1: \documentclass[11pt]{article}
2: %\usepackage{showkeys}
3: \usepackage{mathptmx}
4: \usepackage{times}
5: \usepackage{latexsym,epsfig,bm}
6: \usepackage{amssymb}
7: \usepackage{color}
8: \usepackage{amsthm}
9: \usepackage{mathrsfs}
10:
11: \pagestyle{myheadings}
12: \markboth{}{Alexander Komech, Andrew Komech:
13: {\rm
14: %Global well-posedness for the
15: Schr\"odinger equation
16: coupled to nonlinear oscillator}}
17:
18:
19: \def\version{April 27, 2007}
20:
21: %\nonstopmode
22:
23: \textwidth 17cm
24: \textheight 22cm
25: \topmargin -1cm
26: \oddsidemargin -3mm
27: \evensidemargin -3mm
28:
29: % Euler Script symbols:
30: \DeclareSymbolFont{EUR}{U}{eur}{m}{n}
31: \SetSymbolFont{EUR}{bold}{U}{eur}{b}{n}
32: \DeclareSymbolFontAlphabet{\eur}{EUR}
33:
34: \DeclareSymbolFont{EUB}{U}{eur}{b}{n}
35: \SetSymbolFont{EUB}{bold}{U}{eur}{b}{n}
36: \DeclareSymbolFontAlphabet{\eub}{EUB}
37:
38: %\DeclareSymbolFont{EUS}{U}{eus}{m}{n}
39: %\SetSymbolFont{EUS}{bold}{U}{eus}{b}{n}
40: %\DeclareSymbolFontAlphabet{\eus}{EUS}
41:
42: % Frakfur
43: %\DeclareSymbolFont{EUF}{U}{euf}{m}{n}
44: %\SetSymbolFont{EUF}{bold}{U}{euf}{b}{n}
45: %\DeclareSymbolFontAlphabet{\euf}{EUF}
46:
47: % Blackboard
48: \DeclareSymbolFont{AMSb}{U}{msb}{m}{n}
49: \DeclareSymbolFontAlphabet{\mathbb}{AMSb}
50:
51:
52:
53: \newcommand{\re}{\ref}
54: \newcommand{\la}{\label}
55:
56: \newcommand{\ov}{\overline}
57:
58: \newcommand{\5}{\hspace{0.5mm}}
59:
60: \newcommand{\beqn}{\begin{eqnarray}}
61: \newcommand{\eeqn}{\end{eqnarray}}
62: \newcommand{\be}{\begin{equation}}
63: \newcommand{\ee}{\end{equation}}
64:
65: \newcommand{\pa}{\partial}
66: \newcommand{\fr}{\frac}
67:
68: \newcommand{\ds}{\displaystyle}
69:
70: \newcommand{\ve}{\varepsilon}
71: \newcommand{\Lam}{\Lambda}
72:
73:
74: \newcommand{\eps}{\epsilon}
75: \newcommand{\de}{\delta}
76:
77: \newcommand{\na}{\nabla}
78:
79: \newcommand\tr{\mathop{\rm Tr}}
80: \newcommand\rk{\mathop{\rm rk}}
81: \newcommand\rank{\mathop{\rm rank}}
82: \newcommand\corank{\mathop{\rm corank}}
83: \newcommand\range{\mathop{\rm Range\,}}
84: \newcommand\supp{\mathop{\rm supp}}
85: \newcommand\volume{\mathop{\rm vol}}
86: \newcommand\p{\partial}
87: \newcommand{\at}[1]{\vert\sb{\sb{#1}}}
88: \newcommand{\At}[1]{\biggr\vert\sb{\sb{#1}}}
89: \newcommand\pr{\sp{\prime}}
90: \newcommand\primed{\sp{\prime}}
91: \newcommand\ppr{\sp{\prime\prime}}
92: \newcommand\pprimed{\sp{\prime\prime}}
93: \newcommand{\vect}[1]{{\bold #1}}
94:
95: \newcommand{\ft}{\footnote}
96:
97: \def\R{{\mathbb R}}
98: \def\hvar{{\hbar}}
99: \newcommand\C{{\mathbb C}}
100: %\providecommand\C{{\mathbb C}}
101: %\renewcommand\C{{\mathbb C}}
102:
103: \newcommand\N{{\mathbb N}}\newcommand\Z{{\mathbb Z}}
104: \newcommand{\Abs}[1]{\left\vert#1\right\vert}
105: \newcommand{\abs}[1]{\vert #1 \vert}
106: \newcommand{\Norm}[1]{\left\Vert #1 \right\Vert}
107: \newcommand{\norm}[1]{\Vert #1 \Vert}
108: \newcommand{\bnorm}[1]{\!\left\bracevert\! #1 \!\right\bracevert\!}
109: \newcommand{\tangent}[1]{T #1}
110: \newcommand{\cotangent}[1]{T\sp\ast{#1}}
111: \newcommand{\normal}[1]{N #1}
112: \newcommand{\conormal}[1]{N\sp\ast{#1}}
113: \newcommand\Const{{C{\hskip -1.5pt}onst}\,}
114: \newcommand\sothat{{\rm :}\ }
115: \newcommand\sgn{\mathop{\rm sgn}}
116: \providecommand\longlongrightarrow{-\!\!\!-\!\!\!-\!\!\!\longrightarrow}
117:
118: \newcommand\nut{\sb{\!o}}
