δ_{a_{i}, λ_{i}}}{\partial λ_{i}} + c_{2} \frac{H (a_{i}, a_{i})}{λ_{i}^{n - 4}}) + o (\sum_{j \neq i} (ε_{i j} + \frac{H (a_{i}, a_{j})}{{(λ_{i} λ_{j})}^{\frac{n - 4}{2}}}) + \frac{1}{λ_{i}^{n - 4}}) . \end{array}$ Observe that $\begin{matrix} δ_{a, λ}^{\frac{n + 4}{n - 4}} λ \frac{\partial δ_{a, λ}}{\partial λ} = \frac{n - 4}{2} λ^{n} \frac{1 - λ^{2} | x - a |^{2}}{{(1 + λ^{2} | x - a |^{2})}^{n + 1}} . \end{matrix}$

Let $μ > 0$ such that $B (a_{i}, μ) \subset B (y_{ℓ_{i}}, ρ)$ . We have $\begin{matrix} \int_{Ω} K (x) δ_{a_{i}, λ_{i}}^{\frac{n + 4}{n - 4}} λ \frac{\partial δ_{a_{i}, λ_{i}}}{\partial λ_{i}} d x = \int_{B (a_{i}, μ)} K (x) δ_{a_{i}, λ_{i}}^{\frac{n + 4}{n - 4}} λ_{i} \frac{\partial δ_{a_{i}, λ_{i}}}{\partial λ_{i}} d x + O (\frac{1}{λ_{i}^{n}}) . \end{matrix}$ After a change of variables $z = λ_{i} (x - a_{i})$ , $\begin{matrix} \int_{B (a_{i}, μ)} K (x) δ_{a_{i}, λ_{i}}^{\frac{n + 4}{n - 4}} λ_{i} \frac{\partial δ_{a_{i}, λ_{i}}}{\partial λ_{i}} d x = \frac{n - 4}{2} \int_{B (0, λ_{i} μ)} K (a_{i} + \frac{z}{λ_{i}}) \frac{1 - | z |^{2}}{{(1 + | z |^{2})}^{n + 1}} d z . \end{matrix}$ Expanding K around $a_{i}$ , we obtain $\begin{matrix} K (a_{i} + \frac{z}{λ_{i}}) = K (a_{i}) + \sum_{j = 1}^{[β]} \frac{D^{j} K (a_{i}) {(\frac{z}{λ_{i}})}^{j}}{j!} + O ({| \frac{z}{λ_{i}} |}^{β}), \end{matrix}$ since K is of class $C^{[β]}$ around each critical point. Using the fact that $\int_{B (0, λ_{i} μ)} K (a_{i}) \frac{1 - | z |^{2}}{{(1 + | z |^{2})}^{n + 1}} d z = O (\frac{1}{λ_{i}^{n}})$ , we get $\begin{array}{l} \int_{B (a_{i}, μ)} K (x) δ_{a_{i}, λ_{i}}^{\frac{n + 4}{n - 4}} λ_{i} \frac{\partial δ_{a_{i}, λ_{i}}}{\partial λ_{i}} d x \\ = \frac{n - 4}{2} \sum_{j = 1}^{[β]} \int_{B (0, λ_{i} μ)} \frac{D^{j} K (a_{i}) {(z)}^{j}}{λ_{i}^{j} j!} \frac{1 - | z |^{2}}{{(1 + | z |^{2})}^{n + 1}} d z \\ + O (\frac{1}{λ_{i}^{β}} \int_{R^{n}} | z_{k} |^{β} \frac{1 - | z |^{2}}{{(1 + | z |^{2})}^{n + 1}} d z) + O (\frac{1}{λ_{i}^{n}}) . \end{array}$

Observe that, $\begin{matrix} \int_{B (0, λ_{i} μ)} D K (a_{i}) (z) \frac{1 - | z |^{2}}{{(1 + | z |^{2})}^{n + 1}} d z = 0 . \end{matrix}$ Moreover, under ${(f)}_{β}$ -condition, we have $\begin{matrix} | D^{j} K (a_{i}) | = O (| a_{i} - y_{ℓ_{i}} |^{β - j}) . \end{matrix}$ Thus, $\begin{matrix} \int_{B (a_{i}, μ)} K (x) δ_{a_{i}, λ_{i}}^{\frac{n + 4}{n - 4}} λ_{i} \frac{\partial δ_{a_{i}, λ_{i}}}{\partial λ_{i}} d x = O (\sum_{j = 2}^{[β]} \frac{| a_{i} - y_{ℓ_{i}} |^{β - j})}{λ_{i}^{j}}) + O (\frac{1}{λ_{i}^{β}}) . \end{matrix}$ Hence, the estimate (a) of Proposition 3.1 follows.

For the estimate (b), ${(f)}_{β}$ -expansion yields $\begin{matrix} K (a_{i} + \frac{z}{λ_{i}}) = K (y_{ℓ_{i}}) + \sum_{k = 1}^{n} b_{k} {| \frac{z_{k}}{λ_{i}} + {(a_{i} - y_{ℓ_{i}})}_{k} |}^{β} + o ({| \frac{z}{λ_{i}} + (a_{i} - y_{ℓ_{i}}) |}^{β}) \end{matrix}$ Therefore, $\begin{array}{l} \int_{B (a_{i}, μ)} K (x) δ_{a_{i}, λ_{i}}^{\frac{n + 4}{n - 4}} λ_{i} \frac{\partial δ_{a_{i}, λ_{i}}}{\partial λ_{i}} d x \\ = \frac{n - 4}{2} \frac{1}{λ_{i}^{β}} \sum_{k = 1}^{n} b_{k} \int_{B (0, λ_{i} μ)} {| z_{k} + λ_{i} {(a_{i} - y_{ℓ_{i}})}_{k} |}^{β} \frac{1 - | z |^{2}}{{(1 + | z |^{2})}^{n + 1}} d z \\ + o (\frac{1}{λ_{i}^{β}} \int_{R^{n}} | z_{k} |^{β} \frac{1 - | z |^{2}}{{(1 + | z |^{2})}^{n + 1}} d z) + o (| a_{i} - y_{ℓ_{i}} |^{β} \int_{R^{n}} | z_{k} |^{β} \frac{1 - | z |^{2}}{{(1 + | z |^{2})}^{n + 1}} d z) \\ = \frac{n - 4}{2} \frac{1}{λ_{i}^{β}} \sum_{k = 1}^{n} b_{k} \int_{B (0, λ_{i} μ)} | z_{k} |^{β} \frac{1 - | z |^{2}}{{(1 + | z |^{2})}^{n + 1}} d z + O (| a_{i} - y_{ℓ_{i}} |^{β}) \\ + o (\frac{1}{λ_{i}^{β}} \int_{B (0, λ_{i} μ)} | z_{k} |^{β} \frac{| 1 - | z |^{2} |}{{(1 + | z |^{2})}^{n + 1}} d z) + o (\frac{1}{λ_{i}^{n - 4}}) . \end{array}$

Observe that, for $β < n$ , $\begin{matrix} \int_{B (0, λ_{i} μ)} | z_{k} |^{β} \frac{1 - | z |^{2}}{{(1 + | z |^{2})}^{n + 1}} d z = \int_{R^{n}} | z_{k} |^{β} \frac{1 - | z |^{2}}{{(1 + | z |^{2})}^{n + 1}} d z + O (\frac{1}{λ_{i}^{n - β}}) = - c_{1} + O (\frac{1}{λ_{i}^{n - β}}) . \end{matrix}$ For $β = n$ , $\begin{matrix} \int_{B (0, λ_{i} μ)} | z_{k} |^{β} \frac{1 - | z |^{2}}{{(1 + | z |^{2})}^{n + 1}} d z = O (log λ_{i}) . \end{matrix}$

Lastly, for $β > n$ , $\begin{matrix} \int_{B (0, λ_{i} μ)} | z_{k} |^{β} \frac{1 - | z |^{2}}{{(1 + | z |^{2})}^{n + 1}} d z = O (\frac{1}{λ_{i}^{n - β}}) . \end{matrix}$ Therefore, $\begin{array}{l} \int_{Ω} K (x) δ_{a, λ}^{\frac{n + 4}{n - 4}} λ \frac{\partial δ_{a, λ}}{\partial λ} d x \\ (3.1) & = (- \frac{n - 4}{2} c_{1} \frac{\sum_{k = 1}^{n} b_{k}}{λ_{i}^{β}} + o (\frac{1}{λ_{i}^{β}}), if β < n) + O (| a_{i} - y_{ℓ_{i}} |^{β}) + o (\frac{1}{λ_{i}^{n - 4}}) . \end{array}$ This conclude the proof of Proposition 3.1. □
Proposition 3.2.
Assume that K satisfies ${(f)}_{β}$ condition for $β > 1$ . Let $u = \sum_{i = 1}^{p} α_{i} P δ_{a_{i}, λ_{i}} \in \tilde{V} (p, ε)$ . For any $i = 1, \dots, p$ and $k = 1, \dots, n$ , we have the following expansions.
$\begin{array}{l} ⟨ \partial J (u), α_{i} \frac{1}{λ_{i}} \frac{\partial P δ_{a_{i}, λ_{i}}}{\partial {(a_{i})}_{k}} ⟩ \\ = - c_{3} α_{i}^{2} J (u) \frac{b_{k}}{λ_{i} K (a_{i})} β sign {(a_{i} - y_{ℓ_{i}})}_{k} {| {(a_{i} - y_{ℓ_{i}})}_{k} |}^{β - 1} \\ + O (\sum_{j = 2}^{min (n, β)} \frac{| a_{i} - y_{ℓ_{i}} |^{β - j}}{λ_{i}^{j}}) + O (\frac{1}{λ_{i}^{min (n, β)}}) + O (\frac{1}{λ_{i}^{n - 1}}) \\ (3.2) & + O (\sum_{j \neq i} | \frac{1}{λ_{i}} \frac{\partial ε_{i j}}{\partial a_{i}} |) . \end{array}$ Moreover, if $λ_{i} | a_{i} - y_{ℓ_{i}} |$ is bounded and $β < n + 1$ , we have

