- Research
- Open access
- Published:
Artificial boundary condition for one-dimensional nonlinear Schrödinger problem with Dirac interaction: existence and uniqueness results
Boundary Value Problems volume 2018, Article number: 16 (2018)
Abstract
We consider the nonlinear Schrödinger equation with Dirac interaction in a half-line domain of \(\mathbb{R}\). Endowed with artificial boundary condition, we discuss the global well-posedness of the equation.
1 Introduction
We consider a nonlinear Schrödinger equation (NLS) with Dirac distribution defect [1–4]
where \(\boldsymbol {\Omega}\subset\mathbb{R}\), \(u=u(x,t)\) is the unknown solution maps \(\boldsymbol {\Omega}\times\mathbb{R}_{+}\) into \(\mathbb{C}\), \(\delta_{a}\) is the Dirac distribution at the point \(a\in \boldsymbol {\Omega }\), namely, \(\langle \delta_{a},v\rangle = v(a) \) for \(v \in \mathbf {H}^{1}(\boldsymbol {\Omega})\), and \(q \in\mathbb{R}\) represents its intensity parameter. Such distribution is introduced in order to model physically the defect at the point \(x=a\) (see [3–6]). The function g represents a generalization of the classical nonlinear Schrödinger equation (see for example [7–9]).
This specific model (1) is of recent research from a physical point of view in nonlinear optics plasma physics, water wave, quantum mechanics, hydrodynamics; see for example [10–15].
In nonlinear optics, equation (1) models a soliton propagating in a medium with a point defect [4, 16] or a wide soliton with a much narrower one in a bimodal fiber [17]. In the case when \(a=0\) and \(g(s)=s\), this model coincides with the Gross–Pitaevskii equation; see, for instance [1, 2, 4, 18, 19] and the references therein. See also [6, 20, 21] for some recent results dealing with NLS models.
The well-posedness of the solutions of the NLS equation (1) has been studied in the literature. In the case when \(q=0\) and \(\boldsymbol {\Omega}=\mathbb{R}\), the global existence in \(\mathbf {H}^{1}(\mathbb {R})\) and in \(\mathbf {L}^{2}(\mathbb{R})\) were proved in [22–24]. For bounded \(\boldsymbol {\Omega}\subset \mathbb{R}\) and with the standard boundary conditions (Dirichlet, Neumann and periodic), NLS equation (1) possesses a unique global solution in \(\mathbf {H}^{1}(\boldsymbol {\Omega})\), as was proved in [7]. In the last case \(q\neq0\) and \(g(s)=s\), in [4] the well-posedness in \(\mathbf {H}^{1}(\mathbb{R})\) of the solution of NLS equation (1) was proved.
The aim of this paper is to investigate the NLS equation (1) with Dirac interaction defect (\(q\neq0\)) in the case of a half-line in \(\mathbb{R}\). We consider for example \(\boldsymbol {\Omega}=\mathopen{]}-\infty,0[\) (the same calculations remain true for a positive half-line choice of Ω). The equation is endowed with a non-standard boundary condition at the point 0 in order to avoid the perturbations of the solutions caused by the boundary \(\{0\}\). The condition is actually necessary to achieve numerical solutions of the equation, as was demonstrated in [25–27]. The initial data is then supposed of a compact support in Ω. Our main results that have been proved entail that the NLS equation (1) has a unique solution in \(\mathbf {H}^{1}(\boldsymbol {\Omega})\). The demonstration is based on the Galerkin method. The continuous dependence of the solutions with regard to the initial data is also looked into.
The remainder of this paper is organized as follows. In Section 2, we provide some problem formulation and necessary technical results. In Section 3, we demonstrate the global well-posedness of the NLS equation in \(\mathbf {H}^{1}(\boldsymbol {\Omega})\). A few concluding remarks are given in Section 4.
2 Problem formulation and preliminaries
We discuss the nonlinear Schrödinger equation (NLS)
where \(\boldsymbol {\Omega}=\mathopen{]}-\infty,0[\), \(a<0\) and \(\delta_{a}\) is a Dirac interaction defect at point a. The smooth functional \(g\in \mathcal{C}^{1} ([0,+\infty[,\mathbb{R} )\) verifies the following conditions: There exist constants \(C\geq0\), \(\alpha_{1} > 0\), \(\alpha_{2} > 0\) and \(\theta\in[0,2[\) such that
We associate with equation (2) a non-standard boundary condition at the point \(x=0\),
where the operator \(\partial_{n}\) is the normal derivative, the phase function \(\mathbb{V}\) defined by \(\mathbb{V}(x,t)= {\int _{0}^{t}g ( \vert u(x,s) \vert ^{2} ) \,ds}\) and operator \(\partial_{t}^{1/2}\) represent a \(\frac{1}{2}\) order Riemann–Liouville fractional derivative defined by
This result (4) is obtained by [25–27] with an initial data that has compact support in Ω, it represents an artificial boundary condition on \(x=0\) to the NLS equation (2) with \(q=0\). This condition is added in order to avoid the perturbation effect on the solutions resulting from the reflection at the limit point \(\{0\}\).
