\documentclass[reqno]{amsart} \usepackage{hyperref} \AtBeginDocument{{\noindent\small {\em Electronic Journal of Differential Equations}, Vol. 2005(2005), No. 122, pp. 1--31.\newline ISSN: 1072-6691. URL: http://ejde.math.txstate.edu or http://ejde.math.unt.edu \newline ftp ejde.math.txstate.edu (login: ftp)} \thanks{\copyright 2005 Texas State University - San Marcos.} \vspace{9mm}} \begin{document} \title[\hfilneg EJDE-2005/122\hfil Exact boundary controllability] {Exact boundary controllability for higher order nonlinear Schr\"{o}dinger equations with constant coefficients} \author[J. C. Ceballos V., R. Pavez F., O. P. Vera V.\hfil EJDE-2005/122\hfilneg] {Juan Carlos Ceballos V., Ricardo Pavez F.,\\ Octavio Paulo Vera Villagr\'{a}n} \address{Juan Carlos Ceballos V. \hfill\break Departamento de Matem\'{a}tica, Universidad del B\'{\i}o-B\'{\i}o, Collao 1202, Casilla 5-C, Concepci\'{o}n, Chile} \email{jceballo@ubiobio.cl \quad Fax 56-41-731018} \address{Ricardo Pavez F.\hfill\break Departamento de Matem\'{a}tica, Universidad del B\'{\i}o-B\'{\i}o, Collao 1202, Casilla 5-C, Concepci\'{o}n, Chile} \email{rpavez@ubiobio.cl \quad Fax 56-41-731018} \address{Octavio Paulo Vera Villagr\'{a}n \hfill\break Departamento de Matem\'{a}tica, Universidad del B\'{\i}o-B\'{\i}o, Collao 1202, Casilla 5-C, Concepci\'{o}n, Chile} \email{overa@ubiobio.cl \quad octavipaulov@yahoo.com\quad Fax 56-41-731018} \date{} \thanks{Submitted January 20, 2005. Published October 31, 2005.} \thanks{J. C. Ceballos was supported by grant 0528081/R from Proyectos de Investigacion Internos, \hfill\break\indent Universidad del B\'{\i}o-B\'{\i}o. Concepci\'{o}n. Chile} \subjclass[2000]{35K60, 93C20} \keywords{KdVK equation; boundary control; Hilbert uniqueness method; \hfill\break\indent Ingham's inequality; smoothing properties} \begin{abstract} The exact boundary controllability of the higher order nonlinear Schr\"{o}dinger equation with constant coefficients on a bounded domain with various boundary conditions is studied. We derive the exact boundary controllability for this equation for sufficiently small initial and final states. \end{abstract} \maketitle \numberwithin{equation}{section} \newtheorem{theorem}{Theorem}[section] \newtheorem{lemma}[theorem]{Lemma} \newtheorem{remark}[theorem]{Remark} \allowdisplaybreaks \section{Introduction} We consider the initial-value problem \begin{equation} \label{eHSCHROD} \begin{gathered} i u_{t} + \alpha u_{xx} + i \beta u_{xxx} + |u|^{2} u =0, \quad x, t\in \mathbb{R}\\ u(x, 0) = u_{0}(x) \end{gathered} \end{equation} where $\alpha , \,\beta \in \mathbb{R},$ $\beta \neq 0 $ and $u$ is a complex valued function. The above equation is a particular case of the equation \begin{equation} \label{eQ} \begin{gathered} i u_{t} + \alpha u_{xx} + i \beta u_{xxx} + \gamma |u|^{2} u + i \delta |u|^{2} u_{x} + i \epsilon u^{2} \overline{u}_{x}=0, \quad x, t\in \mathbb{R}\\ u(x, 0) = u_{0}(x) \end{gathered} \end{equation} where $\alpha ,\, \beta ,\, \gamma , \,\delta $, with $\beta \neq 0 $ and $u$ is a complex valued function. This equation was first proposed by Hasegawa and Kodama \cite{h1} as a model for the propagation of a signal in a fiber optic (see also \cite{k1}). The equation \eqref{eQ} can be reduced to other well known equations. For instance, setting $\alpha =1$, $\beta = \epsilon =\gamma =0$ in \eqref{eQ} we have the semi linear Schr\"{o}dinger equation, i. e., \begin{equation} u_{t} - i u_{xx} - i \gamma |u|^{2} u =0.\label{eQ1} \end{equation} If we let $\beta = \gamma =0$ and $\alpha =1$ in \eqref{eQ} we obtain the derivative nonlinear Schr\"{o}dinger equation \begin{equation} u_{t} - i u_{xx} - \delta |u|^{2} u_{x} - \epsilon u^{2} \overline{u}_{x}=0.\label{eQ2} \end{equation} Letting $\alpha = \gamma = \epsilon =0$ in \eqref{eQ}, the equation that arises is the complex modified Korteweg-de Vries equation, \begin{equation} u_{t} + \beta u_{xxx} + \delta |u|^{2} u_{x} =0.\label{eQ3} \end{equation} The initial-value problem for the equations \eqref{eQ1}, \eqref{eQ2} and \eqref{eQ3} has been extensively studied, see for instance \cite{b1,c3,k2,l2,l4,l5,m1,r1} and references therein. In 1992, Laurey \cite{l1} considered the equation \eqref{eQ} and proved local well-posedness of the initial-value problem associated for data in $H^{s}(\mathbb{R})$ with $s>3/4$, and global well-posedness in $H^{s}(\mathbb{R})$ where $s\geq 1$. In 1997, Staffilani \cite{s2} established local well-posedness for data in $H^{s}(\mathbb{R})$ with $s\geq 1/4$, improving Laurey's result. Similar results were given in \cite{c1,c2} for \eqref{eQ} where $w(t)$, $\beta(t)$ are real functions. For the case of the \eqref{eHSCHROD} if we consider the Gauge transformation \begin{equation*} u(x, t) = e^{i \frac{\alpha}{3} x + i 2 \frac{\alpha^{3}}{27}} v(x - \frac{\alpha^{2}}{3} t, t) \equiv e^{\theta} v(\eta, \xi) \end{equation*} where $\theta=i \frac{\alpha}{3} x + i 2 \frac{\alpha^{3}}{27},$ $\eta=x - \frac{\alpha^{2}}{3} t$ and $\xi =t,$ then \begin{gather*} u_{t} = i 2 \frac{\alpha^{3}}{27} e^{\theta} v - \frac{\alpha^{2}}{3} e^{\theta} v_{\eta} + e^{\theta} v_{\xi}\\ u_{xx} = - \frac{\alpha^{2}}{9} e^{\theta} v + i \frac{2}{3} \alpha e^{\theta} v_{\eta} + e^{\theta} v_{\eta \eta}\\ u_{xxx} = - i \frac{\alpha^{3}}{27} e^{\theta} v - \frac{1}{3} \alpha^{2} e^{\theta} v_{\eta} + i \alpha e^{\theta} v_{\eta \eta} + e^{\theta} v_{\eta \eta \eta}. \end{gather*} Replacing in \eqref{eHSCHROD} and considering $\beta=1$(rescaling the equation) we obtain \begin{equation} \label{eKdVm} \begin{gathered} i v_{\xi} + i v_{\eta \eta \eta} + |v|^{2} v - \frac{4}{27} \alpha^{3} v=0, \quad x, t\in \mathbb{R}\\ v(x, 0) = v_{0}(x)\equiv u_{0}(x) e^{- i \frac{\alpha}{3}} \end{gathered} \end{equation} Thus \eqref{eHSCHROD} is reduced to a complex modified Korteweg-de Vries type equation. In this paper, we consider the boundary control of the Schr\"{o}dinger equation \begin{equation} \label{e101} i u_{t} + \alpha u_{xx} + i \beta u_{xxx} + |u|^{2} u + i \delta u_{x}=0 \end{equation} where $\alpha , \,\beta , \,\delta \in \mathbb{R}$, $\beta \neq 0 $ and $u$ is a complex valued function on the domain $(a, b),$ $t>0,$ and with the boundary condition \begin{equation} \label{e102}u(a, t)=h_{0},\quad u(b, t)=h_{1},\quad u_{x}(a, t)=h_{2},\quad u_{x}(b, t)=h_{3}. \end{equation} In this paper we want to study directly the exact boundary controllability problem for the higher order Schr\"{o}dinger equation by adapting the method of \cite{l5} which combines the Hilbert Uniqueness Method (HUM) and multiplier techniques. This method has been successfully applied to study controllability of wave and plate equations, Schr\"{o}dinger and KdV equations (see for instance \cite{b1,c3,f1,h2,k2,k3,l2,l4,l6,m1} and references therein). The first result of this paper concerns boundary controllability of the higher order linear Schr\"{o}dinger equation. \begin{theorem} \label{thm1.1} Let $H_{p}^{2}=\{w\in H^{2}(0, 2 \pi): w(0)=w(2 \pi), \,w'(0)=w'(2 \pi)\}$ and $T>0.