$\begin{array}{l} ⟨ \partial J (u), α_{i} \frac{1}{λ_{i}} \frac{\partial P δ_{a_{i}, λ_{i}}}{\partial {(a_{i})}_{k}} ⟩ \\ = - (n - 2) α_{i}^{2} J (u) \frac{b_{k}}{K (a_{i}) λ_{i}^{β}} \int_{R^{n}} {| z_{k} + λ_{i} {(a_{i} - y_{ℓ_{i}})}_{k} |}^{β} \\ (3.3) & \times \frac{z_{k}}{{(1 + | z |^{2})}^{n + 1}} d z + o (\frac{1}{λ_{i}^{β}}) + O (\sum_{j \neq i} | \frac{1}{λ_{i}} \frac{\partial ε_{i j}}{\partial a_{i}} |) . \end{array}$
Here $c_{3} = ((n - 2) c_{n}^{\frac{2 n}{n - 4}}) \int_{R^{n}} \frac{| z |^{2}}{{(1 + | z |^{2})}^{n + 1}} d z$ .
Proof.
We argue as in the proof of Proposition 3.1, $\begin{array}{l} ⟨ \partial J (u), α \frac{1}{λ} \frac{\partial P δ_{a, λ}}{\partial a_{k}} ⟩ \\ = \frac{2 α^{2} J (u)}{K (a)} (- \int_{Ω} K (x) δ_{a, λ}^{\frac{n + 4}{n - 4}} \frac{1}{λ} \frac{\partial δ_{a, λ}}{\partial a_{k}} d x + O (\frac{\frac{\partial H}{\partial a_{k}} (a, a)}{λ^{n - 1}}) + o (\frac{1}{λ^{n - 1}})) \\ + O (\sum_{j \neq i} | \frac{1}{λ_{i}} \frac{\partial ε_{i j}}{\partial a_{i}} |) \\ = - \frac{2 α^{2} J (u)}{K (a)} \int_{B (a, μ)} K (x) δ_{a, λ}^{\frac{n + 4}{n - 4}} \frac{1}{λ} \frac{\partial δ_{a, λ}}{\partial a_{k}} d x + O (\frac{1}{λ^{n - 1}}) + O (\sum_{j \neq i} | \frac{1}{λ_{i}} \frac{\partial ε_{i j}}{\partial a_{i}} |) . \end{array}$ Observe that $\begin{matrix} δ_{a, λ}^{\frac{n + 4}{n - 4}} \frac{1}{λ} \frac{\partial δ_{a, λ}}{\partial a_{k}} = (n - 4) c_{n}^{\frac{2 n}{n - 4}} \frac{λ^{n + 1} {(x - a)}_{k}}{{(1 + λ^{2} | x - a |^{2})}^{n + 1}} . \end{matrix}$ A change of variables $z = λ (x - a)$ yields $\begin{matrix} \int_{B (a, μ)} K (x) δ_{a, λ}^{\frac{n + 4}{n - 4}} \frac{1}{λ} \frac{\partial δ_{a, λ}}{\partial a_{k}} d x = (n - 4) c_{n}^{\frac{2 n}{n - 4}} \int_{B (0, λ μ)} K (a + z / λ) \frac{z_{k}}{{(1 + | z |^{2})}^{n + 1}} d z . \end{matrix}$ To get the first expansion of Proposition 3.2, we expand K as follows $\begin{matrix} K (a + z / λ) = K (a) + \sum_{j = 1}^{min (n, β)} \frac{D^{j} K (a) {(z / λ)}^{j}}{j!} + O (\frac{| z |^{min (β, n)}}{λ^{min (β, n)}}) . \end{matrix}$ Using the fact that $\int_{B (0, λ μ)} K (a) \frac{z_{k}}{{(1 + | z |^{2})}^{n + 1}} d z = 0$ , we get $\begin{array}{l} \int_{B (a, μ)} K (x) δ_{a, λ}^{\frac{n + 4}{n - 4}} \frac{1}{λ} \frac{\partial δ_{a, λ}}{\partial a_{k}} d x \\ = (n - 4) c_{n}^{\frac{2 n}{n - 4}} \frac{1}{λ} \int_{B (0, λ μ)} \frac{D K (a) (z) \cdot z_{k}}{{(1 + | z |^{2})}^{n + 1}} d z \\ + O (\sum_{j = 2}^{min (n, β)} \frac{1}{λ^{j}} \int_{R^{n}} \frac{| D^{j} K (a) | | z |^{j + 1}}{{(1 + | z |^{2})}^{n + 1}} d z) + O (\frac{1}{λ^{min (n, β)}} \int_{R^{n}} \frac{| z |^{min (n, β) + 1}}{{(1 + | z |^{2})}^{n + 1}} d z) . \end{array}$ Observe that $\begin{array}{l} \int_{B (0, λ μ)} \frac{D K (a) (z) \cdot z_{k}}{{(1 + | z |^{2})}^{n + 1}} d z \\ = \sum_{j = 1}^{n} \frac{\partial K (a)}{\partial x_{j}} \int_{B (0, λ μ)} \frac{z_{j} z_{k}}{{(1 + | z |^{2})}^{n + 1}} d z \\ = \frac{1}{n} \frac{\partial K (a)}{\partial x_{k}} \int_{B (0, λ μ)} \frac{| z |^{2}}{{(1 + | z |^{2})}^{n + 1}} d z = \frac{1}{n} \frac{\partial K (a)}{\partial x_{k}} \int_{R^{n}} \frac{| z |^{2}}{{(1 + | z |^{2})}^{n + 1}} d z + O (\frac{1}{λ^{n}}), \end{array}$ since $\int_{B (0, λ μ)} \frac{z_{j} z_{k}}{{(1 + | z |^{2})}^{n + 1}} d z = 0, \forall j \neq k$ . Using now the fact that $a \in B (y, ρ)$ , we derive from ${(f)}_{β}$ -condition that $\begin{array}{l} \frac{\partial K}{\partial x_{k}} (a) & = b_{k} β sign {(a - y)}_{k} {| {(a - y)}_{k} |}^{β - 1} + \frac{\partial R}{\partial x_{k}} (a - y) \\ = b_{k} β sign {(a - y)}_{k} {| {(a - y)}_{k} |}^{β - 1} + o (| x - y |^{β - 1}) . \end{array}$

Moreover, for every $j = 2, \dots, \tilde{β}$ , $\begin{matrix} | D^{j} K (a) | = O (| a - y |^{β - j}) . \end{matrix}$ Thus, $\begin{array}{l} \int_{B (a, μ)} K (x) δ_{a, λ}^{\frac{n + 4}{n - 4}} \frac{1}{λ} \frac{\partial δ_{a, λ}}{\partial a_{k}} d x = & c_{3} \frac{b_{k}}{λ} sign {(a - y)}_{k} {| {(a - y)}_{k} |}^{β - 1} + o (\frac{| a - y |^{β - 1}}{λ}) \\ + O (\sum_{j = 2}^{\tilde{β}} \frac{| a - y |^{β - j}}{λ^{j}}) + O (\frac{1}{λ^{min (n, β)}}) . \end{array}$ This finishes the proof of (a) of Proposition 3.2. Concerning the estimate (b), it follows from the above arguments and the following estimate $\begin{matrix} \int_{B (a, μ)} K (x) δ_{a, λ}^{\frac{n + 4}{n - 4}} \frac{1}{λ} \frac{\partial δ_{a, λ}}{\partial a_{k}} d x = (n - 4) c_{n}^{\frac{2 n}{n - 4}} \frac{b_{k}}{λ} \int_{R^{n}} {| z_{k} + λ {(a - y)}_{k} |}^{β} \frac{z_{k}}{{(1 + | z |^{2})}^{n + 1}} d z + o (\frac{1}{λ^{β}}) . \end{matrix}$ This finishes the proof of Proposition 3.2. □

4. Lack of compactness and critical points at infinity

In the first part of this section, we focus on $\tilde{V} (1, ε)$ ; the neighborhood of critical points at infinity consisting by single masses. We study the concentration phenomenon in this set and we identify the related critical points at infinity. Let $ρ > 0$ small enough such that for any $y \in K$ , the expansion ${(f)}_{β}$ holds in $B (y, ρ)$ and let: $\begin{array}{l} {\tilde{V}}_{1} (1, ε) = {u = α P δ_{(a, λ)} \in \tilde{V} (1, ε), a \in B (y, ρ), y \in K with β = β (y) < n - 4}, \\ {\tilde{V}}_{2} (1, ε) = {u = α P δ_{(a, λ)} \in \tilde{V} (1, ε), a \in B (y, ρ), y \in K with β = β (y) = n - 4}, \\ {\tilde{V}}_{3} (1, ε) = {u = α P δ_{(a, λ)} \in \tilde{V} (1, ε), a \in B (y, ρ), y \in K with β = β (y) > n - 4} \end{array}$

As in [4], see also [2], the characterization of the critical points at infinity in $\tilde{V} (1, ε)$ is obtained through the construction of a suitable decreasing pseudo-gradient satisfying the P.S condition as long as the concentration point $a (s)$ does not enter in a neighborhood of $y \in K_{< n - 4}^{+} \cup K_{n - 4}^{+} \cup K_{> n - 4}$ .