We consider \(u_{0} \in \mathbf {H}^{1}(\boldsymbol {\Omega})\) to be an initial data, such that its support is compact in Ω (see [25–27])
Next, using the method applied in [28], we obtain our first technical result.
Lemma 1
Consider the initial value problem in \(\mathbb{C}^{m}\)
where M is a square matrix of order m, P represents a polynomial function of \(\mathbb{C}^{m}\), \(H_{0} \in\mathbb{C}^{m}\) is a constant vector. Then the problem (7) has a unique local solution \(H\in L^{\infty}(0,T;\mathbb{C}^{m})\).
Proof
We integrate (7) between 0 and t, to get
where \(I_{t}^{1/2}\) represents the Riemann–Liouville fractional integral operator of \(\frac{1}{2}\) order defined by
Since \(H\in L^{\infty}(0,T;\mathbb{C}^{m})\) we have \([I^{1/2}_{t} H(t)]_{t=0}=0\). Then equation (8) becomes
We show that there exists a unique function H verifying equation (9) by applying the Banach fixed-point theorem. Let \(T>0\), we denote
where \(\Vert \cdot \Vert _{2}\) is the Euclidean norm in \(\mathbb{C}^{m}\). We are looking for a fixed point of the functional
Φ sends the closed ball \(B_{X_{T}} (0, R)\) into itself. Let \(R= 2 \Vert H_{0} \Vert _{2} \) and let \(H\in X_{T} \) such that \(\Vert H \Vert _{X_{T}}\leq R\). Using (9), there exists a constant \(C_{1}(R)>0\) such that
where \(\Vert \mathbf{M} \Vert _{2} = \sup_{ \Vert Y \Vert _{2} =1 } \Vert \mathbf{M} Y \Vert _{2}\). If you choose T such that \((\frac{2 \Vert \mathbf{M} \Vert _{2} \sqrt{T}}{\sqrt{\pi}} + \frac{C_{1}(R)T}{R} )\leq\frac{1}{2}\) we see that the functional Φ sends the closed ball \(B_{X_{T}} (0, R)\) into itself.
Φ is a contraction mapping in \(B_{X_{T}} (0, R)\). Let \(H, L \in B_{X_{T}} (0, R)\) such that \(\Vert H \Vert _{X_{T}}\leq R\) and \(\Vert L \Vert _{X_{T}}\leq R\). Let \(t< T\) where T satisfies \((\frac{2 \Vert \mathbf{M} \Vert _{2} \sqrt{T}}{\sqrt{\pi}} + \frac {C_{1}(R)T}{R} )\leq\frac{1}{2}\). We have
Since P is a polynomial function and \(B_{X_{T}} (0, R)\) is bounded, P has Lipschitz continuity in \(B_{X_{T}} (0, R)\). This shows that there exists a constant \(C_{2}(R)>0\) such that
By choosing \(C(R)=\min ( \frac{C_{1}(R)}{R}, C_{2}(R) )\), there exists \(T>0\) such that \((\frac{2 \Vert \mathbf{M} \Vert _{2} \sqrt{T}}{\sqrt{\pi}} + \frac{C_{1}(R)T}{R} )\leq\frac{1}{2}\) and \((\frac{2 \Vert \mathbf{M} \Vert _{2} \sqrt{T}}{\sqrt{\pi}} + C_{2}(R) T )\leq \frac{1}{2}<1\). This shows that the functional Φ is a contraction mapping in \(B_{X_{T}} (0, R)\). By applying the Banach fixed-point theorem, there exists a unique function H verifying equation (9). □
The following lemma was proved in [9, 29].
Lemma 2
Let \(\phi\in \mathbf {H}^{1/4}(0,\mathbf {T})\) and \(\psi\in \mathbf {H}^{3/4}(0,\mathbf {T})\), such that \(\psi(0)=0\), be two functions extended by zero outside \([0,T]\). Then we have the following inequalities:
Remark 1
Let \(\phi\in \mathbf {H}^{1/4}(t_{1},t_{2})\) be a function extended by zero outside \([t_{1},t_{2}]\); we have
We state that the following lemma proved in [9].
Lemma 3
Let the complex function \(w\in \mathbf {H}^{1}(\boldsymbol {\Omega})\) be defined in \(\boldsymbol {\Omega}=\mathopen{]}-\infty,0[\). Then we have
We introduce our technical result.
Lemma 4
Let \(\boldsymbol {\Omega}=\mathopen{]}-\infty,0[\), we assume that the sequence \((\lambda_{m} )_{m}\) of \(\mathbf {H}^{1}(\boldsymbol {\Omega})\) is such that \(\Vert \lambda_{m} \Vert _{H^{1}(\Omega)}\leq C\) and \(\lambda_{m} \longrightarrow 0\) in \(\mathbf {L}^{2}(\boldsymbol {\Omega})\) when \(m\longrightarrow+\infty\). Then \(\lambda_{m}(0)\longrightarrow0\) when \(m\longrightarrow+\infty\).