$ Then, for any $y_{0}, y_{T}\in (H_{p}^{2})'$ (the dual space of $H_{p}^{2}$), there exist $h_{k}\in L^{2}(0, T)$ ($k=0, 1, 2$) such that the solution $y\in C([0, T]: (H_{p}^{2})')$ of the boundary initial-value higher order Schr\"{o}dinger equation \begin{gather} \label{e103} i y_{t} + i \beta y_{xxx} + \alpha y_{xx}=0, \quad (x, t)\in (0, 2 \pi)\times (0, T);\\ \label{e104} \partial_{x}^{k}y(2 \pi , t) - \partial_{x}^{k}y(0, t)=h_{k}(t),\quad k=0, 1, 2;\\ \label{e105} y( . , 0)=y_{0} \end{gather} satisfies $y( . , T)=y_{T}$. \end{theorem} We see that explicit controls may be given. Unfortunately, the state $y$ is only known to belong to $C([0, T]: (H_{p}^{2})')$ so it seems quite difficult to deduce from Theorem \ref{thm1.1} controllability results for higher order nonlinear Schr\"{o}dinger equation \eqref{e101}. The second result relates exact boundary controllability for the linear higher order Schr\"{o}dinger equation with boundary control on $y_{x}$ at $x=L$. In this part a condition on the coefficients $\alpha $ and $\beta$ given by the second and the third order derivatives that appear in $(HSCHROD)$ is needed. A condition on the length $L$ of the domain appears. \begin{theorem} \label{thm1.2} Let $|\alpha |<3 \beta $, $\delta>0$ and \[ \mathcal{N} =\Big\{2 \pi \beta \sqrt{\frac{k^{2} + k l + l^{2}} {3 \beta \delta + \alpha^{2}}} : k, l\in \mathbb{N}^{*} \Big\}. \] Then for any $T>0$ and $L\in (0, +\infty )\setminus \mathcal{N}$, and for any $y_{0}, y_{T}\in L^{2}(0, L)$, there exists $h\in L^{2}(0, T)$ such that the mild solution $y\in C([0, T]: L^{2}(0, L)) \cap L^{2}(0, T: H^{1}(0, L))$ of the system \begin{gather} \label{e106} i y_{t} + i \beta y_{xxx} + \alpha y_{xx} + i \delta y_{x}=0\\ \label{e107} y(0, t)=y(L, t)=0\\ \label{e108} y_{x}(L, t)=h(t)\\ \label{e109} y(x, 0)=y_{0}(x) \end{gather} satisfies $y( . , T)=y_{T}$. \end{theorem} To prove this we use the Hilbert uniqueness method and the multiplier method. It turns out that the study of \eqref{e106}-\eqref{e109} as a boundary initial-value problem is more delicate than the study of \eqref{e103}-\eqref{e105}, and -because of the extra term $y_{x}$ in \eqref{e108}- the observability result holds true if and only if $L\notin \mathcal{N}$. On the other hand, the solution $y$ belongs this time to a functional space in which we may give a sense to the nonlinear term $|y|^{2} y$ in \eqref{eHSCHROD}. By means of the Banach Contraction Fixed Point Theorem and Theorem \ref{thm1.2} we get the main result of the paper, that is the exact boundary controllability of the higher order nonlinear Schr\"{o}dinger equation on a bounded domain. \begin{theorem} \label{thm1.3} Let $|\alpha |<3 \beta $, $\delta>0$, $T>0$ and $L>0$. Then, there exists $r_{0}>0$ such that for any $y_{0}$, $y_{T}\in L^{2}(0, L)$ with $\|y_{0}\|_{L^{2}(0, L)}0$. Then, for all $T>2\pi/\gamma $ there are two positive constants $C_{1}, C_{2}$ depending only on $\gamma $ and $T$ such that \begin{equation} \label{e208}C_{1}(T, \gamma)\sum_{k=-\infty}^{\infty}|a_{k}|^{2}\leq \int_{0}^{T} \big|\sum_{k=-\infty}^{\infty}a_{k} e^{i t \lambda_{k}}\big|\,dx \leq C_{2}(T, \gamma)\sum_{k=-\infty}^{\infty}|a_{k}|^{2} \end{equation} for every complex sequence $(a_{k})_{k\in \mathbb{Z}}\in l^{2}$, where \begin{equation}\label{e209} \begin{gathered} C_{1}(T, \gamma)=\frac{2 T}{\pi} \big( 1 - \frac{4 \pi^{2}}{T^{2} \gamma^{2}} \big)>0, \\ C_{2}(T, \gamma)=\frac{8 T}{\pi} \big( 1 + \frac{4 \pi^{2}}{T^{2} \gamma^{2}} \big)>0 \end{gathered} \end{equation} and $l^{2}$ is the Hilbert space of square summable sequences, sequences $\{a_{k}\}$ such that $\sum_{k\in \mathbb{N}}|a_{k}|^{2}<\infty$. \end{lemma} Finally, we denote by $c$, a generic constant, not necessarily the same at each occasion, which depends in an increasing way on the indicated quantities. \section{Exact boundary controllability of the higher order linear Schr\"{o}dinger equation by means of control on data $[\partial^{k}y( . , t)]_{0}^{2 \pi}$ for $k=0, 1, 2$} For simplicity, in this section, we restrict ourselves to the case where the space domain $[0, L]$ is $[0, 2 \pi]$; although Theorem \ref{thm1.1} holds for arbitrary $L>0$. \begin{lemma} \label{lem3.1} Let $A$ denote the operator $Au=(- \beta \partial^{3} + i \alpha \partial^{2})u$ on the domain $D(A)=H_{p}^{3}\subseteq L^{2}(0, 2 \pi )$. Then $A$ generates a strongly continuous unitary group $(S(t))_{t\in \mathbb{R}}$ on $L^{2}(0, 2 \pi )$. \end{lemma} \begin{proof} Let $A:D(A)\subseteq L^{2}(0, 2 \pi ) \mapsto L^{2}(0, 2 \pi )$ such that $u \mapsto Au=- \beta \partial^{3}u + i \alpha \partial^{2}u$. We have \begin{align*} \langle Au, v\rangle & = \langle - \beta \partial^{3}u + i \alpha \partial^{2}u, v\rangle\\ &= - \beta \langle \partial^{3}u, v\rangle + i \alpha \langle \partial^{2}u, v\rangle \\ & = \beta \langle u, \partial^{3}v\rangle + i \alpha \langle u, \partial^{2}v\rangle\\ &= \langle u, \beta \partial^{3}v\rangle + \langle u, -i \alpha \partial^{2}v\rangle \\ & = \langle u, - (-\beta \partial^{3}v + i \alpha \partial^{2}v)\rangle \\ & = \langle u, - Av\rangle \end{align*} then $A^{*}=- A$. Hence, by the Stone theorem \cite{p1}, $A$ is the infinitesimal generator of a unitary group of class $C_{0}$ (all groups of class $C_{0}$ are strongly continuous) on $L^{2}(0, 2 \pi)$. \end{proof} \noindent\textbf{Definition.} % 3.1.} Let $T>0$. For $u_{T}={\sum_{n\in\mathbb{Z}}c_{n} e^{i n t }\in L^{2}(0, 2 \pi)}$, the mild solution of the uncontrolled problem \begin{equation} \label{eP} \begin{gathered} \partial_{t}u + \beta \partial^{3}u - i \alpha \partial^{2}u = 0,\quad x\in (0, 2 \pi),\; t\in \mathbb{R} ;\\ \partial^{k}u(0, t) = \partial^{k}u(2 \pi, t),\quad k=0, 1, 2;\\ u( . , T)= u_{T}( . ) \end{gathered} \end{equation} is given by \begin{equation} \label{e301} u(x, t)= \sum_{n\in \mathbb{Z}}c_{n} e^{i (\beta n^{3} - \alpha n^{2}) (t - T) + i n x} \end{equation} \begin{remark} \label{rmk3.1} \rm Let $u(x, t) = {\sum_{n\in \mathbb{Z}}\widehat{u}(n, t) e^{i n x}}$, then \[ u(x, t)= \sum_{n\in \mathbb{Z}}c_{n} e^{i [ (\beta n^{3} - \alpha n^{2}) (t - T) + n x]}. \] In fact, \begin{gather*} \partial _{t}u(x, t) = \sum_{n\in \mathbb{Z}}\partial _{t}\widehat{u}(n, t) e^{i n x}\\ \partial^{2}u(x, t) = \sum_{n\in \mathbb{Z}}(i n)^{2} \widehat{u}(n, t) e^{i n x}= -\sum_{n\in \mathbb{Z}}n^{2} \widehat{u}(n, t) e^{i n x}\\ \partial^{3}u(x, t) = \sum_{n\in \mathbb{Z}}(i n)^{3} \widehat{u}(n, t) e^{i n x}= - i\sum_{n\in \mathbb{Z}}n^{3} \widehat{u}(n, t) e^{i n x}, \end{gather*} hence, if $u$ is the solution of \eqref{eP}, we obtain \[ \sum_{n\in \mathbb{Z}}\partial_{t}\widehat{u}(n, t) e^{i n x} - i \beta \sum_{n\in \mathbb{Z}}n^{3} \widehat{u}(n, t) e^{i n x} + i \alpha \sum_{n\in \mathbb{Z}}n^{2} \widehat{u}(n, t) e^{i n x} = 0. \] Multiplying by $e^{- i m x}$($m\in \mathbb{Z}$) and integrating over $x\in (0, 2 \pi)$ we obtain \[ \sum_{n\in \mathbb{Z}}\partial_{t}\widehat{u}(n, t) - i (\beta n^{3} - \alpha n^{2}) \widehat{u}(n, t)\int_{0}^{2 \pi}e^{i (n - m) x}dx =0. \] Using that \[ \int_{0}^{2 \pi}e^{i (n - m) x}dx = \begin{cases} 0, & \mbox{if } n\neq m \\ 2 \pi , & \mbox{if } n=m \end{cases} \] we have that ${\sum_{n\in\mathbb{Z}}\partial_{t}\widehat{u}(n, t) - i (\beta n^{3} - \alpha n^{2}) \widehat{u}(n, t)=0}$, then $\partial_{t}\widehat{u}(n, t) - i (\beta n^{3} - \alpha n^{2}) \widehat{u}(n, t)=0$ where \[ \partial_{t}\left[e^{- i (\beta n^{3} - \alpha n^{2}) t} \widehat{u}(n, t)\right ]=0. \] Integrating over $t\in [0, T]$ yields \[ \widehat{u}(n, t) = \widehat{u}(n, 0) e^{ i (\beta n^{3} - \alpha n^{2}) t} \] multiplying by $e^{i n x}$ and applying ${\sum_{n\in\mathbb{Z}}}$ we obtain \begin{align*} u(x, t)&=\sum_{n\in \mathbb{Z}}\widehat{u}(n, t)e^{i n x} \\ & = \sum_{n\in \mathbb{Z}}\widehat{u}(n, 0) e^{ i [ (\beta n^{3} - \alpha n^{2}) t + n x ]}\\ & = \sum_{n\in \mathbb{Z}}\widehat{u}(n, 0) e^{i (\beta n^{3} - \alpha n^{2}) T} e^{ i [ (\beta n^{3} - \alpha n^{2}) (t - T) + n x ]}\\ & = \sum_{n\in \mathbb{Z}}c_{n} e^{ i (\beta n^{3} - \alpha n^{2}) (t - T) + i n x}. \end{align*} where $c_{n}=\widehat{u}(n, 0) e^{i (\beta n^{3} - \alpha n^{2}) T}$ and $u(x, T)=u_{T}={\sum_{n\in \mathbb{Z}}c_{n} e^{i n x}}$. \end{remark} For the rest of this article, $u$ will denote the solution of \eqref{eP} associated with $u_{T}$. We show the following result for the non-homogeneous problem. \begin{theorem} \label{thm3.1} Let $H_{p}^{2}=\{w\in H^{2}(0, 2 \pi): w(0)=w(2 \pi), w'(0)=w'(2 \pi)\}$ and $T>0$. Then for any $y_{0}$, $y_{T}\in ( H_{p}^{2} )'$ (the dual space of $H_{p}^{2}$), there exist $h_{k}\in L^{2}(0, T)$ ($k=0, 1, 2$) such that the solution $y\in C([0, T]: ( H_{p}^{2} )')$ of the boundary initial-value higher order Schr\"{o}dinger equation \begin{equation} \label{eP1} \begin{gathered} \partial_{t}y + \beta \partial^{3}y - i \alpha \partial^{2}y = 0,\quad (x, t)\in (0, 2 \pi)\times (0, T);\\ \partial^{k}y(2 \pi, t) - \partial^{k}y(0, t) = h_{k}(t),\quad k=0, 1, 2;\\ y( . , 0)= y_{0} \end{gathered} \end{equation} satisfies $y( . ,T)=y_{T}$. \end{theorem} \begin{remark} \label{rmk3.2} \rm Given $y_{0}\in (H_{p}^{2})'$, $h_{k}\in L^{2}(0, T)$ ($k=0, 1, 2$), we want to find $y$ such that it satisfies \eqref{eP1}. We first prove that \eqref{eP1} admits a unique solution $y\in C([0, T]: (H_{p}^{2})')$ in a certain sense, and this solution is the classical one whenever $y\in D(A)$, and $h_{k}(k=0, 1, 2)$ are smooth enough and vanish at 0. \end{remark} \begin{lemma} \label{lem3.2} \textbf{(1)} Assume