119: \newcommand\loc{\sb{\rm loc}}
120: \newcommand\comp{\sb{\rm comp}}
121: \newcommand\mn{\sb{\!\it min}}
122: \newcommand\mx{\sb{\!\it max}}
123: \def\pprime{{\prime\prime}}
124:
125: \DeclareMathSymbol{\varGamma}{\mathord}{letters}{"00}
126: \DeclareMathSymbol{\varDelta}{\mathord}{letters}{"01}
127: \DeclareMathSymbol{\varTheta}{\mathord}{letters}{"02}
128: \DeclareMathSymbol{\varLambda}{\mathord}{letters}{"03}
129: \DeclareMathSymbol{\varXi}{\mathord}{letters}{"04}
130: \DeclareMathSymbol{\varPi}{\mathord}{letters}{"05}
131: \DeclareMathSymbol{\varSigma}{\mathord}{letters}{"06}
132: \DeclareMathSymbol{\varUpsilon}{\mathord}{letters}{"07}
133: \DeclareMathSymbol{\varPhi}{\mathord}{letters}{"08}
134: \DeclareMathSymbol{\varPsi}{\mathord}{letters}{"09}
135: \DeclareMathSymbol{\varOmega}{\mathord}{letters}{"0A}
136:
137: % Italic ``theorems''
138: \font\thf cmssdc10 at 11pt
139: \theoremstyle{plain}
140: \newtheorem{theorem}{Theorem}[section]
141: \newtheorem{hypothesis}[theorem]{Hypothesis}
142: \newtheorem{lemma}[theorem]{Lemma}
143: \newtheorem{proposition}[theorem]{Proposition}
144: \newtheorem{corollary}[theorem]{Corollary}
145: \newtheorem{claim}[theorem]{Claim}
146: \newtheorem{definition}[theorem]{Definition}
147:
148: % Roman ``theorems''
149: \theoremstyle{definition}
150: %\newtheorem{assumption}[lemma]{Assumption}
151: \newtheorem{assumption}[theorem]{Assumption}
152: \newtheorem{condition}[theorem]{Condition}
153:
154: % Humble things: remarks and examples.
155: \theoremstyle{remark}
156: \newtheorem{remark}[theorem]{Remark}
157: \newtheorem{example}[theorem]{Example}
158:
159: %\newenvironment{nam}[args]{begdef}{enddef}
160:
161: %\newcommand\const{\mathop{\rm const}}
162: \newcommand\const{\mathop{{\rm const}}}
163: \newcommand\wlim{\mathop{w\mbox{-}{\rm lim}}}
164: %\renewcommand{\thetheorem}{\arabic{section}.\arabic{theorem}}
165: \renewcommand{\theequation}{\thesection.\arabic{equation}}
166:
167: \font\sectfont=cmssbx10 at 12pt
168: \font\titlefont=cmssbx10 at 20pt
169:
170: \makeatletter\@addtoreset{equation}{section}
171: \makeatother
172:
173:
174: \def\os{{o}}
175: \def\ol{{O}}
176: \def\dist{\mathop{\rm dist}\nolimits}
177: %\def\spec{\,{\rm spec}\,}
178: \def\spec{\sigma}
179: \def\mod{\mathop{\rm mod}\nolimits}
180: \renewcommand{\Re}{\mathop{\eur{R\hskip -1pt e}}\nolimits}
181: \renewcommand{\Im}{\mathop{\eur{I\hskip -1pt m}}\nolimits}
182: \newcommand{\xre}{\sb{\scriptscriptstyle{1,0}}}
183: \newcommand{\xim}{\sb{\scriptscriptstyle{0,1}}}
184:
185: \newcommand{\sppoint}{\sp{\eur d}}
186: \newcommand{\spcont}{\sp{\eur c}}
187: \newcommand{\sbpoint}{\sb{\eur d}}
188: \newcommand{\sbcont}{\sb{\eur c}}
189: \newcommand{\conj}{{\tt C}}
190:
191: % notice- graphics packages
192: %\usepackage[dvips]{epsfig}
193: \usepackage{epsfig}
194: \usepackage{graphicx}
195: \providecommand{\mhy}{\mbox{-}}
196:
197: \newenvironment{equations}{\begin{eqnarray}}{\end{eqnarray}}
198:
199: \renewcommand{\labelenumi}{{\rm (}{\it \roman{enumi}}{\rm )}}
200:
201: \begin{document}
202:
203: \title{Global well-posedness
204: for the Schr\"odinger equation
205: \\
206: coupled to a nonlinear oscillator
207: }
208:
209: \author{
210: {\sc Alexander Komech}
211: \footnote{
212: On leave from Department of Mechanics and Mathematics,
213: Moscow State University, Moscow 119992, Russia.