Proof
Since \(\lambda_{m} \in \mathbf {H}^{1}(\boldsymbol {\Omega})\), by using Lemma 3, we have
Hence the result. □
Finally, we state the last lemma, which has been proved in [30].
Lemma 5
Let \(K\times]0,T[\) be an open bounded subset of \(\mathbb{R}_{x} \times \mathbb{R}_{t}\), \(g_{\mu}\) and g are functions in \(L^{q}(K\times ]0,T[)\), \(1< q<\infty\), such that
Then \(g_{\mu}\rightharpoonup g\) in the weak topology of \(L^{q}(K\times]0,T[)\).
3 Well-posedness of NLS equation
In this section, we are able to announce and prove our main result.
Theorem 1
Let \(u_{0}\in \mathbf {H}^{1}(\boldsymbol {\Omega})\) be an initial data with compact support in Ω. Then there exists a unique function \(u\in \mathcal{C}^{0}([0,+\infty[;\mathbf {H}^{1}(\boldsymbol {\Omega}))\cap \mathcal{C}^{1}([0,+\infty[;[\mathbf {H}^{1}(\boldsymbol {\Omega})]^{\prime})\) solution to the NLS equation (2)–(4)–(6), where \(\mathcal{C}^{k}(\mathbf {I};\mathbf {E})\) represents the space of k times continuously differentiable functions on I in E, \([\mathbf {H}^{1}(\boldsymbol {\Omega} )]'\) is the dual of \(\mathbf {H}^{1}(\boldsymbol {\Omega})\).
Remark 2
If u is a solution of the NLS equation (2)–(4)–(6), then \(\tilde{u}: (x,t)\longmapsto u(-x,t) \) is also a solution of the following NLS equation:
where \(\tilde{\boldsymbol {\Omega}}=\mathopen{]}0,+\infty[\) and \(\tilde{\mathbb{V}}(x,t)=\int_{0}^{t}g ( \vert \tilde{u}(x,s) \vert ^{2} ) \,ds \).
The remainder of this section contains the proof of Theorem 1.
For the sake of simplicity, we make a change of the unknown solution to the nonlinear equation (2)–(4)–(6),
Therefore, the NLS equation (2)–(4)–(6) is rewritten in the form
Remark 3
From (12), we have the following properties:
-
(i)
\({ \vert v(x,t) \vert = \vert u(x,t) \vert }\),
-
(ii)
\({u(x,t)=\exp (i \int_{0}^{t}g ( \vert v(0,s) \vert ^{2} ) \,ds ) v(x,t)}\),
-
(iii)
\({ \Vert v(t) \Vert _{H^{m}(\Omega)}= \Vert u(t) \Vert _{H^{m}(\Omega)}}\) for all \(m\geq0\).
We use the Galerkin method to show that there exists a solution \(v\in \mathcal{C}^{0}([0,+\infty[;\mathbf {H}^{1}(\boldsymbol {\Omega}))\cap \mathcal{C}^{1}([0,+\infty[;[\mathbf {H}^{1}(\boldsymbol {\Omega})]^{\prime})\) of the NLS equation (13). This method is divided into three steps as shown below. Also, we prove the uniqueness of this solution in the last subsection. As a result, by Remark 3 there exists a unique function \(u\in \mathcal{C}^{0}([0,+\infty[;\mathbf {H}^{1}(\boldsymbol {\Omega}))\cap \mathbf {\mathcal{C}}^{1}([0,+\infty[;[\mathbf {H}^{1}(\boldsymbol {\Omega})]^{\prime})\), a solution of (2)–(4)–(6).
3.1 First step: approximate problem
Let \((\varphi_{k})_{k}\) be an orthonormal basis of functions in \(\mathbf {H}^{1}(\boldsymbol {\Omega})\). For \(m\geq1\), we set \(\mathcal{H}_{m} = \operatorname{Span}(\varphi _{1},\ldots, \varphi_{m})\) and we define the orthogonal projection operator \(P_{m}\) by
where \(\langle\cdot,\cdot\rangle_{H^{1}(\Omega)}\) is the scalar product in \(\mathbf {H}^{1}(\boldsymbol {\Omega})\). For \(m\geq1\), we shall approximate v in (13) by
which satisfies, for all \(k \in\{1,2,\ldots,m \}\),
We get the system
where F is a polynomial function, A and B are square matrices of order m defined by
and
such that at \(t=0\) the components of \(H_{m}(0)\) are
The matrix A is Hermitian and positive-definite, then it is invertible. Therefore
We set
and
The system (19) becomes
where P is a polynomial function of \(\mathbb{C}^{m}\). By using Lemma 1, we see that the system (20) has a unique local solution. Then we obtain a unique function \(H_{m} = (h_{1m},\ldots,h_{mm})\) in \([0, T_{m}]\), a solution of (17) with the initial condition (18). As a result, the approximate problem (16) has a unique solution \(v_{m}\) such that \(v_{m}: [0,T_{m}]\rightarrow\mathcal{H}_{m} \). The existence of a maximal solution \(v_{m}\) (defined on \([0, T_{\mathrm{max}}[\)) is obtained by iterating m, where \(T_{\mathrm{max}}\) is the maximum time of the existence such that \(v_{m}: [0,T_{\mathrm{max}}[\rightarrow\mathcal{H}_{m} \). Then we have \(T_{\mathrm{max}}<+\infty\) and \(\lim_{t \to T_{\mathrm{max}}} \vert v_{m} \vert =+\infty\), or \(T_{\mathrm{max}}=+\infty\).