that $h_{k}\in C_{0}^{2}([0, T])=\{h\in C^{2}([0, T]: \mathbb{C}): h(0)=0\}$ and $y_{0}\in H_{p}^{3}$. Then there exists a unique solution $y\in C([0, T]: H^{3}(0, 2 \pi))\cap C^{1}([0, T]: L^{2}(0, 2 \pi))$ of \eqref{eP1}. Moreover, for any $u_{T}\in H_{p}^{3}$ and any $t\in [0, T]$ we have \begin{equation} \begin{aligned} &\int_{0}^{2 \pi}u(x, t) \overline{y(x, t)}\,dx \\ & = \int_{0}^{2 \pi}u(x, 0) \overline{y_{0}(x)}\,dx - (\beta - i \alpha)\int_{0}^{t}\partial^{2}u(0, s) \overline{h_{0}(s)}\,ds \\ &\quad +\beta \int_{0}^{t}\partial u(0, s) \overline{h_{1}(s)}\,ds + \int_{0}^{t}u(0, s) (\overline{\beta h_{2}(s) + i \alpha h_{1}(s)})\,ds. \end{aligned} \label{e302} \end{equation} \textbf{(2)} For $u_{T}\in H_{p}^{2}$, $u\in C([0, T]: H_{p}^{2})$ and $\partial^{2}u(0, . )$ makes sense in $L^{2}(0, T)$. \noindent\textbf{(3)} Assume now that $y_{0}\in (H_{p}^{2})'$ and $h_{k}\in L^{2}(0, T)$($k=0, 1, 2$). Then, there exists a unique $y \in C([0, T]: (H_{p}^{2})')$ such that for all $u_{T}\in H_{p}^{2}$ and for all $t\in [0, T]$, \begin{equation} \label{e303} \begin{aligned} &\langle u( . , t), y(t)\rangle _{H_{p}^{2}\times (H_{p}^{2})'} \\ & = \langle u( . , 0), y_{0}\rangle _{H_{p}^{2}\times (H_{p}^{2})'} - (\beta - i \alpha)\int_{0}^{t}\partial^{2}u(0, s) \overline{h_{0}(s)}\,ds \\ &\quad + \beta \int_{0}^{t}\partial u(0, s) \overline{h_{1}(s)}\,ds + \int_{0}^{t}u(0, s) (\overline{\beta h_{2}(s) + i \alpha h_{1}(s)})\,ds \end{aligned} \end{equation} \end{lemma} \begin{proof} {\bf (1)} Let $\phi_{i}\in C^{\infty}([0, 2 \pi])$($i=0, 1, 2$) be such that \[ \phi_{i}^{(k)}(0) = 0\quad \mbox{and}\quad {\phi_{i}^{(k)}(2 \pi)}= \begin{cases} -1,& i=k \\ 0, & i\neq k. \end{cases} \] We consider the change of function $z(x, t) = \sum_{i=0}^{2} [ h_{i}(t) \phi_{i}(x) + (S(t)y_{0})(x) + y(x, t) ]$, then \begin{align*} z(2 \pi , t) - z(0, t) & = \sum_{i=0}^{2}h_{i}(t) \phi_{i}(2 \pi) + (S(t)y_{0})(2 \pi) + y(2 \pi, t) \\ &\quad - \sum_{i=0}^{2}h_{i}(t) \phi_{i}(0) + (S(t)y_{0})(0) + y(0, t)\\ & = -h_{0}(t) + (S(t)y_{0})(2 \pi) - (S(t)y_{0})(0) + y(2 \pi, t) - y(0, t)\\ & = -h_{0}(t) + (S(t)y_{0})(2 \pi) - (S(t)y_{0})(0) + h_{0}(t)\\ & = (S(t)y_{0})(2 \pi) - (S(t)y_{0})(0) \end{align*} using that $y_{0}\in H_{p}^{3}$ we obtain $z(2 \pi , t) =z(0, t)$. The other initial conditions are calculated in a similar way. Hence, this change of the function yields an equivalent problem to \eqref{eP1}: Find $z$ such that \begin{equation} \label{eP2} \begin{gathered} \partial _{t}z + \beta \partial^{3}z - i \alpha \partial^{2}z =f(x, t)\\ = \sum_{i=0}^{2}\left[h_{i}'(t) \phi_{i}(x) + \beta h_{i}(t) \phi_{i}^{(3)}(x) - i \alpha h_{i}(t) \phi_{i}^{(2)}(x)\right ] \\ \partial^{k}z(2 \pi, t) = \partial ^{k}z(0, t),\quad k=0, 1, 2 \\ z( . , 0)= 0 \end{gathered} \end{equation} Since $f\in C^{1}([0, T]: L^{2}(0, 2 \pi))$, this non-homogeneous problem admits a unique solution (see \cite{p1}), $z\in C([0, T]: H_{p}^{3})\cap C^{1}([0, T]: L^{2}(0, 2 \pi))$. This proves the first assertion in (1). Let $u_{T}\in H_{p}^{3}$, then $u\in C([0, T]: H_{p}^{3})\cap C^{2}([0, T]: L^{2}(0, 2 \pi))$. Multiplying the equation \eqref{eP} by $\overline{y}$ and integrating in $x\in [0, 2 \pi ]$ and $t\in[0, T]$ we have \[ \int_{0}^{t}\int_{0}^{2 \pi}\overline{y}[\partial_{s}u]\,dx\,ds + \beta \int_{0}^{t}\int_{0}^{2 \pi}\overline{y}[\partial^{3}u]\,dx\,ds - i \alpha \int_{0}^{t}\int_{0}^{2 \pi}\overline{y}[\partial^{2}u]\,dx\,ds=0. \] Each term is treated separately. Integrating by parts, \begin{align*} &\int_{0}^{t}\int_{0}^{2 \pi}\overline{y}[\partial_{s}u]\,dx\,ds \\ &= \int_{0}^{2 \pi}\overline{y(x, t)} u(x, t)\,dx - \int_{0}^{2 \pi}\overline{y(x, 0)} u(x, 0)\,dx - \int_{0}^{t}\int_{0}^{2 \pi}[\partial_{s}\overline{y}] u\,dx\,ds\,, \end{align*} \begin{align*} \beta \int_{0}^{t}\int_{0}^{2 \pi}\overline{y}[\partial^{5}u]\,dx\,ds &= \beta \int_{0}^{t}\overline{h_{0}(s)}[\partial^{2}u(0, s)]\,ds - \beta \int_{0}^{t}\overline{h_{1}(s)}[\partial u(0, s)]\,ds\\ &\quad +\beta \int_{0}^{t}\overline{h_{2}(s)} u(0, s)\,ds - \int_{0}^{t}\int_{0}^{2 \pi}[\partial^{3}\overline{y}] u\,dx\,ds\,, \end{align*} \begin{align*} &- i \alpha \int_{0}^{t}\int_{0}^{2 \pi}\overline{y}[\partial^{2}u]\,dx\,ds \\ &= - i \alpha \int_{0}^{t}\overline{h_{0}(s)}[\partial^{2}u(0, s)]\,ds + i \alpha \int_{0}^{t}\overline{h_{1}(s)} u(0, s)\,ds - i \alpha \int_{0}^{t}\int_{0}^{2 \pi}[\partial^{2}\overline{y}] u\,dx\,ds\,. \end{align*} Therefore, \begin{align*} & \int_{0}^{2 \pi}\overline{y(x, t)}\;u(x, t)\,dx - \int_{0}^{2 \pi}\overline{y(x, 0)} u(x, 0)\,dx - \int_{0}^{t}\int_{0}^{2 \pi}[\partial_{s}\overline{y}] u\,dx\,ds\\ &+ \beta \int_{0}^{t}\overline{h_{0}(s)}[\partial^{2}u(0, s)]\,ds - \beta \int_{0}^{t}\overline{h_{1}(s)}[\partial u(0, s)]\,ds + \beta \int_{0}^{t}\overline{h_{2}(s)}\;u(0, s)\,ds\\ &- \int_{0}^{t}\int_{0}^{2 \pi}[\partial^{3}\overline{y}] u\,dx\,ds - i \alpha \int_{0}^{t}\overline{h_{0}(s)}[\partial^{2}u(0, s)]\,ds +i \alpha \int_{0}^{t}\overline{h_{1}(s)} u(0, s)\,ds\\ &-i \alpha \int_{0}^{t}\int_{0}^{2 \pi}[\partial^{2}\overline{y}] u\,dx\,ds =0\,, \end{align*} where \begin{align*} &\int_{0}^{2 \pi}u(x, t) \overline{y(x, t)}\,dx\\ & = \int_{0}^{2 \pi}u(x, 0) \overline{y_{0}(x)}\,dx - (\beta - i \alpha)\int_{0}^{t}[\partial^{2}u(0, s)] \overline{h_{0}(s)}\,ds\\ &\quad +\beta \int_{0}^{t}[\partial u(0, s)] \overline{h_{1}(s)}\,ds + \int_{0}^{t}u(0, s) (\overline{\beta h_{2}(s) + i \alpha h_{1}(s)})\,ds. \end{align*} Result (1) follows. Now, we proof (2). By \eqref{e301}, for $t_{1}, t_{2}\in [0, T]$ \begin{gather*} u(x, t_{1})=\sum_{n\in \mathbb{Z}}c_{n} e^{i (\beta n^{3} - \alpha n^{2}) (t_{1} - T) + i n x }, \\ u(x, t_{2})=\sum_{n\in \mathbb{Z}}c_{n} e^{i (\beta n^{3} - \alpha n^{2}) (t_{2} - T) + i n x}; \end{gather*} hence \begin{align*} &u(x, t_{1}) - u(x, t_{2}) \\ &= \sum_{n\in \mathbb{Z}}c_{n} e^{i (\beta n^{3} - \alpha n^{2}) T}\big(e^{i (\beta n^{3} - \alpha n^{2}) t_{1}} - e^{i (\beta n^{3} - \alpha n^{2}) t_{2}}\big) e^{i n x}. \end{align*} From \eqref{e203}, if $u_{T}\in H_{p}^{2}$ then ${\sum_{n\in \mathbb{Z}}|n^{2} c_{n}|^{2}<\infty }$ and ${\sum_{n\in \mathbb{Z}}|n c_{n}|^{2}<\infty }$. Using Lebesgue's Theorem \cite{r2}, \[ |u(x, t_{1}) - u(x, t_{2})| = \sum_{n\in \mathbb{Z}}\big| (n^{2} + n) c_{n} \big(e^{i (\beta n^{3} - \alpha n^{2}) t_{1}} - e^{i (\beta n^{3} - \alpha n^{2}) t_{2}}\big)\big|^{2} \] which approaches $0$ as $t_{1}\to t_{2}$. We conclude that $u\in C([0, T]: H_{p}^{2})$. Hence $u(0, . )$, $\partial u(0, . )$ exist in $C([0, T])\subseteq L^{2}(0, T)$. The same argument shows that if $u_{T}\in H_{p}^{3}$, $u\in C([0, T]: H_{p}^{3})$ and \begin{equation} \label{e304}\partial^{2}u(0, t) = \sum_{n\in \mathbb{Z}} \big(- n^{2} c_{n} e^{- i (\beta n^{3} - \alpha n^{2}) T} \big) e^{i (\beta n^{3} - \alpha n^{2}) t}. \end{equation} The sum in \eqref{e304} makes sense in $L^{2}(0, T)$ wherever ${\sum_{n\in \mathbb{Z}} ( n^{2} |c_{n}| )^{2}< \infty }$, that is, $u_{T}\in H_{p}^{2}$. From now on, $\partial^{2}u(0, . )$ denotes for $u_{T}\in H_{p}^{2}$, the sum in \eqref{e304}. \end{proof} \begin{remark} \label{rmk3.3} \rm The linear map $u_{T}\mapsto \partial^{2}u(0, . )$ is continuous since \begin{equation} \label{e305} \big\| \sum_{n\in \mathbb{Z}} \big( n^{2} c_{n} e^{- i (\beta n^{3} - \alpha n^{2}) T}\big) e^{i (\beta n^{3} - \alpha n^{2}) t}\big\| \leq \big([ \frac{T}{2 \pi}] + 1\big ) \sum_{n\in \mathbb{Z}}[ n^{2} |c_{n}| ]^{2} \end{equation} where $[x]$ denotes the integral part of a real number $x$. Identifying $L^{2}(0, 2 \pi)$ with its dual by means of the conjugate linear map $y\mapsto \langle . , y\rangle _{L^{2}(0, 2 \pi )}$, we have the following dense and compact embedding (see \cite{l7}) \begin{equation} \label{e306}H_{p}^{2}\hookrightarrow L^{2}(0, 2 \pi) \hookrightarrow (L^{2}(0, 2 \pi))'\hookrightarrow (H_{p}^{2})'. \end{equation} Moreover, \begin{equation} \label{e307}\langle u, y\rangle _{H_{p}^{2}\times (H_{p}^{2})'} = \langle u, y\rangle _{L^{2}(0, 2 \pi)}=\int_{0}^{2 \pi}u \overline{y}\,dx \end{equation} for $u\in H_{p}^{2}$ and $y\in L^{2}(0, 2 \pi)$. Then \begin{align*} &\langle u( . , t), y(t)\rangle _{H_{p}^{2}\times (H_{p}^{2})'} \\ & = \langle u( . , 0), y_{0}\rangle _{H_{p}^{2}\times (H_{p}^{2})'} - (\beta - i \alpha)\int_{0}^{t}[\partial^{2}u(0, s)] \overline{h_{0}(s)}\,ds \\ &\quad +\beta \int_{0}^{t}\partial u(0, s) \overline{h_{1}(s)}\,ds + \int_{0}^{t}u(0, s) (\overline{\beta h_{2}(s) + i \alpha h_{1}(s)})\,ds \end{align*} for $h_{k}\in C_{0}^{2}([0, T]) $ ($k=0, 1, 2$) and $y_{0}, u_{T}\in H_{p}^{3}$. Since $H_{p}^{3}$ is dense in $H_{p}^{2}$, using (2), we see that \eqref{e303} also is true for $u_{T}\in H_{p}^{2}$. \end{remark} \noindent\textbf{Definition.