214: Supported in part
215: by DFG grant 436\,RUS\,113/615/0-1, FWF grant P19138-N13,
216: Max-Planck Institute for Mathematics in the Sciences (Leipzig)
217: and Alexander von Humboldt Research Award (2006).}
218: \\
219: {\it\small Faculty of Mathematics, Vienna University, Wien A-1090, Austria}
220: \\
221: {\sc Andrew Komech}
222: \footnote{
223: Supported in part
224: by Max-Planck Institute for Mathematics in the Sciences (Leipzig)
225: and by the NSF Grants DMS-0434698 and DMS-0600863.
226: }
227: \\
228: {\it\small
229: Mathematics Department, Texas A\&M University,
230: College Station, TX, USA}
231: }
232: \date{\version}
233:
234: \maketitle
235:
236: \begin{abstract}
237: The Schr\"odinger equation
238: with the nonlinearity concentrated at a single point
239: proves to be an interesting and important model
240: for the analysis of long-time behavior of solutions,
241: such as the asymptotic stability of solitary waves
242: and properties of weak global attractors.
243: In this note, we prove global well-posedness
244: of this system in the energy space $\scriptstyle{H\sp 1}$.
245: \end{abstract}
246:
247: %\tableofcontents
248:
249: \section{Introduction and main results}
250: \label{sect-introduction}
251:
252: We are going to prove the well-posedness in
253: $H\sp 1$
254: for the nonlinear Schr\"odinger equation
255: with the nonlinearity concentrated at a single point:
256: \begin{equation}\label{snls}
257: i\dot\psi(x,t)
258: =-\psi''(x,t)-\delta(x)F(\psi(0,t)),
259: \qquad
260: \quad x\in\R,
261: \end{equation}
262: where the dots and the primes
263: stand for the partial derivatives in $t$ and $x$,
264: respectively.
265: The equation describes the Schr\"odinger field
266: coupled to a nonlinear oscillator.
267: This equation is a convenient playground for developing the tools
268: for the analysis of long-time behavior of solutions to
269: $U(1)$-invariant Hamiltonian systems with dispersion.
270: The asymptotic stability of the solitary manifold
271: for equation (\ref{snls})
272: has been considered in \cite{bkks}.
273: Here we complete this result,
274: giving the proof of the global well-posedness of (\ref{snls})
275: in the energy space.
276:
277: Let us mention that for the Klein-Gordon equation
278: with the nonlinearity of the same type
279: the global attraction was addressed in
280: %\cite{ubk-mpi},
281: \cite{ubk-cr},
282: \cite{ubk-arma}.
283:
284: We assume that
285: \begin{equation}\label{FU}
286: F(\psi)=-\nabla\sb\psi U(\psi),
287: \qquad\psi\in\C,
288: \end{equation}
289: for some real-valued potential $U\in C\sp 2(\C)$,
290: where $\nabla\sb\psi$ is the real derivative
291: with respect to $(\Re\psi,\Im\psi)$.
292: Equation (\ref{snls}) is a Hamiltonian system
293: with the Hamiltonian
294: \begin{equation}
295: \mathscr{H}(\psi)
296: =\int\sb{\R}
297: \frac{\abs{\psi'(x)}^2}{2}
298: \,dx
299: +U(\psi(0)),
300: \qquad
301: \psi\in H^1=H\sp 1(\R).
302: \end{equation}
303: The Hamiltonian form of (\ref{snls}) is
304: \begin{equation}\label{Ham}
305: \dot\Psi=J D \mathscr{H}(\Psi),
306: \end{equation}
307: where
308: \begin{equation}
309: \Psi=\left[\begin{array}{c}\Re\psi\\\Im\psi\end{array}\right],
310: \qquad
311: J=\left[\begin{array}{cc}0&1\\-1&0\end{array}\right],
312: \end{equation}
313: and $D \mathscr{H}$ is the Fr\'echet derivative on the Hilbert space $H^1$.
314: The value of the Hamiltonian functional
315: is conserved for classical finite energy solutions of (\ref{snls}).
316: We assume that equation (\ref{snls}) possesses
317: $U(1)$-symmetry,
318: thus requiring that
319: \begin{equation}\label{Uu}
320: U(\psi)=u(\abs{\psi}^2),
321: \qquad
322: \psi\in\C.