3.2 Second step: a priori estimates
3.2.1 Estimate in \(\mathbf {L}^{2}(\boldsymbol {\Omega})\)
We multiply (13) by \(-i\overline{v}\) and we integrate in space domain Ω. We then integrate by parts the second term and consider the real part, to get
We integrate this equality between 0 and t with \(t\in[0,+\infty [\), to get
By using the boundary condition for (13) and Lemma 2, we show that \(\int_{0}^{t}\operatorname{\mathcal{R}e} (i \overline {v(0,s)} \partial_{n} v(0, s) )\,ds \leq0\). Hence
This gives the a priori estimate in \(\mathbf {L}^{2}(\boldsymbol {\Omega})\)
3.2.2 Estimate in \(\mathbf {H}^{1}(\boldsymbol {\Omega})\)
We multiply the first equality of (13) by \(\overline{v_{t}}\) and we integrate in Ω. By considering the real part, we obtain
where
with G is defined by (3).
From the following, we set that K can be any positive constant depending only on q, \(\alpha_{1}\), \(\alpha_{2}\), θ and \(K_{0}\).
We have
where \(\lambda(t)=\operatorname{sign} (\frac{d}{dt} \Vert v \Vert ^{2}_{L^{2}(\Omega)} )\). By using (3), (25) and equality (23), we obtain
Therefore
We use Lemma 3 and (22), to get
where \(T_{\mathrm{max}}>0\) is the maximum time of existence of \((v_{m})_{m}\), which gives
Integrating the equality (27) between 0 and t, we get
Using Lemma 2, we obtain
Using equation (24) of \(\Psi(v(t))\), we have
We will majorize the second member of (28).
By using the Young inequality, we get
We apply the Agmon and Young inequalities using (22), and we have
By considering (3) and by using the Gagliardo–Nirenberg and the Young inequalities, we have
Then, by using (28), (29), (30) and (31), we obtain
By taking the supremum on the left side of this inequality, we obtain
Then \(T_{\mathrm{max}}=+\infty\) and the sequence \((v_{m})_{m}\) remains bounded in \(\mathcal{C}_{b}([0,+\infty[; \mathbf {H}^{1}(\boldsymbol {\Omega}))\). Hence, there exists \(K_{1} > 0\), which is dependent on the equation data, such that
We now specify the space where \((v'_{m})_{m}\) remains bounded. By (16), we see that
where \(P_{m}^{*}\) is the operator of \([\mathbf {H}^{1}(\boldsymbol {\Omega})]^{\prime}\) in \(\mathcal{H}_{m}\) defined by \(\forall\eta\in[\mathbf {H}^{1}(\boldsymbol {\Omega})]^{\prime}\), \(\forall\omega\in \mathbf {H}^{1}(\boldsymbol {\Omega})\), \(\langle P_{m}^{*}\eta,\omega\rangle _{H^{1}(\Omega),[H^{1}(\Omega)]'}=\langle\eta,P_{m}\omega\rangle _{[H^{1}(\Omega)]',H^{1}(\Omega)} \). We see that \(P_{m}^{*}\) is a bounded operator on \([\mathbf {H}^{1}(\boldsymbol {\Omega} )]^{\prime}\). The operator \(\partial_{x}^{2}: \mathbf {H}^{1}(\boldsymbol {\Omega})\longrightarrow[\mathbf {H}^{1}(\boldsymbol {\Omega} )]^{\prime}\) is continuous, then \((P_{m}^{*}(v_{mxx}))_{m}\) is a bounded sequence in \([\mathbf {H}^{1}(\boldsymbol {\Omega})]^{\prime}\). On the other hand, the sequence \((\delta_{a} v_{m})_{m}\) remains bounded in \([\mathbf {H}^{\frac{3}{4}}(\boldsymbol {\Omega} )]'\) and the sequences \(P_{m}^{*}(g( \vert v_{m} \vert ^{2} ) v_{m})_{m}\) and \(P_{m}^{*}(g( \vert v_{m}(0,t) \vert ^{2} ) v_{m})_{m}\) remain bounded in \(\mathbf {H}^{1}(\boldsymbol {\Omega})\). Hence, we see that \((v^{\prime}_{m})_{m}\) remains bounded in \(\mathcal{C}_{b} ([0,+\infty[; [\mathbf {H}^{1}(\Omega)]^{\prime})\).