Let δ be a small positive constant and let $θ_{1}, θ_{2}$ and $θ_{3}$ be the following three cut-off functions $\begin{array}{l} θ_{1} : R ⟶ R \\ t ⟼ \{\begin{matrix} 1 & if | t | ⩽ \frac{δ}{2} \\ 0 & if | t | ⩾ δ . \end{matrix} \\ θ_{2} : R ⟶ R \\ t ⟼ \{\begin{matrix} 1 & if \frac{δ}{2} ⩽ | t | ⩽ \frac{1}{δ} \\ 0 & if | t | \in [0, \frac{δ}{4}] \cup [\frac{2}{δ}, + \infty [. \end{matrix} \\ θ_{3} : R ⟶ R \\ t ⟼ \{\begin{matrix} 1 & if | t | ⩾ \frac{1}{δ} \\ 0 & if | t | ⩽ \frac{1}{2 δ} . \end{matrix} \end{array}$

∙ Pseudo-gradient in ${\tilde{V}}_{1} (1, ε)$ :

Let $W_{1}$ be the following vector field. $\forall u = α P δ_{(a, λ)} \in {\tilde{V}}_{1} (1, ε)$ , $\begin{array}{l} W_{1} (u) = & - θ_{1} (λ | a - y |) (\sum_{k = 1}^{n} b_{k}) α λ \frac{\partial P δ_{(a, λ)}}{\partial λ} \\ + θ_{3} (λ | a - y |) (\sum_{k = 1}^{n} b_{k}) sign {(a - y)}_{k} α \frac{1}{λ} \frac{\partial P δ_{(a, λ)}}{\partial a_{k}} \\ + θ_{2} (λ | a - y |) (\sum_{k = 1}^{n} b_{k}) \int_{R^{n}} {| z_{k} + λ {(a - y)}_{k} |}^{β} \frac{z_{k}}{{(1 + | z |^{2})}^{n + 1}} d z α \frac{1}{λ} \frac{\partial P δ_{(a, λ)}}{\partial a_{k}} . \end{array}$ We claim that $\begin{matrix} (4.1) & ⟨ \partial J (u), W_{1} (u) ⟩ ⩽ - c (\frac{1}{λ^{β}} + \frac{| \nabla K (a) |}{λ}) . \end{matrix}$ Indeed, if $λ | a - y | ⩽ δ$ , by Proposition 3.1, we have $\begin{matrix} (4.2) & ⟨ \partial J (u), - (\sum_{k = 1}^{n} b_{k}) α λ \frac{\partial P δ_{(a, λ)}}{\partial λ} ⟩ ⩽ - c \frac{{(\sum_{k = 1}^{n} b_{k})}^{2}}{λ^{β}}, \end{matrix}$ since $| a - y |^{β} = o (\frac{1}{λ^{β}})$ as δ small enough. Observe that under $(f_{β})$ -condition, we have $\begin{matrix} (4.3) & | \nabla K (a) | = O (| a - y |^{β - 1}), \end{matrix}$ thus $\begin{matrix} (4.4) & \frac{| \nabla K (a) |}{λ} = O (\frac{1}{λ^{β}}), if λ | a - y | is bounded . \end{matrix}$ Therefore, we can appear $- \frac{| \nabla K (a) |}{λ}$ in the upper bound of (4.2) and we obtain $\begin{matrix} (4.5) & ⟨ \partial J (u), - (\sum_{k = 1}^{n} b_{k}) α λ \frac{\partial P δ_{(a, λ)}}{\partial λ} ⟩ ⩽ - c (\frac{1}{λ^{β}} + \frac{| \nabla K (a) |}{λ}) . \end{matrix}$

If $λ | a - y | \in [\frac{δ}{4}, \frac{2}{δ}]$ , by the second expansion of Proposition 3.2, we obtain $\begin{array}{l} ⟨ \partial J (u), \sum_{k = 1}^{n} b_{k} \int_{R^{n}} {| z_{k} + λ {(a - y)}_{k} |}^{β} \frac{z_{k}}{{(1 + | z |^{2})}^{n + 1}} d z α \frac{1}{λ} \frac{\partial P δ_{(a, λ)}}{\partial a_{k}} ⟩ \\ ⩽ - \frac{c}{λ^{β}} {(\int_{R^{n}} {| z_{k_{a}} + λ {(a - y)}_{k_{a}} |}^{β} \frac{z_{k_{a}}}{{(1 + | z |^{2})}^{n + 1}} d z)}^{2} + o (\frac{1}{λ^{β}}), \end{array}$ where $| {(a - y)}_{k_{a}} | = {max}_{1 ⩽ k ⩽ n} | {(a - y)}_{k} |$ . Since $λ | {(a - y)}_{k_{a}} | ⩾ \frac{1}{\sqrt{n}} \frac{δ}{4}$ , we get $\begin{matrix} {(\int_{R^{n}} {| z_{k_{a}} + λ {(a - y)}_{k_{a}} |}^{β} \frac{z_{k_{a}}}{{(1 + | z |^{2})}^{n + 1}} d z)}^{2} ⩾ c_{δ} > 0 . \end{matrix}$ Thus, $\begin{array}{l} ⟨ \partial J (u), \sum_{k = 1}^{n} b_{k} \int_{R^{n}} {| z_{k} + λ {(a - y)}_{k} |}^{β} \frac{z_{k}}{{(1 + | z |^{2})}^{n + 1}} d z α \frac{1}{λ} \frac{\partial P δ_{(a, λ)}}{\partial a_{k}} ⟩ & ⩽ - \frac{c_{1}}{λ^{β}}, \\ ⩽ - c (\frac{1}{λ^{β}} + \frac{| \nabla K (a) |}{λ}) . \end{array}$

Lastly, if $λ | a - y | ⩾ \frac{1}{2 δ}$ , by the first expansion of Proposition 3.2, we have $\begin{array}{l} ⟨ \partial J (u), \sum_{k = 1}^{n} b_{k} sign {(a - y)}_{k} α \frac{1}{λ} \frac{\partial P δ_{(a, λ)}}{\partial a_{k}} ⟩ ⩽ & - c \sum_{k = 1}^{n} b_{k}^{2} \frac{| a - y |^{β - 1}}{λ} \\ + O (\sum_{j = 2}^{\tilde{β}} \frac{| a - y |^{β - j}}{λ^{j}}) + O (\frac{1}{λ^{β}}) . \end{array}$ Oberve that for every $j = 2, \dots, \tilde{β}$ $\begin{matrix} \frac{| a - y |^{β - j}}{λ^{j}} = o (\frac{| a - y |^{β - 1}}{λ}), as δ small enough . \end{matrix}$ Also, $\begin{matrix} (4.6) & \frac{1}{λ^{β}} = o (\frac{| a - y |^{β - 1}}{λ}), as δ small enough . \end{matrix}$ Then, we obtain $\begin{matrix} ⟨ \partial J (u), \sum_{k = 1}^{n} b_{k} sign {(a - y)}_{k} α \frac{1}{λ} \frac{\partial P δ_{(a, λ)}}{\partial a_{k}} ⟩ ⩽ - c \frac{| a - y |^{β - 1}}{λ}, \end{matrix}$ since $\sum_{k = 1}^{n} | {(a - y)}_{k} |^{β - 1} \sim | a - y |^{β - 1}$ . Now by (4.3) and (4.6), we derive from the above inequality that $\begin{matrix} ⟨ \partial J (u), \sum_{k = 1}^{n} b_{k} sign {(a - y)}_{k} α \frac{1}{λ} \frac{\partial P δ_{(a, λ)}}{\partial a_{k}} ⟩ ⩽ - c (\frac{1}{λ^{β}} + \frac{| \nabla K (a) |}{λ}) . \end{matrix}$ Hence claim (4.1) follows.

∙ Pseudo-gradient in ${\tilde{V}}_{2} (1, ε)$ :

Let $W_{2}$ be the following vector field. $\forall u = α P δ_{(a, λ)} \in {\tilde{V}}_{2} (1, ε)$ , $\begin{array}{l} W_{2} (u) = & θ_{1} (λ | a - y |) α λ \frac{\partial P δ_{(a, λ)}}{\partial λ} + θ_{2} (λ | a - y |) \sum_{k = 1}^{n} b_{k} \int_{R^{n}} \frac{z_{k} | z_{k} + λ {(a - y)}_{k} |^{β}}{{(1 + | z |^{2})}^{n + 1}} d z \\ + θ_{3} (λ | a - y |) \sum_{k = 1}^{n} b_{k} sign {(a - y)}_{k} α \frac{1}{λ} \frac{\partial P δ_{(a, λ)}}{\partial a_{k}} . \end{array}$ Observe that, if $λ | a - y | ⩽ δ$ , by the expansion of Proposition 3.1, we get $\begin{array}{l} ⟨ \partial J (u), α λ \frac{\partial P δ_{(a, λ)}}{\partial λ} ⟩ ⩽ & - c \frac{H (a, a)}{λ^{n - 4}} + O (| a - y |^{n - 4}) \\ ⩽ & - \frac{c}{λ^{n - 4}}, \end{array}$ since $| a - y |^{n - 4} = o (\frac{1}{λ^{n - 4}})$ as δ small enough and H is a positive regular function on $Ω^{2}$ . Thus by (4.4), we obtain $\begin{matrix} ⟨ \partial J (u), α λ \frac{\partial P δ_{(a, λ)}}{\partial λ} ⟩ ⩽ - c (\frac{1}{λ^{n - 4}} + \frac{| \nabla K (a) |}{λ}) . \end{matrix}$ If $λ | a - y | ⩾ \frac{δ}{2}$ , we proceed exactly as in ${\tilde{V}}_{1} (1, ε)$ . We therefore obtain $\begin{matrix} (4.7) & ⟨ \partial J (u), W_{2} (u) ⟩ ⩽ - c (\frac{1}{λ^{n - 4}} + \frac{| \nabla K (a) |}{λ}) . \end{matrix}$

∙ Pseudo-gradient in ${\tilde{V}}_{3} (1, ε)$ :

