Mass concentration and local uniqueness of ground states for \(L^2\)-subcritical nonlinear Schrödinger equations (Q667878)

The paper is devoted to studying the following time-independent nonlinear Schrödinger equation \[ -\Delta u+V(x)u=\mu u+\rho^{p-1}u^p,\qquad u\in \mathcal{H}(\mathbb{R}^d), \tag{1} \] where $d\geqslant1$, $\mu\in\mathbb{R}$, $p\in\big(1, 1+\frac{4}{d}\big)$, $\rho>0$, and the space $\mathcal{H}(\mathbb{R}^d)$ is defined as \begin{align*} & \mathcal{H}(\mathbb{R}^d) := \Big \{u(x)\in H^1(\mathbb{R}^d)\,\ \Big|\,\ \int_{\mathbb{R}^d} V(x)|u(x)|^2 \,dx<\infty \Big\}, \\ &\|u\|_{\mathcal{H}}^2= \int_{\mathbb{R}^d}\big(|\nabla u(x)|^2+[1+V(x)]|u(x)|^2\big)\,dx. \end{align*} The trapping potential $V(x)\geqslant0$ is assumed to satisfy \[ V(x)\in L_{loc}^\infty(\mathbb{R}^d)\cap C^\alpha(\mathbb{R}^d),\quad \alpha\in(0,1), \quad \inf_{x\in\mathbb{R}^d}V(x)=0,\quad \lim_{|x|\to\infty}V(x) = \infty. \] \par A minimizer of the minimization problem \[ e(\rho):=\inf\big\{E_{\rho}(u): u\in \mathcal{H}(\mathbb{R}^d),\|u\|_2^2=1\big\}, \] solves equation (1) for some suitable Lagrange multiplier $\mu$. Here $E_{\rho}(u)$ is the Gross-Pitaevskii (GP) energy functional defined by \[ E_{\rho}(u):=\frac{1}{2}\int_{\mathbb{R}^d}|\nabla u|^2\,dx +\frac{1}{2}\int_{\mathbb{R}^d}V(x)|u|^2\,dx -\frac{\rho^{p-1}}{p+1}\int_{\mathbb{R}^d}|u|^{p+1}\,dx. \] Consider also the following classical Gagliardo-Nirenberg type inequality \begin{align*} & C_{GN} \leqslant \frac{\|\nabla u\|_2^{\frac{d}{2}(p-1)}\|u\|_2^{p+1-\frac{d}{2}(p-1)}} {\|u\|_{p+1}^{p+1}} \qquad\text{for any}\quad u\in H^1(\mathbb{R}^d)\setminus\{0\}, \\ & C_{GN} :=\|w\|_2^{p-1}\Big(1-\frac{p-1}{p+1}\frac{d}{2}\Big) \Big[\frac{d(p-1)}{2(p+1)-d(p-1)}\Big]^{\frac{d}{4}(p-1)}, \end{align*} and $w$ is the unique positive solution of \[ \Delta w-w+w^p=0,\,\,\ w \in H^1(\mathbb{R}^d). \tag{2} \] Given $\rho\in(0,\infty)$, define the set of nontrivial weak solutions for (1): \begin{align*} &S_{\mu,\rho}:=\Big\{ u\in \mathcal{H}\setminus \{(0)\}: \ \langle F_{\mu,\rho}'(u),\varphi\rangle=0,\, \forall \ \varphi\in \mathcal{H} \Big\}, \\ & F_{\mu,\rho}(u) :=\frac{1}{2}\int_{\mathbb{R}^d}|\nabla u|^2\,dx +\frac{1}{2}\int_{\mathbb{R}^d}\big(V(x)-\mu\big)|u|^2\,dx -\frac{\rho^{p-1}}{p+1}\int_{\mathbb{R}^d}|u|^{p+1}\,dx. \end{align*} The set of ground states for (1) and the set of minimizers for $e(\rho)$ are \begin{align*} &G_{\mu,\rho}:=\Big\{ u\in S_{\mu,\rho}: F_{\mu,\rho}(u)\leqslant F_{\mu,\rho}(v)\, \text{ for all }\, v \in S_{\mu,\rho} \Big\}, \\ &M_\rho:=\Big\{ u_\rho\in\mathcal{H}(\mathbb{R}^d): \text{$u_\rho$ is a minimizer of $e(\rho)$} \Big\}. \end{align*} The first result is as follows. \par Theorem 1. \begin{itemize} \item [(i)] For a.e. $\rho\in(0,\infty)$, all minimizers of $e(\rho)$ satisfy equation (1) with a fixed Lagrange multiplier $\mu=\mu_\rho$. \item [(ii)] For $a.e.