3.2.3 Estimate of \(x v(t)\) in \(\mathbf {L}^{2}(\boldsymbol {\Omega})\)
In order to pass to the limit in the nonlinear term, we require the inclusion of the approximated solution \(v_{m}\) in \(\mathbf {H}^{1}(\boldsymbol {\Omega})\cap \mathbf {L}^{2} (\boldsymbol {\Omega}; (1+x^{2})\,dx )\). The reason behind this necessity is the well-known compact injection: \(\mathbf {H}^{1}(\boldsymbol {\Omega})\cap \mathbf {L}^{2} (\boldsymbol {\Omega}; (1+x^{2})\,dx )\) in \(\mathbf {L}^{2}(\boldsymbol {\Omega})\). We have already proved the estimate of \(v_{m}\) in \(\mathbf {H}^{1}(\boldsymbol {\Omega})\) and so this section shows the estimate of \(v_{m}\) in \(\mathbf {L}^{2} (\boldsymbol {\Omega}; (1+x^{2})\,dx )\).
We multiply (13) by \(-i x^{2} \overline{v}\) and we integrate in Ω. By using integration by parts and by considering the real part, we get
Since \(v\in \mathbf {H}^{1}(\boldsymbol {\Omega})\) we have \(\frac{1}{2} \operatorname{\mathcal{R}e} [i x^{2} v_{x}(x,t)\overline{v(x,t)} ]_{-\infty}^{0}=0 \). Then, by using the Cauchy–Schwarz inequality and (32), we have
By applying the Young inequality, we get
with \(\epsilon>0\). By applying the Gronwall lemma, for all \(T\in \mathopen{]}0,+\infty[\) we have
Since \(u_{0}\) is with compact support in Ω, we have \(\Vert x u_{0} \Vert _{L^{2}(\Omega)}^{2} \leq K\).
Hence, there exists a constant \(K(T)>0\) such that
This gives the sequence
3.3 Third step: passing to the limit
Since the sequence \((v_{m})_{m}\) remains bounded in \(\mathcal {C}_{b}([0,+\infty[; \mathbf {H}^{1}(\boldsymbol {\Omega}))\), for \(\forall T>0\), \((v_{m})_{m}\) is bounded in \(\mathbf {L}^{\infty}(0,\mathbf {T};\mathbf {H}^{1}(\boldsymbol {\Omega}))\). By the Banach–Alaoglu theorem, we deduce that \((v_{m})_{m}\) admits a subsequence still denoted \((v_{m})_{m}\) such that
By using (33) and since the embedding \(\mathbf {H}^{1}(\boldsymbol {\Omega})\cap \mathbf {L}^{2} (\boldsymbol {\Omega}; (1+x^{2})\,dx ) \hookrightarrow \mathbf {L}^{2}(\boldsymbol {\Omega})\) is compact, we have
On the other hand, the sequence \((\frac{d v_{m}}{dt})_{m}\) is bounded in \(\mathbf {L}^{\infty}(0,\mathbf {T};[\mathbf {H}^{1}(\boldsymbol {\Omega})]')\). Then it admits a subsequence which converges weakly ⋆ to \(h\in \mathbf {L}^{\infty}(0,\mathbf {T};[\mathbf {H}^{1}(\boldsymbol {\Omega})]')\). We have \(h=\frac{d v}{dt}\). Indeed, we have \(\frac{d v_{m}}{dt}\longrightarrow h \) in \(\mathcal{D}'((0,T)\times \boldsymbol {\Omega}) \). Otherwise, using (34) we obtain \(\frac{d v_{m}}{dt}\longrightarrow\frac{d v}{dt}\) in \(\mathcal {D}'((0,T)\times\Omega) \). By the uniqueness of the limit in \(\mathcal{D}'((0,T)\times \boldsymbol {\Omega})\), we obtain \(h=\frac{d u}{dt}\). This implies that
Now, we consider \(\omega\in\mathcal{D}([0, T ])\) such that \(\omega (T) = 0\). We pass to the limit in each term in the equation
where \(\langle\cdot,\cdot\rangle= \langle\cdot,\cdot\rangle _{[H^{1}(\Omega)]',H^{1}(\Omega)}\).