Let $W_{3}$ be the following vector field. $\forall u = α P δ_{(a, λ)} \in {\tilde{V}}_{3} (1, ε)$ , $\begin{matrix} W_{3} (u) = θ_{1} (λ^{n - 4} | a - y |^{β}) α λ \frac{\partial P δ_{(a, λ)}}{\partial λ} + (1 - θ_{1}) (λ^{n - 4} | a - y |^{β}) \sum_{k = 1}^{n} b_{k} sign {(a - y)}_{k} α \frac{1}{λ} \frac{\partial P δ_{(a, λ)}}{\partial a_{k}} . \end{matrix}$ We claim that $\begin{matrix} (4.8) & ⟨ \partial J (u), W_{3} (u) ⟩ ⩽ - c (\frac{1}{λ^{min (n - 1, β)}} + \frac{| \nabla K (a) |}{λ}) . \end{matrix}$ Indeed, if $λ^{n - 4} | a - y |^{β} ⩽ δ$ in the expansion of Proposition 3.1, we have $\begin{matrix} O (| a - y |^{β}) = o (\frac{1}{λ^{n - 4}}), taking δ small enough, \end{matrix}$ Thus, $\begin{matrix} ⟨ \partial J (u), α λ \frac{\partial P δ_{(a, λ)}}{\partial λ} ⟩ ⩽ - \frac{c}{λ^{n - 4}} . \end{matrix}$ Moreover, $\begin{matrix} \frac{| \nabla K (a) |}{λ} = O (\frac{| a - y |^{β - 1}}{λ}) = O (\frac{1}{λ^{1 + \frac{(n - 4) (β - 1)}{β}}}) = o (\frac{1}{λ^{n - 4}}) \end{matrix}$ Therefore, we can appear $\frac{\nabla K (a)}{λ}$ in the latest upper bound. Hence $\begin{array}{l} ⟨ \partial J (u), α λ \frac{\partial P δ_{(a, λ)}}{\partial λ} ⟩ ⩽ & - c (\frac{1}{λ^{n - 2}} + \frac{| \nabla K (a) |}{λ}) \\ ⩽ & - c (\frac{1}{λ^{min (n - 1, β)}} + \frac{| \nabla K (a) |}{λ}) . \end{array}$ Now, if $λ^{n - 4} | a - y |^{β} > \frac{δ}{2}$ , by the first expansion of Proposition 3.2, we have $\begin{array}{l} ⟨ \partial J (u), \sum_{k = 1}^{n} b_{k} sign {(a - y)}_{k} α \frac{1}{λ} \frac{\partial P δ_{(a, λ)}}{\partial a_{k}} ⟩ ⩽ & - c \sum_{k = 1}^{n} b_{k}^{2} \frac{| a - y |^{β - 1}}{λ} + O (\sum_{j = 2}^{\tilde{β}} \frac{| a - y |^{β - j}}{λ^{j}}) \\ + O (\frac{1}{λ^{min (β, n)}}) + O (\frac{1}{λ^{n - 1}}) . \end{array}$ Oberve that for every $j = 2, \dots, n$ $\begin{matrix} \frac{| a - y |^{β - j}}{λ^{j}} = o (\frac{| a - y |^{β - 1}}{λ}), as λ goes to + \infty . \end{matrix}$ Indeed, $\begin{matrix} \frac{| a - y |^{β - j}}{λ^{j}} \frac{λ}{| a - y |^{β - 1}} = \frac{1}{{(λ | a - y |)}^{j - 1}} ⩽ {(\frac{2}{δ})}^{\frac{j - 1}{β}} \frac{1}{λ^{(j - 1) (1 - \frac{n - 4}{β})}}, \end{matrix}$ which goes to zero when λ goes to $+ \infty$ . In addition, $\begin{matrix} (4.9) & \frac{1}{λ^{n - 1}} = o (\frac{| a - y |^{β - 1}}{λ}), and \frac{1}{λ^{min (β, n)}} = o (\frac{| a - y |^{β - 1}}{λ}) as λ \to + \infty . \end{matrix}$ Therefore, $\begin{matrix} ⟨ \partial J (u), \sum_{k = 1}^{n} b_{k} sign {(a - y)}_{k} α \frac{1}{λ} \frac{\partial P δ_{(a, λ)}}{\partial a_{k}} ⟩ ⩽ - c \frac{| a - y |^{β - 1}}{λ} \end{matrix}$ and by (4.2) and (4.9), we obtain $\begin{matrix} (4.10) & ⟨ \partial J (u), \sum_{k = 1}^{n} b_{k} sign {(a - y)}_{k} α \frac{1}{λ} \frac{\partial P δ_{(a, λ)}}{\partial a_{k}} ⟩ ⩽ - c (\frac{1}{λ^{min (n - 1, β)}} + \frac{| \nabla K (a) |}{λ}) . \end{matrix}$ Hence our claim (4.8) follows.

Proof of Theorem 1.3.
Let W the vector field in $\tilde{V} (1, ε)$ defined by convex combination of $W_{1}, W_{2}$ and $W_{3}$ . By (4.1), (4.7) and (4.8), we have $\begin{matrix} ⟨ \partial J (u), W (u) ⟩ ⩽ - c (\frac{1}{λ^{min (n - 1, β)}} + \frac{| \nabla K (a) |}{λ}), \end{matrix}$ for every $u = α P δ_{(a, λ)} \in \tilde{V} (1, ε)$ . Using the argument of ([7], Appendix B), we obtain $\begin{matrix} ⟨ \partial J (u + \bar{v}), W (u) + \frac{\partial \bar{v}}{\partial (α, a, λ)} W (u) ⟩ ⩽ - c (\frac{1}{λ^{min (n - 1, β)}} + \frac{| \nabla K (a) |}{λ}) . \end{matrix}$

In the above construction of W, we observe that the Palais–Smale condition is satisfied along the decreasing flow lines of the pseudo-gradient W as long as the concentration points $a (s)$ of the flow do not enter in some neighborhood of any critical point $y \in K_{< n - 4}^{+} \cup K_{n - 4}^{+} \cup K_{> n - 4}$ , since $λ (s)$ decreases on the flow line in this region. However, if $a (s)$ is near a critical point $y \in K_{< n - 4}^{+} \cup K_{n - 4}^{+} \cup K_{> n - 4}$ , $λ (s)$ increases on the flow line and goes to $+ \infty$ . Thus, we obtain a critical point at infinity.

In this statement, the functional J can be expended after a suitable change of variables as $\begin{matrix} J (α P δ_{a, λ} + \bar{v}) = J (\tilde{α} P δ_{\tilde{a}, \tilde{λ}}) = \frac{S_{n}}{{\tilde{α}}^{\frac{8}{n - 4}} {(K (\tilde{a}))}^{\frac{n - 4}{4}}} (1 + \frac{1}{{\tilde{λ}}^{β}}) . \end{matrix}$ Thus, the index of such critical point at infinity is $n - \tilde{i} (y)$ . Since J behaves in this region as $\frac{1}{K^{\frac{n - 4}{4}}}$ . This conclude the proof of Theorem 1.3. □

In the second part of this section, we focus on $\tilde{V} (p, ε), p ⩾ 2$ . We characterize the critical points at infinity in these sets in order to give a complete description of the loss of compactness of problem (1.1) under ${(f)}_{β}$ -condition, where $β \in (1, n - 4]$ . Let $\begin{array}{l} {\tilde{V}}_{< n - 4} (p, ε) = & {u = \sum_{i = 1}^{p} α_{i} P δ_{(a_{i}, λ_{i})} + v \in \tilde{V} (p, ε), a_{i} \in B (y_{ℓ_{i}}, ρ), y_{ℓ_{i}} \in K \\ with β = β (y_{ℓ_{i}}) < n - 4, \forall i = 1, \dots, p}, \\ {\tilde{V}}_{n - 4} (p, ε) = & {u = \sum_{i = 1}^{p} α_{i} P δ_{(a_{i}, λ_{i})} + v \in \tilde{V} (p, ε), a_{i} \in B (y_{ℓ_{i}}, ρ), y_{ℓ_{i}} \in K \\ with β = β (y_{ℓ_{i}}) = n - 4, \forall i = 1, \dots, p} . \end{array}$

We introduce now the following two Lemmas.
Lemma 4.1.
There exists a pseudo-gradient $W_{1}$ in ${\tilde{V}}_{< n - 4} (p, ε)$ such that for any $u = \sum_{i = 1}^{p} α_{i} P δ_{a_{i}, λ_{i}} \in {\tilde{V}}_{< n - 4} (p, ε)$ , we have $\begin{matrix} ⟨ \partial J (u), W_{1} (u) ⟩ ⩽ - c (\sum_{i = 1}^{p} (\frac{1}{λ_{i}^{β}} + \frac{| \nabla K (a_{i}) |}{λ_{i}}) + \sum_{j \neq i} ε_{i j}) . \end{matrix}$ Moreover, the only situation when the $λ_{i} (s), i = 1, \dots, p, s ⩾ 0$ , are not bounded is when $a_{i} (s)$ goes to $y_{ℓ_{i}} \in K_{< n - 4}^{+}, \forall i = 1, \dots, p$ with $y_{ℓ_{i}} \neq y_{ℓ_{j}}, \forall i \neq j$ .
Lemma 4.2.
There exists a pseudo-gradient $W_{2}$ in ${\tilde{V}}_{n - 4} (p, ε)$ such that for any $u = \sum_{i = 1}^{p} α_{i} P δ_{a_{i}, λ_{i}} \in {\tilde{V}}_{n - 4} (p, ε)$ , we have $\begin{matrix} ⟨ \partial J (u), W_{2} (u) ⟩ ⩽ - c (\sum_{i = 1}^{p} (\frac{1}{λ_{i}^{n - 4}} + \frac{| \nabla K (a_{i}) |}{λ_{i}}) + \sum_{j \neq i} ε_{i j}) . \end{matrix}$ Moreover, the only situation when the $λ_{i} (s), i = 1, \dots, p, s ⩾ 0$ , are not bounded is when $a_{i} (s)$ goes to $y_{ℓ_{i}} \in K_{n - 4}^{+}, \forall i = 1, \dots, p$ with $y_{ℓ_{i}} \neq y_{ℓ_{j}}, \forall i \neq j$ and $ρ (y_{ℓ_{1}}, \dots, y_{ℓ_{p}}) > 0$ .