$ $\rho\in(0,\infty)$, $G_{\mu_\rho,\rho}=M_\rho$. \end{itemize} \par Define \[ Z:=\big\{x\in\mathbb{R}^d:V(x)=0\big\}=\big\{x_1, x_2, \dots, x_m\big\}, \qquad \ m\geqslant1. \] Assume that there exists some $V_i(x)\in C^2_{\text{loc}}(\mathbb{R}^d)$ such that $V_i(tx)=t^{r_i}V_i(x)$ in $\mathbb R^d$ for any $t>0$, $\lim_{|x|\to\infty} V_i(x) = +\infty$, and \[ \lim_{x\to0}\frac{V(x+x_i)}{ V_i(x)}=1,\,\,\,i=1,2,\dots,m. \tag{3} \] Define \begin{align*} &Q_i(y):=\int_{\mathbb{R}^d}V_i(x+y)w^2\,dx,\,\,\,i=1,2,\dots,m, \quad r:=\max_{1\leqslant i\leqslant m}r_i, \\ &\bar{Z}:=\big\{x_i\in Z: r_i= r\big\} \subset Z, \quad \bar{\lambda}_0:=\min_{ i\in \Gamma}\bar{\lambda}_i, \bar{\lambda}_i:=\min_{y\in \mathbb{R}^d}Q_i(y),\quad\Gamma:=\big\{i: x_i\in \bar Z\big\}, \\ &\lambda:=\frac{1}{2}\frac{4-d(p-1)}{2(p+1)-d(p-1)}, \quad Q(y):=\int_{\mathbb{R}^d}V_0(x+y)w^2\,dx,\quad V_0(x):=V_i(x), \quad \bar{\lambda}_i=\bar{\lambda}_0, \\ &Z_0:=\big\{x_i\in \bar{Z}: \bar{\lambda}_i= \bar{\lambda}_0\big\},\,\,\,K_0:=\big\{y: Q(y)=\bar{\lambda}_i=\bar{\lambda}_0\}. \end{align*} The second result is the following theorem. \par Theorem 2. Suppose $ V(x)\in C^2(\mathbb{R}^d)$ satisfies (3), and there exists a constant $\kappa>0$ such that \[ V(x)\leqslant C e^{\kappa |x|} \text{ if }\, |x| \text{ is large}. \] Set $a^*:=\|w\|_2^2$, where $w$ is the unique positive solution of (2). Let $u_k$ be a nonnegative minimizer of $e(\rho_k)$, where $\rho_k\to \infty$ as $k\to\infty$. Then there exists a subsequence, still denoted by $\{u_k\}$, such that $u_k$ satisfies \[ \bar{u}_k(x):=\sqrt{a^*}\varepsilon_k^\frac{d}{2}u_k(\varepsilon_kx+x_k) \to w(x)\text{ uniformly in }\mathbb{R}^d\text{ as }k\to\infty, \] where \[ \varepsilon_k:=\Big(\frac{\rho_k}{\sqrt{a^*}}\Big)^{-\frac{2(p-1)}{4-d(p-1)}}, \] and $x_{k}$ is the unique local maximum point of $u_k$ satisfying \[ \frac{x_k-x_0}{\varepsilon_k}\to y_0\text{ for some }x_0\in Z_0\text{ and }y_0\in K_0\text{ as }k\to\infty. \] Furthermore, $u_k$ decays exponentially in the sense that \[ \bar{u}_k(x)\leqslant Ce^{-\frac{|x|}{2}} \text{ and } |\nabla \bar{u}_k|\leqslant Ce^{-\frac{|x|}{4}} \text{ as } |x|\to\infty, \] where $C>0$ is a constant independent of $k$. \par Suppose that $Z_0$ contains only one element $x_0$ and satisfies (3) with $x_i=x_0$, $r_i\geqslant 2$. Assume also that there exists a constant $R_0$ small enough such that \[ \frac{\partial V(x+x_0)}{\partial x_i}=\frac{\partial V_0(x)}{\partial x_i}+W_i(x)\, \text{ and }\,\ |W_i(x)|\leqslant C|x|^{s_i}\, \text{ in }\, B_{R_0}(0),\,\ \] where $s_i>r-1$ for $i=1,2,\dots,m$. The third result is as follows. \par Theorem 3. Suppose $V(x)\in C^2(\mathbb{R}^d)$ satisfies all above assumptions and $y_0$ is the unique and non-degenerate critical point of $Q(y)$. Then there exists a unique nonnegative minimizer for $e(\rho)$ as $\rho\to \infty$.