Passing to the limit for the term \(I_{m} = \int_{0}^{T} \langle i v'_{m},\omega(t)\varphi_{k} \rangle_{[H^{1}(\Omega )]',H^{1}(\Omega)} \,dt\): By integration by parts with respect to time, we get
By using (34), we obtain
Passing to the limit for the term \(J_{m}=\int_{0}^{T} \langle v_{mxx},\omega(t)\varphi_{k} \rangle_{[H^{1}(\Omega )]',H^{1}(\Omega)} \,dt\): Applying the Green formula, we have
By using (34) we have
It remains to demonstrate
Indeed,
By using (35) and by applying Lemma 4, we then obtain (39). This gives
Passing to the limit for the term \(L_{m}=\int_{0}^{T} \langle g ( \vert v_{m} \vert ^{2} ) v_{m},\omega(t)\varphi _{k} \rangle_{[H^{1}(\Omega)]',H^{1}(\Omega)} \,dt\): By using (34), (35) and (36), we have \(v_{m} \longrightarrow v\) strongly in \(\mathcal {C}^{0}([0,T];\mathbf {L}^{2}(\boldsymbol {\Omega}))\). Let K be a compact of Ω, since \(v_{m} g( \vert v_{m} \vert ^{2})\) belongs to a bounded set of \(\mathbf {L}^{\infty}(0,\mathbf {T},\mathbf {L}^{2}(\mathbf {K}))\) we extract a subsequence of \((v_{m} )_{m}\) (noted again \((v_{m} )_{m}\)) such that \(v_{m} g( \vert v_{m} \vert ^{2})\rightharpoonup w \) weakly ⋆ in \(\mathbf {L}^{\infty}(0,\mathbf {T},\mathbf {L}^{2}(\mathbf {K}))\). By using Lemma 5, we have \(g( \vert v_{m} \vert ^{2}) v_{m} \longrightarrow g ( \vert v \vert ^{2} ) v\) strongly in \(\mathcal{C}^{0}([0,T];\mathbf {L}^{2}(\boldsymbol {\Omega}))\). Hence
Passing to the limit for the term \(K_{m}=\int_{0}^{T} g ( \vert v_{m}(0,t) \vert ^{2} ) \langle v_{m},\omega(t)\varphi _{k} \rangle_{[H^{1}(\Omega)]',H^{1}(\Omega)} \,dt\): We have
Since g is continuous and by using Lemma 4, we have
By using the fact that \((v_{m})_{m}\) is bounded in \(\mathcal{C}_{b} ([0,\mathbf {T}];\mathbf {H}^{1}(\boldsymbol {\Omega}))\), we get
Using the weak convergence ⋆ of \((v_{m})_{m}\) to v in \(\mathbf {L}^{\infty}(0,\mathbf {T};\mathbf {H}^{1}(\boldsymbol {\Omega}))\) and \(g ( \vert v(0,\cdot) \vert ^{2} )\in \mathbf {L}^{\infty}(0,\mathbf {T})\), we obtain
This gives
Passing to the limit for the term \(D_{m}=\int_{0}^{T} \langle\delta_{a} v_{m},\omega(t)\varphi_{k} \rangle _{[H^{1}(\Omega)]',H^{1}(\Omega)} \,dt\): For all \(\varepsilon>0\), we have \(\delta_{a}\in[\mathbf {H}^{\frac {1}{2}+\varepsilon}(\boldsymbol {\Omega})]'\). Then the sequence \((\delta_{a} v_{m}(t))_{m}\) is bounded in \([\mathbf {H}^{\frac {1}{2}+\varepsilon}(\boldsymbol {\Omega})]'\); in particular in \([\mathbf {H}^{\frac{3}{4}}(\boldsymbol {\Omega})]'\). Indeed,
By using Agmon’s inequality, we have
We use the Banach–Alaoglu theorem to obtain
Therefore \(\forall\psi\in \mathbf {L}^{1}(0,\mathbf {T};\mathbf {H}^{3/4}(\boldsymbol {\Omega}))\) we have
In particular \(\forall\psi\in \mathbf {L}^{1}(0,\mathbf {T};\mathbf {H}^{1}(\boldsymbol {\Omega}))\) we have
We just justify that \(\xi=\delta_{a} v\). Note that \(\delta_{a} v \in [\mathbf {H}^{\frac{3}{4}}(\boldsymbol {\Omega})]'\). Using (43) we have
We know that \(v_{m} \longrightarrow v \mbox{ strongly in }\mathcal{C} ([0,\mathbf {T}];\mathbf {L}_{\mathrm{loc}}^{2}(\boldsymbol {\Omega}) ) \) and \((v_{m})_{m}\) is bounded in \(\mathcal{C}^{0}([0,\mathbf {T}]; \mathbf {H}^{1}(\boldsymbol {\Omega} ))\). Then, for all K compact on Ω, we have
Otherwise, we have
By using the uniqueness of the limit in \(\mathcal{D}'\), we get \(\xi=\delta_{a} u\). This implies that
Passing to the limit for equation ( 37 ): By using (37), (38), (40), (41), (42) and (44), we deduce that, for all \(\varphi\in \mathbf {H}^{1}(\boldsymbol {\Omega})\),
For \(\omega\in\mathcal{D}(0, \mathbf {T})\), we see that v obeys
We just justify that for all \(\varphi\in \mathbf {H}^{1}(\boldsymbol {\Omega})\) we have
Then v satisfies the initial condition \(v(0)=u_{0}\). We consider \(\omega\in\mathcal{D}([0,T])\); \(\omega(T)=0\), and multiply equation (46) by \(\omega(t)\). We integrate between 0 and T to obtain
Combining (45) and (47), we get
By choosing \(\omega(0)=1\), we deduce that \(v(0)=u_{0}\). We then justify that
and this satisfies the problem (13).