The proof of Lemmas 4.1 and 4.2 will be given at the end of this section. We now state the proof of Theorem 1.3. Proof of Theorem 1.3.
It follows from the following lemma. □
Lemma 4.3.
Under the assumption that K satisfies $(B)$ and ${(f)}_{β}$ for $β \in (1, n - 4]$ , there exists a pseudo-gradient W in $\tilde{V} (p, ε)$ such that for any $u = \sum_{i = 1}^{p} α_{i} P δ_{a_{i}, λ_{i}} \in \tilde{V} (p, ε)$ , we have $\begin{matrix} (4.11) & ⟨ \partial J (u), \tilde{W} (u) ⟩ ⩽ - c (\sum_{i = 1}^{p} (\frac{1}{λ_{i}^{β}} + \frac{| \nabla K (a_{i}) |}{λ_{i}}) + \sum_{j \neq i} ε_{i j}) . \end{matrix}$ Moreover, the only case where $λ_{i} (s), i = 1, \dots, p, s ⩾ 0$ are not bounded is when $a_{i} (s)$ goes to $y_{ℓ_{i}} \in K_{< n - 4}^{+} \cup K_{n - 4}^{+}, \forall i = 1, \dots, p$ with $y_{ℓ_{i}} \neq y_{ℓ_{j}}, \forall i \neq j$ and $(y_{ℓ_{1}}, \dots, y_{ℓ_{p}}) \in C_{n - 4}^{\infty} \cup C_{< n - 4}^{\infty} \cup (C_{n - 4}^{\infty} \times C_{< n - 4}^{\infty})$ .
Proof of Lemma 4.3.
Let $u = \sum_{i = 1}^{p} α_{i} P δ_{a_{i}, λ_{i}} \in \tilde{V} (p, ε), p ⩾ 2$ . By Lemmas 4.1 and 4.2, it remains only to consider the case where $\begin{matrix} u = \sum_{i \in I_{1}} α_{i} P δ_{a_{i}, λ_{i}} + \sum_{i \in I_{2}} α_{i} P δ_{a_{i}, λ_{i}}, \end{matrix}$ with $I_{1} \neq \emptyset$ , $I_{2} \neq \emptyset$ , $u_{1} : = \sum_{i \in I_{1}} α_{i} P δ_{a_{i}, λ_{i}} \in {\tilde{V}}_{< n - 4} (♯ I_{1}, ε)$ and $u_{2} : = \sum_{i \in I_{2}} α_{i} P δ_{a_{i}, λ_{i}} \in {\tilde{V}}_{n - 4} (♯ I_{2}, ε)$ . We order the $λ_{i}$ ’s, we can assume that $\begin{matrix} λ_{1} ⩽ λ_{2} ⩽ \dots ⩽ λ_{p} . \end{matrix}$ Three cases may occur.

∙ First case: $u_{1} \notin {u = \sum_{j = 1}^{♯ I_{1}} α_{j} P δ_{a_{j}, λ_{j}} \in {\tilde{V}}_{< n - 4} (♯ I_{1}, ε) with y_{ℓ_{j}} \in K_{< n - 4}^{+}, y_{ℓ_{j}} \neq y_{ℓ_{k}}, \forall 1 ⩽ j \neq k ⩽ ♯ I_{1}}$ .

Let ${\tilde{W}}_{1} (u) = W_{1} (u_{1})$ where $W_{1}$ is the pseudo-gradient defined in Lemma 4.1. Observe that the maximum of the $λ_{i} (s), i \in I_{1}$ does not increase through $W_{1}$ . Moreover, by Lemma 4.1, we have $\begin{array}{l} ⟨ \partial J (u), {\tilde{W}}_{1} (u) ⟩ ⩽ & - c (\sum_{i \in I_{1}} (\frac{1}{λ_{i}^{β}} + \frac{| \nabla K (a_{i}) |}{λ_{i}}) + \sum_{i, j \in I_{1}, j \neq i} ε_{i j}) \\ (4.12) & + O (\sum_{i \in I_{1}, j \notin I_{1}} ε_{i j}) . \end{array}$ Let $k_{0}$ be an index in $I_{1}$ such that $\begin{matrix} λ_{k_{0}}^{β (y_{ℓ_{k_{0}}})} = min {λ_{i}^{β (y_{ℓ_{i}})}, i \in I_{1}} . \end{matrix}$ Define $\begin{matrix} {\tilde{I}}_{1} = {i, 1 ⩽ i ⩽ p, s.t. λ_{i}^{β (y_{ℓ_{i}})} ⩾ \frac{1}{10} λ_{k_{0}}^{β (y_{ℓ_{k_{0}}})}} . \end{matrix}$ Observe that $I_{1} \subset {\tilde{I}}_{1}$ . Our first goal is to make appears in the upper bound of (4.12) all indices $i \in {\tilde{I}}_{1}$ . For each index i we define the following vector field. $\begin{matrix} (4.13) & X_{i} (u) = \sum_{k = 1}^{n} b_{k} sign {(a_{i} - y_{ℓ_{i}})}_{k} \frac{1}{λ_{i}} \frac{\partial P δ_{a_{i}, λ_{i}}}{\partial {(a_{i})}_{k}} . \end{matrix}$ Using the first expansion of Lemma 3.2, we have $\begin{array}{l} ⟨ \partial J (u), X_{i} (u) ⟩ ⩽ & - \frac{c_{3}}{K (a_{i})} α_{i}^{2} J (u) \sum_{k = 1}^{n} b_{k}^{2} \frac{| a_{i} - y_{ℓ_{i}} |^{β - 1}}{λ_{i}} \\ (4.14) & + O (\sum_{j = 2}^{[β]} \frac{| a_{i} - y_{ℓ_{i}} |^{β - j}}{λ_{i}^{j}}) + O (\frac{1}{λ_{i}^{β}}) + o (\sum_{j \neq i} ε_{i j}), \end{array}$ since $| \frac{1}{λ_{i}} \frac{\partial ε_{i j}}{\partial a_{i}} | = o (ε_{i j})$ if $| a_{i} - a_{j} | > ρ$ . Let M be a large positive constant. Observe that if $| λ_{i} (a_{i} - y_{ℓ_{i}}) | ⩽ M$ , $\begin{matrix} \frac{| a_{i} - y_{ℓ_{i}} |^{β - j}}{λ_{i}^{j}} = O (\frac{1}{λ_{i}^{β}}), \end{matrix}$ and if $| λ_{i} (a_{i} - y_{ℓ_{i}}) | ⩾ M$ , $\begin{matrix} \frac{| a_{i} - y_{ℓ_{i}} |^{β - j}}{λ_{i}^{j}} = o (\frac{| a_{i} - y_{ℓ_{i}} |^{β - 1}}{λ_{i}}), \end{matrix}$ and $\begin{matrix} \frac{1}{λ_{i}^{β}} = o (\frac{| a_{i} - y_{ℓ_{i}} |^{β - 1}}{λ_{i}}) as M large . \end{matrix}$ Therefore, for $m_{1} > 0$ very small, we get from (4.12) and (4.14) $\begin{array}{l} ⟨ \partial J (u), {\tilde{W}}_{1} (u) + m_{1} \sum_{i \in {\tilde{I}}_{1} ∖ I_{1}} X_{i} (u) ⟩ ⩽ & - c (\sum_{i \in {\tilde{I}}_{1}} (\frac{1}{λ_{i}^{β}} + \frac{| \nabla K (a_{i}) |}{λ_{i}}) + \sum_{i, j \in I_{1}, j \neq i} ε_{i j}) \\ (4.15) & + o (\sum_{j = 1}^{p} \frac{1}{λ_{j}^{β}}), \end{array}$ since under ${(f)}_{β}$ -condition $| \nabla K (a_{i}) | = O (| a_{i} - y_{ℓ_{i}} |^{β - 1})$ and $\begin{matrix} ε_{i j} = O (\frac{1}{{(λ_{i} λ_{j})}^{\frac{n - 2}{2}}}) = o (\frac{1}{λ_{i}^{β}} + \frac{1}{λ_{j}^{β}}), \forall i \in I_{1}, j \notin I_{1} . \end{matrix}$