}% 3.2.} For $y_{0}\in (H_{p}^{2})'$ and $h_{k}\in L^{2}(0, T)$ ($k=0, 1, 2$), we define a weak solution of \eqref{eP1} as a function $y\in C([0, T]: (H_{p}^{2})')$ such that \eqref{e303} holds for all $u_{T}\in H_{p}^{2}$ and all $t\in [0, T]$. \smallskip \noindent\textbf{Claim.} For $t$ fixed in $[0, T]$, \eqref{e303} defines $y(t)\in (H_{p}^{2})'$ in a unique manner. In fact, from the proof of (2) the map $\Xi : H_{p}^{2}\to \mathbb{C}$, $u_T\mapsto \Xi(u_T)$, given by \begin{align*} \Xi (u_{T}) & =- (\beta -i \alpha)\int_{0}^{t}\overline{h_{0}(s)} [\partial^{2}u(0, s)]\,ds + \beta \int_{0}^{t}\overline{h_{1}(s)}[\partial u(0, s)]\,ds \\ &\quad +\int_{0}^{t}( \beta \overline{h_{2}(s)} - i \alpha \overline{h_{1}(s)} ) u(0, s)\,ds \end{align*} is a continuous linear form. On the other hand, the map $\Phi :H_{p}^{2}\mapsto H_{p}^{2}$ with $u_{T}\to \Phi (u_{T}) =u( . , t)$ is an automorphism of the Hilbert space, hence, for each $t\in [0, T], $ $y(t)$ is uniquely defined in $(H_{p}^{2})'$. Moreover, for $t\in [0, T]$, \begin{align*} \|y(t)\|_{(H_{p}^{2})'} &=\sup_{\|u( . , t)\|_{H_{p}^{2}}\leq 1}|\langle u( . , t), y(t)\rangle| \\ &=\sup_{\|u( . , t)\|_{H_{p}^{2}}\leq 1}| \langle u( . , 0), y_{0} \rangle _{H_{p}^{4}\times (H_{p}^{2})'} - (\beta - i \alpha)\int_{0}^{t}[\partial^{2}u(0, s)] \overline{h_{0}(s)}\,ds\\ &\quad + \beta \int_{0}^{t}[\partial u(0, s)] \overline{h_{1}(s)}\,ds +\int_{0}^{t}u(0, s) (\overline{\beta h_{2}(s) + i \alpha h_{1}(s)})\,ds| \\ & \leq \sup_{\|u( . , t)\|_{H_{p}^{2}}\leq 1}| \langle u( . , 0), y_{0}\rangle _{H_{p}^{2}\times (H_{p}^{2})'}| \\ &\quad + (|\beta| +|\alpha |)\;\sup_{\|u( . , t)\|_{H_{p}^{2}}\leq 1} \int_{0}^{t}| \overline{h_{0}(s)} [\partial^{2}u(0, s)] |\,ds \\ &\quad +|\beta | \sup_{\|u( . , t)\|_{H_{p}^{2}}\leq 1} \int_{0}^{t} | \overline{h_{1}(s)} [\partial u(0, s)] |\,ds \\ &\quad+ \sup_{\|u( . , t)\|_{H_{p}^{2}}\leq 1} \int_{0}^{t}|( \beta \overline{h_{2}(s)} - i \alpha \overline{h_{1}(s)} ) u(0, s) |\,ds \\ & \leq \sup_{\|u( . , t)\|_{H_{p}^{2}}\leq 1}\|u( . , 0)\|_{(H_{p}^{2})'} \|y_{0}\|_{H_{p}^{2}} \\ &\quad + (|\beta| + |\alpha |)\; \sup_{\|u( . , t)\|_{H_{p}^{2}}\leq 1} \|\overline{h_{0}(t)}\|_{L^{2}(0, T)} \|\partial^{2}u(0, t)\|_{L^{2}(0, T)} \\ &\quad +|\beta |\sup_{\|u( . , t)\|_{H_{p}^{2}}\leq 1} \|\overline{h_{1}(t)}\|_{L^{2}(0, T)} \|\partial u(0, t)\|_{L^{2}(0, T)}\\ &\quad +\sup_{\|u( . , t)\|_{H_{p}^{2}}\leq 1} \|( \beta \overline{h_{2}(s)} - i \alpha \overline{h_{1}(s)} ) \|_{L^{2}(0, T)} \|u(0, t)\|_{L^{2}(0, T)} \\ & \leq c \big( \|y_{0}\|_{(H_{p}^{2})'} + \|h_{0}\|_{L^{2}(0, T)} + \|h_{1}\|_{L^{2}(0, T)} + \|h_{2}\|_{L^{2}(0, T)} \big) \end{align*} % \label{e308} where $c$ is a positive constant which does not depend on $t$ or on $y_{0}$, $h_{0}$, $h_{1}$, $h_{2}$. Since \begin{equation} \label{e309} y\in C([0, T]: L^{2}(0, 2 \pi))\subseteq C([0, T]: (H_{p}^{2})') \end{equation} for $y\in H_{p}^{3}$ and $(h_{0}, h_{1}, h_{2})\in [C_{0}^{2}([0, T])]^{3}$, and since $H_{p}^{3}$ is dense in $L^{2}(0, T)$ and $C_{0}^{2}([0, T])$ is dense in $L^{2}(0, L)$, it follows from \eqref{e309} that $y\in C([0, T]: (H_{p}^{2})')$. %\end{proof} \begin{lemma}[Observability result] \label{lem3.3} Let $T>0$. There exist positive numbers $C_{1}^{T}$, $C_{2}^{T}$ such that for every $u_{T}\in H_{p}^{2}$ \begin{equation} \label{e310} \begin{aligned} C_{1}^{T} \|u_{T}\|_{H_{p}^{2}(0, 2 \pi)}^{2} & \leq \|u(0, . )\|_{L^{2}(0, T)}^{2} + \|\partial u(0, . )\|_{L^{2}(0, T)}^{2} + \|\partial^{2}u(0, . )\|_{L^{2}(0, T)}^{2} \\ & \leq C_{2}^{T} \|u_{T}\|_{H_{p}^{2}(0, 2 \pi)}^{2} \end{aligned} \end{equation} \end{lemma} \begin{proof} In $L^{2}(0, T)$ we have that \begin{gather*} u(0, t) = \sum_{n\in \mathbb{Z}}c_{n} e^{i (\beta n^{3} - \alpha n^{2}) (t - T)}\\ \partial u(0, t) = \sum_{n\in \mathbb{Z}}i n c_{n} e^{i (\beta n^{3} - \alpha n^{2}) (t - T)}\\ \partial^{2}u(0, t) = \sum_{n\in \mathbb{Z}}- n^{2} c_{n} e^{i (\beta n^{3} - \alpha n^{2}) (t - T)}. \end{gather*} Hence \begin{equation} \label{e311} \begin{aligned} &\|u(0, t)\|_{L^{2}(0, T)}^{2} + \|\partial u(0, t)\|_{L^{2}(0, T)}^{2} + \|\partial^{2}u(0, t)\|_{L^{2}(0, T)}^{2} \\ & \leq \big([ \frac{T}{2 \pi} ] + 1 \big) \sum_{n\in \mathbb{Z}}( 1 + n^{2} + n^{4} )|c_{n}|^{2} \\ & \leq C_{2}^{T}\;\|u_{T}\|_{H_{p}^{2}(0, 2 \pi)}^{2}\quad \mbox{for } u_{T}\in H_{p}^{2} \end{aligned} \end{equation} where $C_{2}^{T}=( [\frac{T}{2 \pi}]+ 1 )$. To prove the left inequality we first take $T'\in (0, T)$ and $\gamma >2 \pi /T'$. Let $N\in \mathbb{N}^{*}$ be such that \[ n\in \mathbb{Z},\quad |n|\geq N\Rightarrow [\beta (n + 1)^{5} - \alpha (n + 1)^{3}] - [\beta n^{5} - \alpha n^{3}]\geq \gamma . \] By Ingham's inequality \cite{i1} there exists $c^{T'}>0$ such that for all sequences $(a_{n})_{n\in \mathbb{Z}}$ in $l^{2}(\mathbb{Z})$, \begin{equation} \label{e312} \sum_{|n|\geq N}|a_{n}|^{2}\leq c^{T'}\int_{0}^{T'}\Big|\sum_{|n|\geq N}a_{n} e^{i (\beta n^{3} - \alpha n^{2}) (t - T)}\Big|^{2}\,dt. \end{equation} Let $\mathcal{Z}_{n}=\mathop{\rm Span} (e^{i n x})$ for $n\in \mathbb{Z}$ and $\mathcal{Z}={\oplus_{n\in \mathbb{Z}}\mathcal{Z}_{n}\subseteq H_{p}^{2}}$. We define a semi-norm $p$ in $\mathcal{Z}$ by: $\forall u\in \mathcal{Z}$, \begin{equation} \label{e313} \begin{aligned} p(u) & = ( |u(0)|^{2} + |\partial u(0)|^{2} + |\partial^{2}u(0)|^{2} )^{1/2} \\ & = \Big(\big|\sum_{n\in \mathbb{Z}}\widehat{u}(n)\big|^{2} + \big|\sum_{n\in \mathbb{Z}}i n \widehat{u}(n)\big|^{2} + \big|\sum_{n\in \mathbb{Z}}-n^{2}\widehat{u}(n)\big|^{2} \Big)^{1/2} \end{aligned} \end{equation} (For $u\in \mathcal{Z}$, $\widehat{u}(n)=0$ for $|n|$ large enough). Let $u_{T}\in \mathcal{Z}\cap ( {\oplus_{|n|T'$, it follows from \eqref{e311}, \eqref{e314} and a result by Komornik (see \cite{k2}) that there exists a constant $C_{1}^{T}>0$ such that for all $u_{T}$ in $\mathcal{Z}$, \begin{equation} \label{e315} \begin{aligned} C_{1}^{T} \|u_{T}\|_{H_{p}^{2}(0, 2 \pi)}^{2} & \leq \int_{0}^{T}[ p(u( . , t)) ]^{2}\,dt \\ & = \|u(0, . )\|_{L^{2}(0, T)}^{2} + \|\partial u(0, . )\|_{L^{2}(0, T)}^{2} + \|\partial^{2}u(0, . )\|_{L^{2}(0, T)}^{2} \end{aligned} \end{equation} and the result follows. \end{proof} We remark that by a density argument we obtain the left inequality in \eqref{e310} in the general case ($u_{T}\in H_{p}^{2}$). \begin{proof}[Proof of Theorem \ref{thm3.1}] Without loss of generality we may assume that $y_{0}=0$. In fact, if $y_{0}$, $y_{T}\in (H_{p}^{2})'$, if there exist $h_{k}\in L^{2}(0, T)$ ($k=0, 1, 2$) such that the weak solution $\widetilde{y}$ of \eqref{eP1} and $\widetilde{y}( . , 0)=0$ satisfies $\widetilde{y}( . , T)=y_{T} - S(T)y_{0}$, then $y_{T}=S(T)y_{0} + \widetilde{y}( . ,t)$ is the weak solution of \eqref{eP1} with the same control functions and its such that $y( . ,T)=y_{T}$. In what follows we assume that $y_{0}=0$. For $u_{T}\in H_{p}^{2}$ we let $\Lambda :H_{p}^{2}\mapsto (H_{p}^{2})'$, \[ u_{T}\mapsto \Lambda (u_{T})= y_{T}. \] where $y$ is the weak solution of \eqref{eP1} and $h_{k}$($k=0, 1, 2)$ are chosen the following way: \begin{gather*} \overline{h_{0}(t)}=\frac{-1}{(\beta + i \alpha)}\partial^{2}u(0, t),\quad \overline{h_{1}(t)}=\frac{1}{\beta}\partial u(0, t),\\ \overline{h_{2}(t)}=i \frac{1}{\beta} u(0, t) + i \frac{\alpha }{\beta^{2}} \partial u(0, t) \end{gather*} As above $u$ stands for the solutions of \eqref{eP} associated with $u_{T}$. Clearly $\Lambda :H_{p}^{2}\mapsto (H_{p}^{2})'$ is a conjugate linear continuous map. Moreover \begin{align*} \langle u_{T}, \Lambda(u_{T})\rangle _{H_{p}^{2}\times (H_{p}^{2})'} & = \int_{0}^{T}( |u(0, t)|^{2} + |\partial u(0, t)|^{2} + |\partial^{2}u(0, t)|^{2})\,dt\\ & \geq C_{1}^{T} \|u_{T}\|_{H_{p}^{2}(0, 2 \pi)}^{2}. \end{align*} By Lemmas \ref{lem3.2} and \ref{lem3.3} it follows from