3.4 Uniqueness and continuous dependence of the solutions
Let \(v(t)\) and \(\widehat{v}(t)\) be two solutions satisfying the problem (13) which follow, respectively, from the initial data \(u_{0}\) and \(\widehat{u}_{0}\). We set \(w(t)=v(t)-\widehat{v}(t)\) with the initial condition \(w(0)=u_{0}-\widehat{u}_{0}\). Then we obtain
with the boundary condition
We multiply (48) by w̅ and we take the imaginary part, to obtain
By using (49), we have
The other right terms of (50) are shown to be bounded by applying the Cauchy–Schwarz inequality and the fact that the injection of \(\mathbf {H}^{1}(\boldsymbol {\Omega})\) in \(\mathbf {L}^{\infty}(\boldsymbol {\Omega})\) is continuous, and by using (3), we have
and by applying Lemma 3, we have
From the above inequalities, we obtain
We apply Gronwall’s lemma and Lemma 2, to obtain
Therefore the uniqueness of the solution of (13) follows immediately.
Finally, from Remark 3, there exists a unique function \(u\in\mathcal{C}^{0}([0,+\infty[;\mathbf {H}^{1}(\boldsymbol {\Omega}))\cap \mathcal {C}^{1}([0,+\infty[;[\mathbf {H}^{1}(\boldsymbol {\Omega})]')\) solution of the NLS equation (2)–(4)–(6) such that
Moreover, the following proposition shows that the continuous dependence of the solutions of the NLS equation (2)–(4)–(6).
Proposition 1
The map
is continuous on bounded subsets of \(\mathbf {H}^{1}(\boldsymbol {\Omega})\) for the strong topology of \(\mathbf {L}^{2}(\boldsymbol {\Omega})\).
Proof
Let \(u(t)\) and \(\widehat{u}(t)\) be two solutions of NLS equation (2)–(4)–(6) which are issued, respectively, from the initial data \(u_{0}\) and \(\widehat{u}_{0}\). By using Remark 3, we have
Then we have
By applying the mean value theorem and (3) and by using Lemma 3, we get
where \(T>0\). Finally, by taking the supremum on the left side of the last inequality and by applying (51), we get
which gives the result. □
Remark 4
The choice of the negative half-line Ω does not affect the well-posedness of the NLS equation. In fact, according to Remark 2, we deduce that if \(\Omega=\mathopen{]}0,+\infty[\), \(a>0\) and \(u_{0}\) has a compact support in Ω, then the NLS equation
has a unique solution \(u\in\mathcal{C}^{0}([0,+\infty[;\mathbf {H}^{1}(\boldsymbol {\Omega} ))\cap\mathcal{C}^{1}([0,+\infty[;[\mathbf {H}^{1}(\boldsymbol {\Omega})]')\). Moreover, the map \(u_{0} \mapsto u(t)\) is continuous on bounded subsets of \(\mathbf {H}^{1}(\boldsymbol {\Omega})\) for the strong topology of \(\mathbf {L}^{2}(\boldsymbol {\Omega})\).
4 Conclusion
In this paper, we have studied a nonlinear Schrödinger equation with Dirac distribution in a half-line domain of \(\mathbb{R}\). For this purpose, a non-standard boundary condition was considered in order to demonstrate the well-posedness of the solution. Then, by using the Galerkin method, we have shown that this equation can have a unique solution in \(\mathbf {H}^{1}(\boldsymbol {\Omega})\).
The remaining question is to investigate the nonlinear Schrödinger equation with the Dirac distribution on a bounded domain of \(\mathbb {R}\) or \(\mathbb{R}^{2}\) with non-standard boundary conditions. This is a delicate issue which needs future research.
References
Pitaevskiĭ, L.P.: Vortex lines in an imperfect Bose gas. Sov. Phys. JETP 13, 451–454 (1961)
Gross, E.P.: Structure of a quantized vortex in boson systems. Nuovo Cimento 20, 454–477 (1961)
Holmer, J., Zworski, M.: Slow soliton interaction with delta impurities. J. Mod. Dyn. 1, 689–718 (2007)
Goodman, R.H., Holmes, P.J., Weinstein, M.I.: Strong NLS soliton–defect interactions. Physica D 192, 215–248 (2004)
Fu, Y., Rădulescu, V.D., Zhang, B.: Hodge decomposition of variable exponent spaces of Clifford-valued functions and applications to Dirac and Stokes equations. Comput. Math. Appl. 70(4), 691–704 (2015)
Goubet, O., Hamraoui, E.: Blow-up of solutions to cubic nonlinear Schrödinger equations with defect: the radial case. Adv. Nonlinear Anal. 6(2), 183–197 (2017)
Ghidaglia, J.-M.: Finite dimensional behavior for weakly damped driven Schrödinger equations. Ann. Inst. Henri Poincaré, Anal. Non Linéaire 5(4), 365–405 (1988)
Bahrouni, A., Ounaies, H., Rădulescu, V.D.: Infinitely many solutions for a class of sublinear Schrödinger equations with indefinite potentials. Proc. R. Soc. Edinb. A 145(3), 445–465 (2015)