In order to appear $- \sum_{i, j \in {\tilde{I}}_{1}, j \neq i} ε_{i j}$ we will decrease all the $λ_{i}, i \in {\tilde{I}}_{1} ∖ I_{1}$ with different speed. Let $Z_{1} (u) = \sum_{i \in {\tilde{I}}_{1} ∖ I_{1}} - 2^{i} λ_{i} \frac{\partial P δ_{a_{i}, λ_{i}}}{\partial λ_{i}}$ . Observe that $\begin{matrix} 2^{i} λ_{i} \frac{\partial ε_{i j}}{\partial λ_{i}} + 2^{j} λ_{j} \frac{\partial ε_{i j}}{\partial λ_{j}} ⩽ - c ε_{i j}, \forall i < j . \end{matrix}$ Thus, using the first expansion of Proposition 3.1, we get $\begin{matrix} ⟨ \partial J (u), Z_{1} (u) ⟩ ⩽ - c \sum_{i, j \in {\tilde{I}}_{1} ∖ I_{1}, j \neq i} ε_{i j} + O (\sum_{i \in {\tilde{I}}_{1} ∖ I_{1}} \frac{1}{λ_{i}^{β}} + \sum_{j = 2}^{[β]} \frac{| a_{i} - y_{ℓ_{i}} |^{β - j}}{λ_{i}^{j}}) . \end{matrix}$ Therefore, for $m_{2} > 2$ very small, we obtain $\begin{array}{l} ⟨ \partial J (u), {\tilde{W}}_{1} (u) + m_{2} Z_{1} (u) + m_{1} \sum_{i \in {\tilde{I}}_{1} ∖ I_{1}} X_{i} (u) ⟩ ⩽ & - c (\sum_{i \in {\tilde{I}}_{1}} (\frac{1}{λ_{i}^{β}} + \frac{| \nabla K (a_{i}) |}{λ_{i}}) + \sum_{i, j \in {\tilde{I}}_{1}, j \neq i} ε_{i j}) \\ (4.16) & + o (\sum_{j = 1}^{p} \frac{1}{λ_{j}^{β}}) . \end{array}$ Now let R be the set of the remainder indices. So $R = {i \in I_{2}, λ_{i}^{β} < \frac{1}{10} λ_{k_{0}}^{β}}$ and denote $\tilde{u} = \sum_{i \in R} α_{i} P δ_{a_{i}, λ_{i}}$ . Observe that $\tilde{u} \in {\tilde{V}}_{n - 2} (♯ R, ε)$ , therefore, we can apply the associated vector field $W_{2} (\tilde{u})$ defined in Lemma 4.2. For ${\tilde{W}}_{2} (u) = W_{2} (\tilde{u})$ we get by Lemma 4.2 $\begin{array}{l} ⟨ \partial J (u), {\tilde{W}}_{2} (u) ⟩ ⩽ & - c (\sum_{i \in R} (\frac{1}{λ_{i}^{β}} + \frac{| \nabla K (a_{i}) |}{λ_{i}}) + \sum_{i, j \in R, j \neq i} ε_{i j}) \\ + O (\sum_{i \in R, j \notin R} ε_{i j}) . \end{array}$ We let in this case $W = {\tilde{W}}_{1} + {\tilde{W}}_{2} + m_{2} Z_{1} + m_{1} \sum_{i \in {\tilde{I}}_{1} ∖ I_{1}} X_{i} (u)$ . It satisfies $\begin{matrix} (4.17) & ⟨ \partial J (u), W (u) ⟩ ⩽ - c (\sum_{i = 1}^{p} (\frac{1}{λ_{i}^{β}} + \frac{| \nabla K (a_{i}) |}{λ_{i}}) + \sum_{j \neq i} ε_{i j}) . \end{matrix}$

∙ Second case: $u_{2} \notin {u = \sum_{j = 1}^{♯ I_{2}} α_{j} P δ_{a_{j}, λ_{j}} \in {\tilde{V}}_{n - 4} (♯ I_{2}, ε) with y_{ℓ_{j}} \in K_{n - 4}^{+}, y_{ℓ_{j}} \neq y_{ℓ_{k}}, \forall 1 ⩽ j \neq k ⩽ ♯ I_{2} and ρ (y_{ℓ_{1}}, \dots, y_{ℓ_{p}}) > 0}$ .

Let ${\tilde{W}}_{2} (u) = W_{2} (u_{2})$ where $W_{2}$ is defined in Lemma 4.2. We then have: $\begin{array}{l} ⟨ \partial J (u), {\tilde{W}}_{2} (u) ⟩ ⩽ & - c (\sum_{i \in I_{2}} (\frac{1}{λ_{i}^{β}} + \frac{| \nabla K (a_{i}) |}{λ_{i}}) + \sum_{i, j \in I_{2}, j \neq i} ε_{i j}) \\ + O (\sum_{i \in I_{2}, j \notin I_{2}} ε_{i j}) . \end{array}$ As in the first case, we denote by $k_{0}$ the index of $I_{2}$ satisfying $\begin{matrix} λ_{k_{0}} = min {λ_{i}, i \in I_{2}} \end{matrix}$ and we define $\begin{matrix} {\tilde{I}}_{2} = {i, 1 ⩽ i ⩽ p, s.t. λ_{i}^{β (y_{ℓ_{i}})} ⩾ \frac{1}{10} λ_{k_{0}}^{n - 4}} \end{matrix}$ and $E = {\tilde{I}}_{2}^{c}$ . Let $\begin{array}{l} Z_{2} (u) = \sum_{i \in {\tilde{I}}_{2} ∖ I_{2}} - 2^{i} λ_{i} \frac{\partial P δ_{a_{i}, λ_{i}}}{\partial λ_{i}}, \\ {\tilde{W}}_{1} (u) = W_{1} (\sum_{i \in E} α_{i} P δ_{a_{i}, λ_{i}}), \end{array}$ and $X_{i} (u)$ the vector field defined in (4.13). By the same computation of the first section, we get for $W = {\tilde{W}}_{2} + {\tilde{W}}_{1} + m_{2} Z_{2} + m_{1} \sum_{i \in {\tilde{I}}_{2} ∖ I_{2}} X_{i} (u)$ , $\begin{matrix} ⟨ \partial J (u), W (u) ⟩ ⩽ - c (\sum_{i = 1}^{p} (\frac{1}{λ_{i}^{β}} + \frac{| \nabla K (a_{i}) |}{λ_{i}}) + \sum_{j \neq i} ε_{i j}) . \end{matrix}$

∙ Third case: $u_{1} \in {u = \sum_{j = 1}^{♯ I_{1}} α_{j} P δ_{a_{j}, λ_{j}} \in {\tilde{V}}_{< n - 4} (♯ I_{1}, ε) with y_{ℓ_{j}} \in K_{< n - 4}^{+} and y_{ℓ_{j}} \neq y_{ℓ_{k}}, \forall 1 ⩽ j \neq k ⩽ ♯ I_{2}}$ and $u_{2} \in {u = \sum_{j = 1}^{♯ I_{2}} α_{j} P δ_{a_{j}, λ_{j}} \in {\tilde{V}}_{n - 4} (♯ I_{2}, ε) with y_{ℓ_{j}} \in K_{n - 4}^{+}, y_{ℓ_{j}} \neq y_{ℓ_{k}}, \forall 1 ⩽ j \neq k ⩽ ♯ I_{2} and ρ (y_{ℓ_{1}}, \dots, y_{ℓ_{p}}) > 0}$ .

Let ${\tilde{W}}_{1} (u) = W_{1} (u_{1})$ and ${\tilde{W}}_{2} (u) = W_{2} (u_{2})$ where $W_{1}$ and $W_{2}$ are the vector fields defined in Lemmas 4.1 and 4.2 respectively. Using the above estimates, we get for $W = {\tilde{W}}_{1} + {\tilde{W}}_{2}$ $\begin{matrix} ⟨ \partial J (u), W (u) ⟩ ⩽ - c (\sum_{i = 1}^{p} (\frac{1}{λ_{i}^{β}} + \frac{| \nabla K (a_{i}) |}{λ_{i}}) + \sum_{j \neq i} ε_{i j}) . \end{matrix}$ This finishes the proof of Lemma 4.3 and then the proof of Theorem 1.3 follows. □
Proof of Lemma 4.1.
We divide ${\tilde{V}}_{1} (p, ε)$ as follows. Let $δ > 0$ and small. $\begin{array}{l} W_{1} (p, ε) : = {u = \sum_{j = 1}^{p} α_{j} P δ_{a_{j}, λ_{j}} \in {\tilde{V}}_{1} (p, ε), y_{l_{j}} \neq y_{l_{k}}, \forall j \neq k, - \sum_{k = 1}^{n} b_{k} (y_{l_{j}}) > 0, \\ and λ_{j} | a_{j} - y_{l_{j}} | < δ, \forall j = 1, \dots, p} . \\ W_{2} (p, ε) : = {u = \sum_{j = 1}^{p} α_{j} P δ_{a_{j}, λ_{j})} \in {\tilde{V}}_{1} (p, ε), y_{l_{j}} \neq y_{l_{k}}, \forall j \neq k, λ_{j} | a_{j} - y_{l_{j}} | < δ, \\ \forall j = 1, \dots, p and there exist at least j_{1} such that - \sum_{k = 1}^{n} b_{k} (y_{l_{j_{1}}}) < 0} . \\ W_{3} (p, ε) : = {u = \sum_{j = 1}^{p} α_{j} P δ_{a_{j}, λ_{j}} \in {\tilde{V}}_{1} (p, ε), y_{l_{j}} \neq y_{l_{k}}, \forall j \neq k, and there exist at least \\ j_{1}, s, t, λ_{j_{1}} | a_{j_{1}} - y_{l_{j_{1}}} | ⩾ \frac{δ}{2}} . \\ W_{4} (p, ε) : = {u = \sum_{j = 1}^{p} α_{j} P δ_{a_{j}, λ_{j}} \in {\tilde{V}}_{1} (p, ε), such that there exist j \neq k with y_{l_{j}} = y_{l_{k}}} . \end{array}$ ∙ Pseudo-gradient in $W_{1} (p, ε)$ . In this region, we have $| a_{j} - y_{l_{j}} |^{β} = o (\frac{1}{λ_{j}^{β}}), \forall j = 1, \dots, p$ and $\frac{1}{{(λ_{j} λ_{k})}^{\frac{n - 2}{2}}} = o (\frac{1}{λ_{j}^{β}}) + o (\frac{1}{λ_{k}^{β}})$ since $β < n - 2$ . Thus, for $V_{1} (u) = \sum_{j = 1}^{p} λ_{j} \frac{\partial P δ_{a_{j}, λ_{j}}}{\partial λ_{j}}$ , we have: $\begin{matrix} ⟨ \partial J (u), V_{1} (u) ⟩ ⩽ - c (\sum_{i = 1}^{p} (\frac{1}{λ_{i}^{β}} + \frac{| \nabla K (a_{i}) |}{λ_{i}}) + \sum_{j \neq i} ε_{i j}), \end{matrix}$ since $\frac{| \nabla K (a_{i}) |}{λ_{i}} = o (\frac{1}{λ_{j}^{β}})$ .