Lax-Milgram's Theorem (see \cite{y1}) that $\Lambda $ is invertible. Then the theorem follows. \end{proof} \begin{remark} \label{rmk3.5}\rm If $T=2 \pi$, Lemma \ref{lem3.3} is trivial. Indeed, for any $u_{T}\in H_{p}^{2}$, \[ \|u_{T}\|_{H_{p}^{4}(0, 2 \pi)}^{2} = \|u(0, . )\|_{L^{2}(0, 2 \pi)}^{2} + \|\partial u(0, . )\|_{L^{2}(0, 2 \pi)}^{2} + \|\partial^{2}u(0, . )\|_{L^{2}(0, 2 \pi)}^{2}. \] \end{remark} \section{Exact boundary controllability of the higher order linear Schr\"{o}dinger equation by means of the control $\partial y(L, t)$} We consider now, the scalar space $\mathbb{R}$. In this section, $L$ stands for some positive number. We shall prove the controllability in $L^{2}(0, L)$ of \begin{equation} \label{eR1} \begin{gathered} \partial _{t}y + \beta \partial^{3}y - i \alpha \partial^{2}y + \delta \partial y= 0 \\ y(0, t) = y(L, t) = 0 \\ \partial y(L, t)=h(t) \\ y( . , 0)= y_{0} \end{gathered} \end{equation} where $h\in L^{2}(0, T)$ stands for the control function. More precisely we shall prove that, for any $L>0$, $T>0$, $y_{0}$, $y_{T}\in L^{2}(0, L)$ there exists $h\in L^{2}(0, T)$ such that a mild solution \begin{equation} y\in C([0, T]: L^{2}(0, L))\cap L^{2}(0, T: H^{1}(0, L))\cap H^{1}(0, T: H^{-2}(0, L)) \end{equation} of \eqref{eR1} which verifies the equation \eqref{eR1} in $\mathcal{D}'(0, T: H^{-2}(0, L))$ and $y_{0}$ in $L^{2}(0, L)$ may be found such that $y( . ,T)=y_{T}$. We begin by showing the well-posedness of the initial-value homogeneous problem with $|\alpha | <3 \beta $ \begin{equation} \label{eR2} \begin{gathered} \partial _{t}y + \beta \partial^{3}y - i \alpha \partial^{2}y + \delta \partial y= 0 \\ y(0, t) = y(L, t) = 0 \\ \partial y(L, t)=0 \\ y( . , 0)= y_{0}\,. \end{gathered} \end{equation} Let $A$ denote the operator $Aw=- \beta w''' +i \alpha w'' - \delta w'$ on the (dense) domain $D(A)\subseteq L^{2}(0, L)$, defined by \[ \label{e401}D(A)=\{w\in H^{3}(0, L): w(0)=w(L)=w'(L)=0\} \] \begin{lemma} \label{lem4.1} Operator $A$ generates a strongly continuous semigroup of contractions on $L^{2}(0, L)$. \end{lemma} \begin{proof} $A$ is closed. Let $w\in D(A)$. Then \begin{align*} &\mathop{\rm Re} \langle w, Aw\rangle _{L^{2}(0, L)} \\ & = \mathop{\rm Re} \int_{0}^{L}[ - \beta w''' + i \alpha w'' - \delta w' ] w(x)\,dx \\ & = \mathop{\rm Re} \Big[- \beta \int_{0}^{L}w'''(x) w(x)\,dx + i \alpha \int_{0}^{L}w'' w(x)\,dx - \delta \int_{0}^{L}w'(x) w(x)\,dx\Big]. \end{align*} Each term is treated separately. Integrating by parts, \begin{gather*} \int_{0}^{L}w'''(x) w(x)\,dx = \frac{1}{2} [w'(0)]^{2},\\ \int_{0}^{L}w''(x) w(x)\,dx = -\int_{0}^{L}[w'(x)]^{2}\,dx \end{gather*} Then \[ \mathop{\rm Re} \langle w, Aw\rangle _{L^{2}(0, L)} = - \frac{\beta}{2} [ w'(0) ]^{2} \leq 0\quad \mbox{if}\quad \beta > \frac{1}{3} |\alpha | \] hence, $A$ is dissipative. It can be seen that $A^{*}(w) =\beta w''' - i \alpha w'' + \delta w'$ with domain $D(A^{*})=\{w\in H^{5}(0, L): w(0)=w(L)=w'(0)=0\}$, so that \[ \mathop{\rm Re} \langle w, A^{*}w\rangle _{L^{2}(0, L)} =- \frac{\beta}{2} [ w'(L) ]^{2}\leq 0 , \quad \mbox{if } \beta > \frac{1}{3} |\alpha | \] and $A^{*}$ is dissipative. Hence, by the Lumer-Phillips Theorem, $A$ is the infinitesimal generator of a $C_{0}$ semigroup of contractions on $L^{2}(0, L)$. The result follows. \end{proof} We denote by $(S(t))_{t\geq 0}$ the semi-group of contractions associated with $A$, and we let $\mathbb{H}$ denote the Banach space $C([0, T]: L^{2}(0, L))\cap L^{2}([0, T]: H^{1}(0, L))$ endowed with the norm \begin{equation} \label{e402} \begin{aligned} \|y\|_{\mathbb{H}} & = \sup_{t\in[0, T]}\|y( . , t)\|_{L^{2}(0, L)} + \Big(\int_{0}^{T}\|y( . , t)\|_{H^{1}(0, L)}^{2}\,dt\Big)^{1/2}\\ & = \sup_{t\in[0, T]}\|y( . , t)\|_{L^{2}(0, L)} + \|y( . , t)\|_{L^{2}(0, T: H^{1}(0, L))}. \end{aligned} \end{equation} Using the multiplier method, we get useful estimates for the mild solutions of \eqref{eR2}. \begin{lemma} \label{lem4.2} Let $|\alpha |<3 \beta $. Then \begin{itemize} \item[(1)] The map $y_{0}\in L^{2}(0, L)\mapsto S( \cdot )y_{0}\in \mathbb{H}$ is continuous. \item[(2)] For $y_{0}\in L^{2}(0, L)$, $\partial y(0, . )$ makes sense in $L^{2}(0, L)$, and for all $y_{0}\in L^{2}(0, L)$, \begin{gather} \label{e403}\|\partial y( . , t)\|_{L^{2}(0, T)} \leq \|y_{0}\|_{L^{2}(0, L)}\\ \label{e404}\|y_{0}\|_{L^{2}(0, L)}^{2} \leq \frac{1}{T} \|S( \cdot )y_{0}\|_{L^{2}((0, T)\times (0, L))}^{2} + \|\partial y(0, . )\|_{L^{2}(0, T)}^{2} \end{gather} \end{itemize} \end{lemma} \begin{proof} (1) For $y_{0}\in L^{2}(0, L)$ we write $y$ the mild solution $S( \cdot )y_{0} $ of $ (R_{2})$. By Lemma \ref{lem4.1}, $ y\in C([0, T]: L^{2}(0, L))$ and \begin{equation} \label{e405}\|y\|_{C([0, T]: L^{2}(0, L))}\leq \|y_{0}\|_{L^{2}(0, L)} \end{equation} To see that $y\in L^{2}(0, T: H^{2}(0, L))$ we first assume that $y\in D(A)$. Let $\xi = \xi(x, t)\in C^{\infty}([0, T]\times [0, L])$. Then, multiplying the equation \eqref{eR2} by $i \xi y$ we have \begin{gather*} i \xi \overline{y} \partial _{t}y + i \xi \overline{y} \partial ^{3}y + \alpha \xi \overline{y} \partial ^{2}y + i \delta \xi \overline{y} \partial y =0\\ - i \xi y \partial_{t}\overline{y} - i \xi y \partial ^{3}\overline{y} + \alpha \xi y \partial ^{2}\overline{y} - i \delta \xi y \partial \overline{y} =0 \end{gather*} (applying conjugates). Subtracting, integrating over $x\in (0, L)$ and using straightforward calculus, we obtain \begin{gather*} i \partial_{t}\int_{0}^{L}\xi |y|^{2}\,dx - i\int_{0}^{L}\partial_{t}\xi |y|^{2}\,dx + i \beta \int_{0}^{L}\xi \overline{y} \partial ^{3}y\,dx + i \beta \int_{0}^{L}\xi y \partial ^{3}\overline{y}\,dx \\ + \alpha \int_{0}^{L}\xi \overline{y} \partial ^{2}y\,dx - \alpha \int_{0}^{L}\xi y \partial ^{2}\overline{y}\,dx - i \delta \int_{0}^{L}\partial \xi |y|^{2}\,dx =0. \end{gather*} Each term is treated separately. Integrating by parts \begin{gather*} \begin{aligned} \int_{0}^{L}\xi \overline{y} \partial ^{3}y\,dx & = \int_{0}^{L}\partial ^{2}\xi \overline{y} \partial y\,dx + 2\int_{0}^{L}\partial \xi |\partial y|^{2}\,dx - \xi(0, t) |\partial y(0, t)|^{2} \\ &\quad + \int_{0}^{L}\xi \partial y \partial ^{2}\overline{y}\,dx \end{aligned}\\ \int_{0}^{L}\xi y \partial ^{3}\overline{y}\,dx = \int_{0}^{L}\partial ^{2}\xi y \partial \overline{y}\,dx + \int_{0}^{L}\partial \xi |\partial y|^{2}\,dx - \int_{0}^{L}\xi \partial y \partial ^{2}\overline{y}\,dx,\\ \int_{0}^{L}\xi \overline{y} \partial ^{2}y\,dx = -\int_{0}^{L}\partial \xi \overline{y} \partial y\,dx - \int_{0}^{L}\xi |\partial y|^{2}\,dx,\\ \int_{0}^{L}\xi y \partial ^{2}\overline{y}\,dx = -\int_{0}^{L}\partial \xi y \partial \overline{y}\,dx - \int_{0}^{L}\xi |\partial y|^{2}\,dx\,. \end{gather*} Then \begin{align*} &i \partial_{t}\int_{0}^{L}\xi |y|^{2}\,dx - i\int_{0}^{L}\partial_{t}\xi |y|^{2}\,dx + i \beta \int_{0}^{L}\partial ^{2}\xi \overline{y} \partial y\,dx + 2 i \beta \int_{0}^{L}\partial \xi |\partial y|^{2}\,dx\\ & - i \beta \xi(0, t) |\partial y(0, t)|^{2} + i \beta \int_{0}^{L}\xi \partial y \partial ^{2}\overline{y}\,dx + i \beta \int_{0}^{L}\partial ^{2}\xi y \partial \overline{y}\,dx + i \beta \int_{0}^{L}\partial \xi |\partial y|^{2}\,dx\\ &- i \beta \int_{0}^{L}\xi \partial y \partial^{2}\overline{y}\,dx - \alpha \int_{0}^{L}\partial \xi \overline{y} \partial y\,dx - \alpha \int_{0}^{L}\xi |\partial y|^{2}\,dx + \int_{0}^{L}\partial \xi y \partial \overline{y}\,dx \\ &+ \int_{0}^{L}\xi |\partial y|^{2}\,dx - i \delta \int_{0}^{L}\partial \xi |y|^{2}\,dx =0\,. \end{align*} Hence, \begin{align*} & i \partial_{t}\int_{0}^{L}\xi |y|^{2}\,dx - i\int_{0}^{L}\partial_{t}\xi |y|^{2}\,dx + i \beta \int_{0}^{L}\partial ^{2}\xi \partial (|y|^{2})\,dx + 3 i \beta \int_{0}^{L}\partial \xi |\partial y|^{2}\,dx\\ & - i \beta \xi(0, t) |\partial y(0, t)|^{2} - 2 i \alpha \mathop{\rm Im}\int_{0}^{L}\partial \xi \overline{y} \partial y\,dx - i \delta \int_{0}^{L}\partial \xi |y|^{2}\,dx =0\,. \end{align*} Thus \begin{align*} & \partial_{t}\int_{0}^{L}\xi |y|^{2}\,dx - \int_{0}^{L}\partial_{t}\xi |y|^{2}\,dx - \beta \int_{0}^{L}\partial ^{3}\xi |y|^{2}\,dx + 3 \beta \int_{0}^{L}\partial \xi |\partial y|^{2}\,dx\\ & - \beta \xi(0, t) |\partial y(0, t)|^{2} - \delta \int_{0}^{L}\partial \xi |y|^{2}\,dx \\ &= 2 \alpha \mathop{\rm Im} \int_{0}^{L}\partial \xi \overline{y} \partial y\,dx \\ & \leq |\alpha |\int_{0}^{L}\partial \xi |y|^{2}\,dx + |\alpha |\int_{0}^{L}\partial \xi |\partial y|^{2}\,dx, \end{align*} where \begin{equation} \label{e406} \begin{aligned} &\partial_{t}\int_{0}^{L}\xi |y|^{2}\,dx + \int_{0}^{L}[ 3 \beta - |\alpha | ] \partial \xi |\partial y|^{2}\,dx - \int_{0}^{L}\partial_{t}\xi |y|^{2}\,dx - \beta \int_{0}^{L}\partial ^{3}\xi |y|^{2}\,dx \\ & - \beta \xi(0, t) |\partial y(0, t)|^{2} - \delta \int_{0}^{L}\partial \xi |y|^{2}\,dx - |\alpha |\int_{0}^{L}\partial \xi |y|^{2}\,dx \leq 0. \end{aligned} \end{equation} Choosing $\xi(x, t) = x$ leads to \[ \partial_{t}\int_{0}^{L}x |y|^{2}\,dx + \int_{0}^{L}[ 3 \beta - |\alpha | ] |\partial y|^{2}\,dx - ( \delta + |\alpha | )\int_{0}^{L}|y|^{2}\,dx \leq 0. \] Integrating over $t\in [0, T]$ we obtain \begin{align*} &\int_{0}^{L}x |y|^{2}\,dx + [ 3 \beta - |\alpha | ] \int_{0}^{T}\int_{0}^{L}|\partial y|^{2}\,dx\,dt \\ &\leq ( \delta +|\alpha | ) \int_{0}^{T}\int_{0}^{L}|y|^{2}\,dx\,dt + \int_{0}^{L}x |y_{0}|^{2}\,dx\\ & \leq ( \delta + |\alpha | )\int_{0}^{T}\int_{0}^{L}|y|^{2}\,dx\,dt + L\int_{0}^{L}|y_{0}|^{2}\,dx. \end{align*} Using that $|\alpha |<3 \beta, $ the second and the third terms in the left hand on the above equation are positive, thus we obtain \begin{align*} &[ 3 \beta - |\alpha | ] \|\partial y\|_{L^{2}(0, T: L^{2}(0, L))}^{2}\\ &\leq \big[ ( |\delta| + |\alpha | ) \|y\|_{L^{2}(0, T: L^{2}(0, L))}^{2} + L \|y_{0}\|_{L^{2}(0, L)}^{2} \big], \end{align*} where \begin{equation}\label{e407} \begin{aligned} &\|\partial y\|_{L^{2}(0, T: L^{2}(0, L))}^{2}\\ &\leq \frac{1}{( 3 \beta - |\alpha | )}\big[( |\delta| + |\alpha | ) \|y\|_{L^{2}(0, T: L^{2}(0, L))}^{2} + L \|y_{0}\|_{L^{2}(0, L)}^{2} \big]\,. \end{aligned} \end{equation} Then, using \eqref{e405}, \begin{equation} \label{e408} \begin{aligned} \|y\|_{L^{2}(0, T: H^{1}(0, L))} & \leq \Big[ T + \frac{1}{( 3 \beta - |\alpha | )} [( |\delta| + |\alpha | ) T + L ]\Big]^{1/2} \|y_{0}\|_{L^{2}(0, L)} \\ & \leq \big[ \frac{1}{( 3 \beta - |\alpha | )}[ (|\delta| + 3 \beta ) T + L ]\big]^{1/2} \|y_{0}\|_{L^{2}(0, L)} \end{aligned} \end{equation} By the density of $D(A)$ in $L^{2}(0, L)$ the result extends to arbitrary $y_{0}\in L^{2}(0, L)$. \smallskip \noindent(2) We also assume $y_{0}\in D(A)$ and taking $\xi(x, t)=1$ in \eqref{e406}, we get \begin{equation} \label{e409}\beta |\partial y(0, t)|^{2} \leq \int_{0}^{L}|y_{0}|^{2}\,dx - \int_{0}^{L}|y|^{2}\,dx\leq \int_{0}^{L}|y_{0}|^{2}\,dx. \end{equation} On the other hand the choice $\xi(x, t)=T - t$ yields \begin{equation} \label{e410}\partial_{t}\int_{0}^{L}(T - t) |y|^{2}\,dx + \int_{0}^{L}|y|^{2}\,dx - \beta (T - t) |\partial y(0, t)|^{2} \leq 0. \end{equation} Integrating over $t\in [0, T]$ we have \[ - T\int_{0}^{L} |y_{0}|^{2}\,dx + \int_{0}^{L}\int_{0}^{L}|y|^{2}\,dx\,dt - \beta \int_{0}^{L}(T - t) |\partial y(0, t)|^{2}\,dt \leq 0. \] Hence \begin{equation} \label{e411} \begin{aligned} \int_{0}^{L} |y_{0}|^{2}\,dx & \leq \frac{1}{T}\int_{0}^{L}\int_{0}^{L}|y|^{2}\,dx\,dt - \frac{\beta}{T}\int_{0}^{L}(T - t) |\partial y(0, t)|^{2}\,dt \\ & \leq \frac{1}{T}\int_{0}^{L}\int_{0}^{L}|y|^{2}\,dx\,dt + \beta \int_{0}^{L}|\partial y(0, t)|^{2}\,dt. \end{aligned} \end{equation} By \eqref{e410} there exists a unique continuous (linear) extension of the map $y_{0}\in D(A)\mapsto \partial y(0, . )\in L^{2}(0, T)$ to the whole space $L^{2}(0, L)$. In what follows we also will denote by $\partial y(0, . )$ the value of this map at any $y_{0}\in L^{2}(0, L)$. It is trivial to see that \eqref{e410} and \eqref{e411} are true for any $y_{0}\in L^{2}(0, L)$. \end{proof} \begin{lemma}[Observability result] \label{lem4.3} Let $|\alpha |<3 \beta $, $\delta>0$ and \begin{equation} \mathcal{N}=\Big\{2 \pi \beta \sqrt{\frac{k^{2} + k l + l^{2})} {3 \beta \delta + \alpha^{2}}} : k, l\in \mathbb{N}^{*}\Big\}. \end{equation} Then, for all $L\in (0, + \infty )\backslash \mathcal{N}$, for all $T>0$, there exists $C=C(L, T)>0$ such that for all $y_{0}\in L^{2}(0, L)$, \begin{equation} \label{e412}\|y_{0}\|_{L^{2}(0, L)}\leq C\;\|\partial y(0, . )\|_{L^{2}(0, T)}. \end{equation} \end{lemma} \begin{proof} (By contradiction) If the statement is false, there exists a sequence $(y_{0}^{n})_{n\geq 0}\in L^{2}(0, L)$ such that $\|y_{0}^{n}\|_{L^{2}(0, L)}=1$ for any $n$, but $\|\partial y^{n}(0, . )\|_{L^{2}(0, T)}\to 0$ as $n\to \infty$, where $y^{n}=S( \cdot )y_{0}^{n}$. Using \eqref{e409} have that $\{y^{n}\}$ is bounded in $L^{2}(0, T: H^{2}(0, L))\;( \hookrightarrow L^{2}(0, T: H^{1}(0, L)\;)$. On the other hand, \begin{equation} \partial_{t}y^{n}=- (\beta \partial ^{3}y^{n} - i \alpha \partial ^{2}y^{n} + \delta \partial y^{n})\quad \mbox{is bounded in } L^{2}(0, T: H^{-2}(0, L)). \end{equation} But $H^{1}(0, L)\stackrel{c}\hookrightarrow L^{2}(0, L)\hookrightarrow H^{-2}(0, L)$, then from Lions-Aubin's Theorem (see \cite{l7}), the set $\{y^{n}\}$ is relatively compact in $L^{2}(0, T: L^{2}(0, L))$. Without loss of generality, we may assume that the sequence $\{y^{n}\}$ is convergent in $L^{2}(0, T: L^{2}(0, L))$. We infer from \eqref{e404} that $\{y_{0}^{n}\}$ is a Cauchy sequence in $L^{2}(0, T)$. Let $ y_{0}={\lim_{n\to 0}y_{0}^{n}} $ and $ y=S( \cdot )y_{0}$. By Lemma \ref{lem4.2}, $\partial y^{n}(0, . )\to \partial y(0, . ) $ in $ L^{2}(0, T)$. Thus, $\|y_{0}\|_{L^{2}(0, L)}=1$ and $\partial y(0, . )=0$, but such function does not exist because of the following lemma. \end{proof} \begin{lemma} \label{lem4.4} For $T>0$ let $\mathcal{F}_{T}$ denote the space of the initial states $y_{0}\in L^{2}(0, L)$ such that the mild solution $y=S( \cdot )y_{0}$ of \eqref{eR2} satisfies $\partial y(0, . )=0$ in $L^{2}(0, T)$. Then, for $L\in (0, \infty)\backslash \mathcal{N}$, $\mathcal{F}_{T}=\{0\}$, for all $T>0$. \end{lemma} \begin{proof} It is obvious that if $T0$, $\mathcal{F}_{T}$ is a finite-dimensional vector space. In fact, if $\{y_{0}^{n}\}$ is a sequence in the unit ball $\mathbb{B}_{\mathcal{F}_{T}} =\{y\in \mathcal{F}_{T}: \|y\|_{L^{2}(0, L)}\leq 1\}$ the same argument as above shows that there exist a convergent subsequence. Since the unit ball is compact, by the Riesz Theorem (see \cite{r2}) $\mathcal{F}_{T}$ is finite dimensional and the claim follows. Let $T'>0$ be given. To prove that $\mathcal{F}_{T'}=\{0\}$, it is sufficient to find $00$ such that $T0$ \begin{equation} \label{eR4} \begin{gathered} \partial _{t}u + \beta \partial^{3}u - i \alpha \partial^{2}u + \delta \partial u=0, \\ u(0, t) = u(L, t) = 0, \\ \partial u(0, t)=0 ,\\ u(T, 0)= u_{T}(x)\,. \end{gathered} \end{equation} The change of variables $\tau T - t$ and $\zeta = L - x$ transform \eqref{eR4} into \eqref{eR2} and vice-versa. Using Lemmas \ref{lem4.1}, \ref{lem4.2} and \ref{lem4.3}, we readily get the following result. \begin{lemma}]Observability result] \label{lem4.7} Let $L, T>0$, $|\alpha |<3 \beta$ and $\delta>0$. For any $u_{T}\in L^{2}(0, L)$ the mild solution of \eqref{eR4} belongs to $\mathbb{H}$, the function $\partial u(L, . )$ makes sense in $L^{2}(0, T)$. If moreover, $L\notin \mathcal{N}$, there exists a constant $C=C(L, T)>0$ such that for any $u_{T}\in L^{2}(0, L)$ we have that \begin{equation} \label{e425} \|\partial u(L, . )\|_{L^{2}(0, T)} \leq \|u_{T}\|_{L^{2}(0, T)}\leq C\;\|\partial u(L, . )\|_{L^{2}(0, T)}. \end{equation} \end{lemma} It remains to apply the Hilbert uniqueness method. \begin{theorem} \label{thm4.1} Let $|\alpha |<3 \beta $, $\delta>0$ and \[ \mathcal{N}=\big\{2 \pi \beta \sqrt{\frac{k^{2} + k l + l^{2})}{3 \beta \delta + \alpha^{2}}} : k, l\in \mathbb{N}^{*}\big\}. \] Then, for any $T>0$ and $L\in (0, + \infty )\backslash \mathcal{N}$, and for any $y_{0}$, $y_{T}\in L^{2}(0, L)$, there exists $h\in L^{2}(0, T)$ such that the mild solution $y\in C([0, T]: L^{2}(0, L))\cap L^{2}(0, T: H^{1}(0, L))$ of \begin{gather} \label{e426} \partial_{t}y + \beta \partial^{3}y - i \alpha \partial^{2}y + \delta \partial y = 0\\ \label{e427} y(0, t)=y(L, T)=0\\ \label{e428} \partial y(L, t)=h(t)\\ \label{e429} y(x, 0)=y_{0}(x) \end{gather} satisfies $y( . , T)=y_{T}$. \end{theorem} \begin{proof} By Remark \ref{rmk4.3} (b) we may assume, without loss of generality, that $y_{0}=0$. (see the proof of Theorem \ref{thm3.1}) Let $(u_{T}, h)\in C_{c}^{\infty}(0, L)\times C_{c}^{\infty}(0, L)$, let $u$ (resp. $y$) be the classical solution of \eqref{e426}-\eqref{e429} (resp. $(R_{1}))$. Multiplying \eqref{e426} by $u$ and integrating over $x\in [0, L]$ we have \begin{equation} \label{e430}\int_{0}^{L}u \partial_{t}y\,dx + \beta \int_{0}^{L}u \partial^{3}y\,dx - i \alpha \int_{0}^{L}u \partial^{2}y\,dx + \delta \int_{0}^{L}u \partial y\,dx = 0 \end{equation} Each term is treated separately. Integrating by parts, \begin{gather*} \int_{0}^{L}u \partial _{t}y\,dx = \partial_{t}\int_{0}^{L}u y\,dx - \int_{0}^{L}\partial _{t}u y\,dx,\\ \int_{0}^{L}u \partial^{3}y\,dx = - \partial u(L, t) h(t) - \int_{0}^{L}\partial^{3}u y\,dx,\\ \int_{0}^{L}u \partial^{2}y\,dx = \int_{0}^{L}\partial^{2}u y\,dx,\\ \int_{0}^{L}u \partial y\,dx = -\int_{0}^{L}\partial u y\,dx\,. \end{gather*} Then in \eqref{e430} we obtain \begin{align*} &\partial_{t}\int_{0}^{L}u y\,dx - \int_{0}^{L}\partial _{t}u y\,dx - \beta \partial u(L, t) h(t) - \beta \int_{0}^{L}\partial^{3}u y\,dx \\ &- i \alpha \int_{0}^{L}\partial^{2}u y\,dx - \delta \int_{0}^{L}\partial u y\,dx = 0 \end{align*} where \begin{equation} \label{e431} \begin{aligned} \partial_{t}\int_{0}^{L}u y\,dx - \beta \partial u(L, t) h(t) &= - 2 i \alpha\int_{0}^{L}\partial u \partial y\,dx\\ &\leq |\alpha |\int_{0}^{L}|\partial u|^{2}\,dx + |\alpha |\int_{0}^{L}|\partial y|^{2}\,dx\,. \end{aligned} \end{equation} Integrating over $t\in [0, T]$ and using that $y_{0}=0$ we obtain \begin{equation} \label{e432} \begin{aligned} &\int_{0}^{L}u_{T}(x) y(x, T)\,dx \\ & \leq \beta \int_{0}^{T}\partial u(L, t) h(t)\,dt + |\alpha |\int_{0}^{T}\int_{0}^{L}|\partial u|^{2}\,dx\,dt +|\alpha |\int_{0}^{T}\int_{0}^{L}|\partial y|^{2}\,dx\,dt \end{aligned} \end{equation} By a density argument we see that \eqref{e432} holds for $u_{T}\in L^{2}(0, L)$ and $h\in L^{2}(0, T)$. Let $\Lambda $ denote the linear continuous map $\Lambda :L^{2}(0, L)\mapsto L^{2}(0, L)$ with $u_{T}\mapsto \Lambda (u_{T}) = y( . ,T)$ and $y$ standing for the solution of \eqref{eR1} associated with the data $h( . )=\partial u(L, . )\in L^{2}(0, T)$. It follows \eqref{e432} and by Lemma \ref{lem4.7} that \[ \langle \Lambda (u_{T}), u_{T}\rangle _{L^{2}(0, L)}=\|\partial u(L, . )\|_{L^{2}(0, T)}^{2}\geq C^{-2}\;\|u_{T}\|_{L^{2}(0, L)}^{2}. \] Therefore, by Lax-Milgram's theorem (see \cite{y1}), $\Lambda $ is invertible. The proof is complete. \end{proof} \begin{remark} \label{rmk4.4} When $y_{0}=0$, the Hilbert uniqueness method yields $u$, a linear continuous selection of the control, namely the map $\Gamma_{0}:L^{2}(0, L)\mapsto L^{2}(0, T)$ with $y_{T}\mapsto \Gamma_{0}(y_{T}) =\partial u(L, . )$ where $u$ denotes the solution of \eqref{eR4} associated with $u_{T}=\Lambda ^{-1}(y_{T})$. \end{remark} \section{Exact boundary controllability for a higher order nonlinear Schr\"{o}dinger equation with constant coefficients on a bounded domain} In this section we prove that the following boundary-control system (for $|\alpha |<3 \beta $ and $\delta>0$) \begin{equation} \label{eQ11} \begin{gathered} \partial _{t}y + \beta \partial^{3}y - i \alpha \partial^{2}y - i |y|^{2} y + \delta \partial y= 0 \\ y(0, t) = y(L, t) = 0\\ \partial y(L, t)=h(t),\quad h\in L^{2}(0, T)\\ y( . , 0)= y_{0} \end{gathered} \end{equation} is exactly controllable in a neighborhood of the null state. More precisely we show that for any $L>0$ and $T>0$ there exists a radius $r_{0}>0$ such that for every $y_{0}, y_{T}\in L^{2}(0, L)$ with $\|y_{0}\|_{L^{2}(0, L)}0$ such that for for all $f\in C^{1}([0, T]: L^{2}(0, L))$, \[ \|\partial y_{2}\|_{L^{2}((0, T)\times (0, L))} \leq c_{2} \|f\|_{L^{1}(0, T: L^{2}(0, L))}. \] In fact, multiplying \eqref{eQ33} by $i x \overline{y}_{2}$, \begin{gather*} i x \overline{y}_{2} \partial_{t}y_{2} + i \beta x \overline{y}_{2} \partial^{3}y_{2} + \alpha x \overline{y}_{2} \partial^{2}y_{2} + i \delta x \overline{y}_{2} \partial y_{2} = i x \overline{y}_{2} f ,\\ - i x y_{2} \partial_{t}\overline{y}_{2} - i \beta x y_{2} \partial^{3}\overline{y}_{2} + \alpha x y_{2} \partial^{2}\overline{y}_{2} - i \delta x y_{2} \partial \overline{y}_{2} = - i x y_{2} f \end{gather*} (applying conjugate). Subtracting and integrating over $x\in [0, L]$ we have \begin{align*} & i\partial_{t}\int_{0}^{L}x |y_{2}|^{2}\,dx + i \beta \int_{0}^{L}x \overline{y}_{2} \partial^{3}y_{2}\,dx + i \beta \int_{0}^{L}x y_{2} \partial^{3}\overline{y}_{2}\,dx \\ &+ \alpha \int_{0}^{L}x \overline{y}_{2} \partial^{2}y_{2}\,dx - \alpha \int_{0}^{L}x y_{2} \partial^{2}\overline{y}_{2}\,dx - i \delta \int_{0}^{L}|y_{2}|^{2}\,dx \\ &= 2 i \mathop{\rm Re}\int_{0}^{L}x \overline{y}_{2} f\,dx\,. \end{align*} Each term is treated separately. Integrating by parts, \begin{gather*} \int_{0}^{L}x \overline{y}_{2} \partial^{3}y_{2}\,dx = 2\int_{0}^{L}|\partial y_{2}|^{2}\,dx + \int_{0}^{L}x \partial y_{2} \partial \overline{y}_{2}\,dx\,, \\ \int_{0}^{L}x y_{2} \partial^{3}\overline{y}_{2}\,dx = \int_{0}^{L}|\partial y_{2}|^{2}\,dx - \int_{0}^{L}x \partial y_{2} \partial \overline{y}_{2}\,dx\,, \\ \int_{0}^{L}x \overline{y}_{2} \partial^{2}y_{2}\,dx = -\int_{0}^{L}\overline{y}_{2} \partial y_{2}\,dx - \int_{0}^{L}x |\partial y_{2}|^{2}\,dx \,, \\ \int_{0}^{L}x y_{2} \partial^{2}\overline{y}_{2}\,dx = -\int_{0}^{L}y_{2} \partial \overline{y}_{2}\,dx - \int_{0}^{L}x |\partial y_{2}|^{2}\,dx\,. \end{gather*} Then \begin{align*} &i\partial_{t}\int_{0}^{L}x |y_{2}|^{2}\,dx + 3 i \beta \int_{0}^{L}|\partial y_{2}|^{2}\,dx - 2 i \alpha \int_{0}^{L}\overline{y}_{2} \partial y_{2}\,dx - i \delta \int_{0}^{L}|y_{2}|^{2}\,dx \\ &= 2 i \mathop{\rm Re}\int_{0}^{L}x \overline{y}_{2} f\,dx\,, \end{align*} or \begin{align*} &\partial_{t}\int_{0}^{L}x |y_{2}|^{2}\,dx + 3 \beta \int_{0}^{L}|\partial y_{2}|^{2}\,dx - 2 \alpha \int_{0}^{L}\overline{y}_{2} [\partial y_{2}]\,dx - \delta \int_{0}^{L}|y_{2}|^{2}\,dx \\ &= 2 i Re\int_{0}^{L}x \overline{y}_{2} f\,dx\,. \end{align*} Hence \begin{align*} & \partial_{t}\int_{0}^{L}x |y_{2}|^{2}\,dx + 3 \beta \int_{0}^{L}|\partial y_{2}|^{2}\,dx \\ &= 2 i Re\int_{0}^{L}x \overline{y}_{2} f\,dx + \int_{0}^{L}|y_{2}|^{2}\,dx + 2 \alpha \int_{0}^{L}\overline{y}_{2} [\partial y_{2}]\,dx\\ & \leq 2\int_{0}^{L}x |y_{2}| |f|\,dx + \delta \int_{0}^{L}|y_{2}|^{2}\,dx + |\alpha |\int_{0}^{L}|y_{2}|^{2}\,dx + |\alpha |\int_{0}^{L}|\partial y_{2}|^{2}\,dx\,. \end{align*} Thus \begin{align*} &\partial_{t}\int_{0}^{L}x |y_{2}|^{2}\,dx + \int_{0}^{L}( 3 \beta - |\alpha | ) |\partial y_{2}|^{2}\,dx \\ &\leq 2\int_{0}^{L}x |y_{2}| |f|\,dx + ( |\delta| + |\alpha | )\int_{0}^{L}|y_{2}|^{2}\,dx\,. \end{align*} Integrating over $t\in [0, T]$ we obtain \begin{align*} & \int_{0}^{L}x |y_{2}|^{2}\,dx + \int_{0}^{T}\int_{0}^{L}( 3 \beta - |\alpha | ) |\partial y_{2}|^{2}\,dx\,dt \\ &\leq 2\int_{0}^{T}\int_{0}^{L}x |y_{2}| |f|\,dx\,dt +( |\delta| + |\alpha | )\int_{0}^{T}\int_{0}^{L}|y_{2}|^{2}\,dx\,dt + \int_{0}^{L}x |y_{0 2}|^{2}\,dx\,. \end{align*} Using \eqref{e502} the result follows. \end{proof} \begin{theorem} \label{thm5.1} Let $|\alpha |<3 \beta$, $\delta>0$, $T>0$ and $L>0$. Then, there exists $r_{0}>0$ such that for any $y_{0}$, $y_{T}\in L^{2}(0, L)$ with $\|y_{0}\|_{L^{2}(0, L)}0$ there exists $r_{0}>0$ small enough such that if $\|y_{0}\|_{L^{2}(0, L)}0$ to be chosen later. Let $\Theta :L^{2}(0, T: H^{1}(0, L))\mapsto \mathbb{H}$, defined by \[ \Theta (y) = S( \cdot )y_{0} + (\Gamma_{1}\circ \Gamma_{0}) ( y_{T} - S(T)y_{0} + \Gamma_{2}(|y|^{2} y)( . , T) ) + \Gamma_{2}(-|y|^{2} y) \] where $\Gamma_{0}$ is well-defined in Remark \ref{rmk4.4}, $\Gamma_{1}$ and $\Gamma_{2}$ are defined in this section. $\Theta $ is well-defined and continuous by Lemmas \ref{lem4.2}, \ref{lem4.6}, and Remark \ref{rmk4.4}. We have that each fixed point of $\Theta$ verifies \eqref{eQ11} in $\mathcal{D}'(0, T: H^{-2}(0, L))$ and $u( . , T)=y_{T}$. To prove the existence of a fixed-point for $\Theta$ we apply the Banach contraction fixed-point theorem to the restriction of $\Theta$ to some closed ball $\overline{\mathbb{B}}(0, R)$ in $L^{2}(0, T: H^{1}(0, L))$ ($R$ will be chosen later). We need that \begin{gather} \label{e506} \Theta (\overline{\mathbb{B}}(0, R))\subseteq \overline{\mathbb{B}}(0, R), \\ \label{e507}\exists \;C_{3}\in ]0, 1[\, \forall y, z\in \overline{\mathbb{B}}(0, R):\quad \|\Theta(y) - \Theta(z)\|\leq C_{3} \|y - z\|, \end{gather} where $\| \cdot \|$ stands for the norm $L^{2}(0, T: H^{1}(0, L))$. Let $\kappa_{1}$ (resp. $\kappa_{2}, \kappa_{2}'$) denotes the norm of $\Gamma_{1}$ (resp. $\Gamma_{2}, \Gamma_{2}$) as a map from $L^{2}(0, T)$(resp. $L^{1}(0, T: L^{2}(0, L))$ into $L^{2}(0, T: H^{1}(0, L))$ (resp. $L^{2}(0, T: H^{1}(0, L))$, $C([0, T]: L^{2}(0, L))$), and $\kappa$ denote the norm of $\Gamma_{0}$ as a map from $L^{2}(0, L)$ into $L^{2}(0, L)$. Set $\kappa_{3}=\sqrt{\frac{( \delta + 3 \beta ) T + L}{3 \beta - |\alpha |}}$. Let $y, z\in L^{2}(0, T: H^{1}(0, L))$. Assume that $\|y\|\leq R$, $\|z\|\leq R$. Then by \eqref{e408} and \eqref{e501}, \begin{equation} \label{e508} \begin{aligned} \|\Theta (y)\| & \leq \sqrt{\frac{( \delta + 3 \beta ) T + L}{3 \beta - |\alpha |}} \|y_{0}\|_{L^{2}(0, L)} \\ &\quad +\kappa_{1} \kappa (\|y_{T}\|_{L^{2}(0, L)} + \|y_{0}\|_{L^{2}(0, L)} + \kappa_{2}' C_{1} \|y\|^{2}) + \kappa_{2} C_{1} \|y\|^{2} \\ & \leq C_{1} (\kappa_{2} + \kappa \kappa_{1} \kappa_{2}') R^{2} + (2 \kappa \kappa_{1} +\kappa_{3}) r. \end{aligned} \end{equation} Hence, we have the first condition on $R$ and $r: $ \begin{equation}\label{e509} C_{1} (\kappa_{2} + \kappa \kappa_{1} \kappa_{2}') R^{2} + (2 \kappa \kappa_{1} + \kappa_{3}) r\leq R. \end{equation} Now write \begin{equation} \label{e510}\Theta(y) - \Theta(z) = \Gamma_{2} (|z|^{2} z - |y|^{2} y) + (\Gamma_{1}\circ \Gamma_{0}) (\Gamma_{2}(|y|^{2} y - |z|^{2} z)( . , T)). \end{equation} Therefore, by \eqref{e501}, \begin{equation} \label{e511}\|\Theta(y) - \Theta(z)\| = 2 C_{1} (\kappa_{2} + \kappa \kappa_{1} \kappa_{2}') R \|y - z\| \end{equation} Condition \eqref{e506} will hold provided that \begin{equation} \label{e512}2 C_{1} (\kappa_{2} + \kappa \kappa_{1} \kappa_{2}') R<1. \end{equation} Let $R$ be some positive number verifying \eqref{e511}. Then \eqref{e508} holds true if we take $r=R/\big(2 (2 \kappa \kappa_{1} + \kappa_{3})\big)$. Setting \[ r_{0}=\frac{1}{4 C_{1} (2 \kappa \kappa_{1} + \kappa_{3}) (\kappa_{2} + \kappa \kappa_{1} \kappa_{2}')}\,, \] we see that $ r\to r_{0} $ as $ R\to 1/(2 C_{1} (\kappa_{2} + \kappa \kappa_{1} \kappa_{2}'))$. It follows that if $\|y_{0}\|_{L^{2}(0, L)}L$ such that $\widetilde{L}\notin \mathcal{N}$ and to apply the theorem to the functions $\widetilde{y}_{0}, \widetilde{y}_{T}\in L^{2}(0, \widetilde{L})$, where $\widetilde{y}_{0}, \widetilde{y}_{T}$ denote the prolongations by zero of the given states $\widetilde{y}_{0}, \widetilde{y}_{T}\in L^{2}(0, L)$, and then to restrict the solution $\widetilde{y}$ to the domain $(0, T)\times (0, L)$. The proof follows. \end{proof} \subsection*{Acknowledgement} The authors want to thank Prof. Carlos Picarte (Universidad del B\'\i o-B\'\i o) for his help in the typesetting the original manuscript. \begin{thebibliography}{00} \bibitem{b1} C. Bardos, G. Lebeau and J. Rauch, {\it Sharp sufficient conditions for the observation, control and stabilization of waves from the boundary,} SIAM J. Control Optim., 30(1992) 1024-1065. \bibitem{b2} D. J. Benney, {\it Long waves on liquid films,} J. Math. Phys, 45(1966) 150-155 \bibitem{b3} H. A. Biagioni and F. Linares, {\it On the Benney-Lin and Kawahara equations}, Journal of mathematical Analysis and applications, 211(1997) 131-152. \bibitem{b4} L Bona and R. Winter, {\it The Korteweg -de Vries equation posed in a quarter-plane,} SIAM J. Math Anal., 14(1983) 1056-1106. \bibitem{b5} H. Brezis, {\it Analyse Fonctionnelle, Théorie et applications.} Masson, 1993. \bibitem{c1} X. Carvajal, {\it Local well-posedness for a higher order nonlinear Schr\"{o}dinger equation in Sobolev space of negative indices,} EJDE, 204(2004) 1-10. \bibitem{c2} X. Carvajal and F. Linares, {\it A higher order nonlinear Schr\"{o}dinger equation with variable coefficients,} Preprint. \bibitem{c3} J. M. Coron, {\it Contrólabilité exacte frontiére de l'equation d'Euler des fluides parfaits incompressibles bidimensionnels,} C. R. Acad. Sci. Paris, t. 317, Série I(1993) 271-276. \bibitem{f1} A. V. Fursikov and O. Y. Imanuvilov, {\it On controllability of certain systems simulating a fluid Flow, in flow Control,} IMA, Math. Appl., vol. 68, Gunzberger ed., Springer-Verlag, New York, (1995) 148-184. \bibitem{h1} A. Hasegawa and Y. Kodama, {\it Higher order pulse propagation in a monomode dielectric guide,} IEEE, J. Quant. Elect., 23(1987) 510-524. \bibitem{h2} L. F. Ho, {\it Observabilité frontiére de l'equation des ondes,} C. R. Acad. Paris, Série 1 Math, 302(1986) 443-446. \bibitem{i1} A. E Ingham, {\it Some trigonometrical inequalities with application to the theory of series,} Math. A., 41(1936) 367-379. \bibitem{k1} Y. Kodama, {\it Optical soliton in a monomode fiber,} J. Phys. Stat., 39(1985) 597-614. \bibitem{k2} V. Komornik, {\it Exact controllability and stabilization, the multiplier method,} R.A.M. 36, John Wiley-Mason, (1994). \bibitem{k3} V. Komornik, D. L. Russel and B.-Y. Zhang, {\it Control and stabilization of the Korteweg-de vries equation on a periodic domain,} submitted to J. Differential Equations. \bibitem{k4} D. J. Korteweg and G. de Vries, {\it On the change of form of long waves advancing in a rectangular canal and on a new type of long stationary waves,} Philos. Mag., 5, 39(1895) 422-423. \bibitem{l1} C Laurey, {\it Le probléme de Cauchy pour une équation de Schr\"{o}dinger non-linéaire de ordre 3,} C. R. Acad. Sci. Paris, 315(1992) 165-168. \bibitem{l2} G. Lebeau, {\it Controle de l'equation de Schr\"{o}dinger,} J. Math. Pures Appl, 71(1992) 267-291. \bibitem{l3} S. P. Lin, {\it Finite amplitude side-band stability of a voicous film,} J. Fluid. Mech. 63(1974)417. \bibitem{l4} J. L. Lions, {\it Controlabilité exacte de systémes distribués,} C. R. Acad. Sci. Paris, 302(1986) 471-475. \bibitem{l5} J. L. Lions, {\it Controllabilité exacte, Perturbations at Stabilisation de Systémes Distribués,} Tome I, Controllabilité exacte, Collections de recherche en mathématiques appliquées, 8, Masson, Paris, 1988. \bibitem{l6} J. L. Lions, {\it Exact controllability, stabilization, and perturbations for distributed systems,} SIAM Rev., 30(1988) 1-68. \bibitem{l7} J. L. Lions, {\it Quelques méthodes de résolution des problémes aux limites non linéaires,} Etudes Mathematiques, Paris, 1969. \bibitem{m1} E. Machtyngier, {\it Exact controllability for the Schr\"{o}dinger equation,} SIAM J. Control Optim., 32(1994) 24-34. \bibitem{p1} A. Pazy, {\it Semigroups of linear operators and applications to partial differential equations,} Springer-Verlag, New York, 1983. \bibitem{r1} L. Rosier, {\it Exact boundary controllability for the Korteweg-de Vries equation on a bounded domain, control optimization and calculus of variations,} 2(1997) 33-55. \bibitem{r2} W. Rudin, {\it Functional analysis,} McGraw-Hill, 1973. \bibitem{r3} D. L. Russel and B.-Y. Zhang, {\it Controllability and stability of the third-order linear dispersion equation on a periodic domain,} SIAM J. Control Optim., 31(1993) 659-673. \bibitem{s1} J. Simon, {\it Compact sets in the space $L^{p}(0, T: \mathbb{B}),$} Annali di Matematica pura ed applicata (IV), 146(1987) 65-96. \bibitem{s2} G. Staffilani, On the generalized Korteweg-de Vries type equations, Differential and integral equations, Vol. 10, 4(1997) 777-796. \bibitem{v1} O. Vera, {\it Gain of regularity for a Korteweg-de Vries - Kawahara type equation,} EJDE, 71(2004) 1-24. \bibitem{v2} O. Vera, {\it Exact boundary controllability for the Korteweg-de Vries-Kawahara equation on a bounded domain.} Submitted. \bibitem{v3} O. Vera, {\it Smoothing Properties for a higher order nonlinear Schr\"{o}dinger equation with constant coefficients.} Submitted. \bibitem{y1} K. Yosida, {\it Functional Analysis,} Springer-Verlag, Berlin Heidelberg New York, 1978. \bibitem{y2} R. M. Young, {\it An introduction to Nonharmonic Fourier Series,} New York, Academic Press, 1980. \bibitem{z1} B.-Y. Zhang, {\it Some results for nonlinear dispersive wave equations with applications to control,} Ph. D. Thesis, University of Wisconsin, Madison, June 1990. \end{thebibliography} \end{document}