Abounouh, M., Al Moatassime, H., Chrifi, A.: Existence of global attractor for one-dimensional weakly damped nonlinear Schrödinger equation with Dirac interaction and artificial boundary condition in half-line. Adv. Differ. Equ. 2017(1), 137 (2017)
Claerbout, J.F.: Coarse grid calculation of waves in inhomogeneous media with application to delineation of complicated seismic structure. Geophysics 35(3), 407–418 (1970)
Burgnies, L., Vanbésien, O., Lippens, D.: Transient analysis of ballistic transport in stublike quantum waveguides. Appl. Phys. Lett. 71, 803 (1997)
Levy, M.: Parabolic Equation Methods for Electromagnetic Wave Propagation. IEE Electromagnetic Waves Series, vol. 45. Institution of Electrical Engineers, London (2000)
Schmidt, F., Deuflhard, P.: Discrete transparent boundary conditions for the numerical solution of Fresnel’s equation. Comput. Math. Appl. 29(9), 53–76 (1995)
Tappert, F.D.: The Parabolic Approximation Method, pp. 224–287. Springer, Berlin (1977)
Sulem, C., Sulem, P.-L.: The Nonlinear Schrödinger Equation: Self-Focusing and Wave Collapse. Applied Mathematical Sciences, vol. 139. Springer, New York (1999)
Malomed, B.A., Azbel, M.: Modulation instability of a wave scattered by a nonlinear center. Phys. Rev. B 47(16), 10402–10406 (1993)
Cao, X.D., Malomed, B.A.: Soliton defect collisions in the nonlinear Schrödinger equation. Phys. Lett. A206, 177–182 (1995)
Holmer, J., Marzuola, J., Zworski, M.: Fast soliton scattering by delta impurities. Commun. Math. Phys. 274(1), 187–216 (2007)
Gazol, A., Passot, T., Sulem, P.L.: Coupling between nonlinear Alfvén waves and reduced magnetohydrodynamics for compressible fluids. Phys. Plasmas 6(8), 3114–3122 (1999)
Holzleitner, M., Kostenko, A., Teschl, G.: Dispersion estimates for spherical Schrödinger equations: the effect of boundary conditions. Opusc. Math. 36(6), 769–786 (2016)
Colorado, E.: On the existence of bound and ground states for some coupled nonlinear Schrödinger–Korteweg–de Vries equations. Adv. Nonlinear Anal. 6(4), 407–426 (2017)
Bao, W.: The nonlinear Schrödinger equation and applications in Bose–Einstein condensation and plasma physics. In: Dynamics in Models of Coarsening, Coagulation, Condensation and Quantization. Lect. Notes Ser. Inst. Math. Sci. Natl. Univ. Singap., vol. 9, pp. 141–239. World Sci. Publ., Hackensack (2007)
Cazenave, T.: Semilinear Schrödinger Equations. Courant Lecture Notes in Mathematics, vol. 10. American Mathematical Society, Providence (2003)
Glassey, R.T.: On the blowing up of solutions to the Cauchy problem for nonlinear Schrödinger equations. J. Math. Phys. 18(9), 1794–1797 (1977)
Antoine, X., Besse, C., Klein, P.: Absorbing boundary conditions for the one-dimensional Schrödinger equation with an exterior repulsive potential. J. Comput. Phys. 228, 312–335 (2009)
Antoine, X., Besse, C., Klein, P.: Absorbing boundary conditions for Schrödinger equations with general potentials and nonlinearities. SIAM J. Sci. Comput. 33, 1008–1033 (2011)
Antoine, X., Arnold, A., Besse, C., Ehrhardt, M., Schädle, A.: A review of transparent and artificial boundary conditions techniques for linear and nonlinear Schrödinger equations. Commun. Comput. Phys. 4, 729–796 (2008)
Idczak, D., Walczak, S.: On a linear-quadratic problem with Caputo derivative. Opusc. Math. 36(1), 49–68 (2016)
Arnold, A.: Numerically absorbing boundary conditions for quantum evolution equations. VLSI Des. 4(1–4), 313–319 (1998)
Lions, J.-L.: Quelques Méthodes de Résolution des Problème aux Limites Non Linéaires. Dunod, Paris (1969)
Acknowledgements
We are thankful to the editor and the anonymous reviewers for many valuable suggestions to improve this paper.
Funding
This work is supported by National Center for Scientific and Technical Research (CNRST) of Morocco (I003/009).
Author information
Authors and Affiliations
Contributions
All authors contributed equally and significantly in writing this article. All authors read and approved the final manuscript.
Corresponding author
Ethics declarations
Competing interests
The authors declare that they have no competing interests.
Additional information
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.
About this article
Cite this article
Abounouh, M., Al Moatassime, H. & Chrifi, A. Artificial boundary condition for one-dimensional nonlinear Schrödinger problem with Dirac interaction: existence and uniqueness results. Bound Value Probl 2018, 16 (2018). https://doi.org/10.1186/s13661-018-0935-9
Received:
Accepted:
Published:
DOI: https://doi.org/10.1186/s13661-018-0935-9