∙ Pseudo-gradient in $W_{2} (p, ε)$ . Let $k_{0}$ an index such that $λ_{k_{0}}^{β} = min {λ_{j}^{β}, such that - \sum_{k = 1}^{n} b_{k} (y_{l_{j}}) < 0}$ . We denote by $I_{1} = {j = 1, \dots, p, such that, λ_{j}^{β} ⩽ \frac{1}{2} λ_{k_{0}}^{β}}$ . Observe that $\tilde{u} : = \sum_{j \in I_{1}} α_{j} P δ_{a_{j}, λ_{j}} \in W_{1} (♯ I_{1}, ε)$ . Using the same previous technics, we get for ${\tilde{V}}_{1} (u) = V_{1} (\tilde{u})$ , $\begin{matrix} ⟨ \partial J (u), {\tilde{V}}_{1} (u) - \sum_{j \notin I_{1}} λ_{j} \frac{\partial P δ_{a_{j}, λ_{j}}}{\partial λ_{j}} ⟩ ⩽ - c (\sum_{i = 1}^{p} (\frac{1}{λ_{i}^{β}} + \frac{| \nabla K (a_{i}) |}{λ_{i}}) + \sum_{j \neq i} ε_{i j}) . \end{matrix}$

∙ Pseudo-gradient in $W_{3} (p, ε)$ . Let $k_{1}$ an index such that $λ_{k_{1}}^{β} = min {λ_{j}^{β}, such that λ_{j} | a_{j} - y_{l_{j}} | ⩾ \frac{δ}{2}}$ and let $J_{1} = {j = 1, \dots, p such that λ_{j}^{β} ⩽ \frac{1}{2} λ_{k_{1}}^{β}}$ . For any $j \in J_{1}$ , we have $λ_{j} | a_{j} - y_{l_{j}} | ⩽ \frac{δ}{2}$ . Let $\tilde{u} : = \sum_{j \in J_{1}} α_{j} P δ_{a_{j}, λ_{j}}$ , $\tilde{u} \in W_{1} (♯ J_{1}, ε) \cup W_{2} (♯ J_{1}, ε)$ . We denote by ${\tilde{W}}_{3} (u) = V (\tilde{u})$ where V is the associated vector field to the above two regions. Using the second expansion of Lemma 3.2 and the previous technics, we get $\begin{matrix} ⟨ \partial J (u), {\tilde{W}}_{3} (u) + \sum_{i \notin J_{1}} (X_{i} (u) - λ_{i} \frac{\partial P δ_{a_{i}, λ_{i}}}{\partial λ_{i}}) ⟩ ⩽ - c (\sum_{i = 1}^{p} (\frac{1}{λ_{i}^{β}} + \frac{| \nabla K (a_{i}) |}{λ_{i}}) + \sum_{j \neq i} ε_{i j}) . \end{matrix}$ Here $X_{i} (u) = α_{i} \sum_{k = 1}^{n} \frac{b_{k}}{λ_{i}} \int_{R^{n}} \frac{| x_{k} + λ_{i} {(a_{i} - y_{ℓ_{i}})}_{k} |^{β}}{(1 + λ_{i} | {(a_{i} - y_{ℓ_{i}})}_{k} |^{β - 1}} \frac{x_{k}}{{(1 + | x |^{2})}^{n + 1}} d x$ if $λ_{i} | {(a_{i} - y_{ℓ_{i}})}_{k} | \in [\frac{δ}{2}, \frac{2}{δ}]$ and $X_{i} (u)$ is defined by (4.13) otherwise.

∙ Pseudo-gradient in $W_{4} (p, ε)$ . For any critical point $y_{ℓ}$ of K, we denote $B_{ρ} = {j = 1, \dots, p such that a_{j} \in B (y_{ℓ}, ρ)}$ . In this region, there exists at least ℓ such that $♯ B_{ℓ} ⩾ 2$ . Let $\begin{matrix} J_{1} = {ℓ, 1 ⩽ ℓ ⩽ ♯ K, such that ♯ B_{ℓ} ⩾ 2} . \end{matrix}$ For any $ℓ \in J_{1}$ , we decrease all $λ_{j}$ ’s, $j \in B_{ℓ}$ as follows. Let $\begin{array}{l} ψ : R ⟶ R \\ t ⟼ \{\begin{matrix} 1 & if | t | ⩾ 1 \\ 0 & if | t | ⩽ γ, \end{matrix} \end{array}$ where $γ > 0$ very small. Define $\overline{ψ} (λ_{j}) = \sum_{i \neq j} ψ (\frac{λ_{j}}{λ_{i}})$ for $j \in B_{k}$ and $k \in J_{1}$ . By the first expansion of Lemma 3.1, we get $\begin{matrix} ⟨ \partial J (u), - \sum_{ℓ \in J_{1}} \sum_{j \in B_{ℓ}} \overline{ψ} (λ_{j}) λ_{j} \frac{\partial P δ_{a_{j}, λ_{j}}}{\partial λ_{j}} ⟩ ⩽ - c \sum_{ℓ \in J_{1}} \sum_{i \neq j \in B_{ℓ}} ε_{i j} + + O (\sum_{i = 1}^{p} \frac{1}{λ_{i}^{β}}) . \end{matrix}$ To obtain the required upper bound, we set $\begin{matrix} I_{1} : = {j; λ_{j} | a_{i} - y_{ℓ_{j}} | ⩾ δ} . \end{matrix}$ If $I_{1} \neq \emptyset$ , we use the above vector field (defined in $W_{3} (p, ε)$ ) and using the expansions of Lemma 3.2, we obtain $\begin{matrix} ⟨ \partial J (u), - \sum_{ℓ \in J_{1}} \sum_{j \in B_{ℓ}} \overline{ψ} (λ_{j}) λ_{j} \frac{\partial P δ_{a_{j}, λ_{j}}}{\partial λ_{j}} + \sum_{i \in I_{1}} X_{i} (u) ⟩ ⩽ - c (\sum_{i = 1}^{p} (\frac{1}{λ_{i}^{β}} + \frac{| \nabla K (a_{i}) |}{λ_{i}}) + \sum_{j \neq i} ε_{i j}) . \end{matrix}$ If $I_{1} = \emptyset$ , we denote by $I_{2}$ the set of indices constructed by 1 and all j such that $λ_{j} \sim λ_{1}$ (of the same order). We write $u = \sum_{i \in I_{2}} α_{i} P δ_{a_{i}, λ_{i}} + \sum_{i \notin I_{2}} α_{i} P δ_{a_{i}, λ_{i}}$ . Observe that $\tilde{u} : = \sum_{i \in I_{2}} α_{i} P δ_{a_{i}, λ_{i}} \in W_{k} (♯ I_{2}, ε), k = 1, 2$ or 3. We then apply the associated vector field denoted ${\tilde{W}}_{4} (u)$ . We obtain $\begin{array}{l} ⟨ \partial J (u), {\tilde{W}}_{4} (u) ⟩ ⩽ & - c (\sum_{i \in I_{2}} (\frac{1}{λ_{i}^{β}} + \frac{| \nabla K (a_{i}) |}{λ_{i}}) + \sum_{i, j \in I_{2}, j \neq i} ε_{i j}) \\ + O (\sum_{i \in I_{2}, j \notin I_{2}} ε_{i j}) \end{array}$ and therefore, $\begin{matrix} ⟨ \partial J (u), {\tilde{W}}_{4} (u) - \sum_{ℓ \in J_{1}} \sum_{j \in B_{ℓ}} \overline{ψ} (λ_{j}) λ_{j} \frac{\partial P δ_{a_{j}, λ_{j}}}{\partial λ_{j}} ⟩ ⩽ - c (\sum_{i = 1}^{p} (\frac{1}{λ_{i}^{β}} + \frac{| \nabla K (a_{i}) |}{λ_{i}}) + \sum_{j \neq i} ε_{i j}) . \end{matrix}$ This finishes the proof of Lemma 4.1. □
Proof of Lemma 4.2.
The situation here is exactly the one of ([16], Proposition 3.7), so we omit the proof here. □

5. Proof of the existence results

5.1. Proof of Theorem 1.2

Using the result of Theorem 1.1, the critical points at infinity of the associated variational problem are in one to one correspondence with the elements ${(y_{1}, \dots, y_{p})}_{\infty}$ , ${(y_{1}, \dots, y_{p})}_{\infty} \in C^{\infty}$ . For each ${(y_{1}, \dots, y_{p})}_{\infty} \in C^{\infty}$ , we denote by $W_{u}^{\infty} {(y_{1}, \dots, y_{p})}_{\infty}$ ; the unstable manifold of the critical points at infinity ${(y_{1}, \dots, y_{p})}_{\infty}$ . Recall that $i {(y_{1}, \dots, y_{p})}_{\infty}$ the index of ${(y_{1}, \dots, y_{p})}_{\infty}$ is equal to the dimension of $W_{u}^{\infty} {(y_{1}, \dots, y_{p})}_{\infty}$ . Using now the gradient flow of $(- \partial J)$ to deform $Σ^{+}$ . It follows then by deformation lemma (see [6]), that $\begin{matrix} (5.1) & Σ^{+} ≃ ⋃_{{(y_{1}, \dots, y_{p})}_{\infty} \in C^{\infty}} W_{u}^{\infty} {(y_{1}, \dots, y_{p})}_{\infty} \cup ⋃_{w, \partial J (w) = 0} W_{u} (w), \end{matrix}$ where ≃ denotes retracts by deformation. It follows from the above deformation retract that the problem (1.1) has necessary a solution w. Otherwise, it follows from the above equation that $\begin{matrix} 1 = χ (Σ^{+}) = \sum_{{(y_{1}, \dots, y_{p})}_{\infty} \in C^{\infty}} {(- 1)}^{i {(y_{1}, \dots, y_{p})}_{\infty}}, \end{matrix}$ where χ denotes the Euler–Poincaré Characteristic. Such an equality contradicts the assumption of Theorem 1.2.

5.2. Proof of Theorem 1.4

Let $\begin{matrix} J_{1} (u) = \frac{1}{{(\int_{Ω} u^{\frac{2 n}{n - 2}} d x)}^{\frac{n}{n - 2}}}, u \in Σ \end{matrix}$ be the Euler Lagrange functional associated to Yamabe problem on Ω. Let $\begin{matrix} S = \frac{1}{{(\int_{R^{n}} δ_{a, λ}^{\frac{2 n}{n - 2}} d x)}^{\frac{n}{n - 2}}} \end{matrix}$ be the best Sobolev constant. S does not depend on a and λ. It is known that $\begin{matrix} S = inf_{u \in Σ} J_{1} (u) \end{matrix}$ and that the infimum is not achieved, see [11].

For $c \in R$ and for any function f on Σ, we define $\begin{matrix} f^{c} = {u \in Σ, s.t., f (u) ⩽ c} . \end{matrix}$ It is easy to see that if $| K - 1 |_{L^{\infty} (Ω)} ⩽ ε_{0}$ for $ε_{0}$ small enough, we have $\begin{matrix} (5.2) & J^{S + \frac{S}{4}} \subset J_{1}^{S + \frac{S}{2}} \subset J^{S + \frac{3 S}{4}} . \end{matrix}$ This is due to the fact that $J (u) = J_{1} (u) (1 + O (| ε_{0} |))$ .

Now let ${(y_{1}, \dots, y_{q})}_{\infty}$ be a critical point at infinity of q masses. It is known that the level of J at ${(y_{1}, \dots, y_{q})}_{\infty}$ is given by $S {(\sum_{k = 1}^{q} \frac{1}{K {(y_{k})}^{(n - 2) / 2}})}^{2 / n}$ , see [10]. Hence goes to $q S$ when $ε_{0}$ is close to zero. Therefore, for $| ε_{0} |$ small enough, we have: $\begin{array}{l} All critical points at infinity of J of q -masses, q ⩾ 2 are above S + \frac{3}{4} S, (*) \end{array}$ and $\begin{array}{l} all critical points at infinity of J of one masse are below S + \frac{S}{4} . (* *) \end{array}$

Therefore, $\begin{matrix} (5.3) & the functional J has no critical points at infinity in J_{({\tilde{S}}_{n} + η)}^{({\tilde{S}}_{n} + 3 η)} . \end{matrix}$ To prove the existence result, we argue by contradiction and we assume that J has no critical points. It follows from (5.3) that $\begin{matrix} J^{{\tilde{S}}_{n} + 3 η} ≃ J^{{\tilde{S}}_{n} + η}, \end{matrix}$ where ≃ denotes retracts by deformation. Thus by (5.2), we derive that $\begin{matrix} (5.4) & J_{1}^{{\tilde{S}}_{n} + 2 η} ≃ J^{{\tilde{S}}_{n} + η} . \end{matrix}$ Now we use the gradient flow of $(- \partial J)$ to deform $J^{{\tilde{S}}_{n} + η}$ . As mentioned above, the only critical points at infinity of J under the level ${\tilde{S}}_{n} + η$ are ${(y)}_{\infty}$ , $y \in K_{< n - 4}^{+} \cup K_{⩾ n - 4}$ . Thus $\begin{matrix} (5.5) & J^{{\tilde{S}}_{n} + η} ≃ ⋃_{y \in K_{< n - 4}^{+} \cup K_{⩾ n - 4}} W_{u}^{\infty} (y) . \end{matrix}$ We apply now the Euler–Poincaré characteristic of both sides of (5.5), we get $\begin{matrix} χ (J^{{\tilde{S}}_{n} + η}) = \sum_{y \in K_{< n - 4}^{+} \cup K_{⩾ n - 4}} {(- 1)}^{n - \tilde{i} (y)} . \end{matrix}$ Thus by (5.4), we obtain $\begin{matrix} (5.6) & χ (J_{1}^{{\tilde{S}}_{n} + 2 η}) = \sum_{y \in K_{< n - 4}^{+} \cup K_{⩾ n - 4}} {(- 1)}^{n - \tilde{i} (y)} . \end{matrix}$ It is known that $J_{1}^{{\tilde{S}}_{n} + 2 η}$ and Ω has the same homotopy type. See ([5] remark 5). Indeed, for η small, $J_{1}^{{\tilde{S}}_{n} + 2 η} \subset V (1, ε)$ (see [20] page 1549). Therefore, there is no critical points of $J_{1}$ in these sets. From (5.6) we get $\begin{matrix} (5.7) & χ (Ω) = \sum_{y \in K_{< n - 4}^{+} \cup K_{⩾ n - 4}} {(- 1)}^{n - \tilde{i} (y)} . \end{matrix}$ Such equality contradicts the assumption of Theorem 1.4. This complete the proof of Theorem 1.4.

References

Abdelhedi and

Chtioui, On the prescribed Paneitz curvature problem on the standard spheres, Advanced Nonlinear Studies 6 (2006), 511–528. doi:10.1515/ans-2006-0402.

Abdelhedi,

Chtioui and

Hajaiej, A complete study of the lack of compactness and existence results of a fractional Nirenberg equation via a flatness hypothesis: Part I, Analysis and PDE 9(6) (2016), 1285–1315. doi:10.2140/apde.2016.9.1285.

Al Ghamdi,

Chtioui and

Rigane, Existence of conformal metrics with prescribed Q-curvature, Abstract and Applied Analysis 2013 (2013), 568245.

Bahri, Critical Point at Infinity in Some Variational Problems, Pitman Res. Notes Math, Ser, Vol. 182, Longman Sci. Tech, Harlow, 1989.

Bahri and

J.M.

Coron, On a nonlinear elliptic equation involving the critical Sobolev exponent: The effect of topology of the domain, Comm. Pure Appli. Math. 41 (1988), 255–294.

Bahri and

Rabinowitz, Periodic orbits of Hamiltonian systems of three body type, Ann. Inst. H. Poincaré Anal. Non Linéaire 8 (1991), 561–649. doi:10.1016/S0294-1449(16)30252-9.

Ben Ayed,

Chen,

Chtioui and

Hammami, On the prescribed scalar curvature problem on 4-manifolds, Duke Math. J. 84 (1996), 633–677. doi:10.1215/S0012-7094-96-08420-3.

Ben Ayed,

El Mehdi and

Hammami, Some existence results for a Paneitz type problem via the theory of critical points at infinity, Journal de Mathématiques Pures et Appliquées 84 (2005), 247–278. doi:10.1016/j.matpur.2004.09.012.

Bensouf and

Chtioui, Conformal metrics with prescribed Q-curvature on

S^{n}

, Calculus of Variations and PDE 41 (2011), 455–481. doi:10.1007/s00526-010-0372-9.

10.

Bouchech and

Chtioui, Multiplicity and existence results for a nonlinear elliptic equation with Sobolev exponent, Advanced Nonlinear Studies 10 (2010), 537–572.

11.

Brezis and

J.M.

Coron, Convergence of solutions of H-systems or how to blow bubbles, Arch. Rational Mech. Anal. 89 (1985), 21–56. doi:10.1007/BF00281744.

12.

S.A.

Chang,

M.J.

Gursky and

P.C.

Yang, Regularity of a fourth order non linear PDE with critical exponent, Amer. J. Math. 121 (1999), 215–257. doi:10.1353/ajm.1999.0011.

13.

S.A.

Chang and

P.C.

Yang, On a fourth order curvature invariant, Spectral problems in Geometry and Arithmetic, Contemporary Math. 237 (1999), 9–28. doi:10.1090/conm/237/1710786.

14.

Chtioui and

Bouchech, On a bi-harmonic equation involving critical exponent: Existence and multiplicity results, Acta Mathematica Scientia 31(4) (2011), 1213–1244. doi:10.1016/S0252-9602(11)60311-1.

15.

Chtioui and

El Mehdi, On a Paneitz type equation in six dimensional domains, Differential and Integral Equations 17 (2004), 681–696.

16.

Chtioui and

Rigane, On the prescribed Q-curvature problem on

S^{n}

, Journal of Functional Analysis 261 (2011), 2999–3043. doi:10.1016/j.jfa.2011.07.017.

17.

Djadli,

Hebey and

Ledoux, Paneitz-type operators and applications, Duke Math. J. 104 (2000), 129–169. doi:10.1215/S0012-7094-00-10416-4.

18.

Djadli,

Malchiodi and

Ould Ahmedou, Prescribing a fourth order conformal invariant on the standard sphere, part I: A perturbation result, Commun. Contemp. Math. 4 (2002), 375–408. doi:10.1142/S0219199702000695.

19.

Djadli,

Malchiodi and

Ould Ahmedou, Prescribing a fourth order conformal invariant on the standard sphere, part II: Blow up analysis and applications, Annali della Scuola Normale Sup. di Pisa. 5 (2002), 387–434.

20.

Ebobisse and

M.O.

Ahmedou, On a nonlinear fourth-order elliptic equation involving the critical Sobolev exponent, Nonlinear Anal. TMA 52 (2003), 1535–1552. doi:10.1016/S0362-546X(02)00273-0.

21.

C.S.

Lin, A classification of solution of a conformally invariant fourth order equation in

R^{n}

, Commentari Mathematici Helvetici 73 (1998), 206–231. doi:10.1007/s000140050052.

22.

Struwe, A global compactness result for elliptic boundary value problem involving limiting nonlinearities, Math. Z. 187 (1984), 511–517. doi:10.1007/BF01174186.

23.

R.C.A.M.

Van Der Vorst, Variational identities and applications to differential systems, Arch. Rational Mech. Anal. 116 (1991), 375–398.