323: \end{equation}
324: It then follows that
325: $F(0)=0$
326: and
327: $F(e^{i s}\psi)=e^{i s}F(\psi)$
328: for $\psi\in\C$, $s\in\R$,
329: and that
330: \begin{equation}\label{def-a}
331: F(\psi)=a(\abs{\psi}^2)\psi,
332: \qquad
333: \psi\in\C,
334: \qquad
335: {\rm where}
336: \quad
337: a(\cdot)=2 u'(\cdot)\in\R.
338: \end{equation}
339: This symmetry implies that
340: $e^{i\theta}\psi(x,t)$ is a solution to (\ref{snls})
341: if $\psi(x,t)$ is.
342: According to the N\"other theorem,
343: the $U(1)$-invariance leads (formally)
344: to the conservation of
345: the charge, given by the functional
346: \begin{equation}\label{charge}
347: Q(\psi)=\frac{1}{2}\int\sb{\R}\abs{\psi}^2\,dx.
348: \end{equation}
349: We also assume that $U(\psi)$ is such that
350: \begin{equation}\label{u-is-such}
351: U(z)\ge A-B\abs{z}^2
352: \qquad
353: {\rm with\ some}
354: \quad
355: A\in\R,\quad B > 0.
356: \end{equation}
357: We will show that
358: equation (\ref{snls}) is globally well-posed in $H\sp 1$.
359: We will consider the solutions of class
360: $\psi\in C\sb b(\R\times\R) $.
361: All the derivatives in equation (\ref{snls})
362: are understood in the sense of distributions.
363:
364: \begin{theorem}[Global well-posedness]
365: \label{theorem-gwp}
366: Let the conditions
367: (\ref{FU}), (\ref{Uu}) and (\ref{u-is-such}) hold with $U\in C^2(\C)$. Then
368: \begin{enumerate}
369: \item
370: For any $\phi\in H^1(\R)$,
371: the equation for (\ref{snls})
372: with the initial data $\psi\at{t=0}=\phi$
373: has a unique solution
374: $\psi\in C(\R,H^1(\R))$.
375:
376: \item
377: The values of the charge and energy functionals
378: are conserved:
379: \begin{equation}
380: \label{star}
381: Q(\psi(t))=Q(\phi),
382: \qquad
383: \mathscr{H}(\psi(t))=\mathscr{H}(\phi),
384: \qquad t\in\R.
385: \end{equation}
386: \item
387: There exists $\Lambda(\phi)>0$
388: such that
389: the following a priori bound holds:
390: \begin{equation}
391: \sup\limits\sb{t\in\R}\norm{\psi(t)}\sb{H^1}
392: \le \Lambda(\phi)<\infty.
393: \label{a-priori-h1}
394: \end{equation}
395: \item
396: The map
397: $\eub{U}:\,\psi(0)\mapsto\psi$
398: is continuous
399: from $H^1$ to $L\sp\infty([0,T],H^1(\R))
400: %\cap C\sp{(1/4)}(\R\times\R)
401: $,
402: for any $T>0$.
403: \end{enumerate}
404: \end{theorem}
405:
406: \begin{theorem}\label{theorem-holder}
407: Under conditions of Theorem~\ref{theorem-gwp},
408: $\psi\in C\sp{(1/4)}(\R\times\R)$.
409: \end{theorem}
410:
411: %\begin{theorem}[Smoothness of solution]
412: %\label{theorem-smoothness}
413: %Let the conditions of Theorem~\ref{theorem-gwp}
414: %be satisfied.
415: %Then
416: %\begin{enumerate}
417: %\item
418: % %For any $\epsilon>0$,
419: % %$\psi\in L\sp\infty(\R,H\sp{1}(\R))
420: % %\cap C(\R,H\sp{1-\epsilon}(\R))\cap C\sp{(1/4)}(\R\times\R)$.
421: %$\psi\in C(\R,H\sp{1}(\R))\cap C\sp{(1/4)}(\R\times\R)$.
422: %\item
423: %The map $\eub{U}:\,\psi(0)\mapsto\psi$
424: %is continuous
425: %from $H^1$ to $L\sp\infty([0,T],H^1)
426: %\cap C\sp{(1/4)}(\R\times\R)$,
427: %for any $T>0$.
428: %\end{enumerate}
429: %\end{theorem}
430:
431: Let us give the outline of the proof.
432: We need a small preparation first:
433: We show that, without loss of generality,
434: it suffices to prove the theorem assuming
435: that $U$ is uniformly bounded together with its
436: derivatives.
437: Indeed, the a priori bounds on the $L\sp\infty$-norm
438: of $\psi$ imply that the nonlinearity $F(z)$
439: may be modified for large values of $\abs{z}$.
440: Then we will prove the existence and
441: uniqueness of the solution
442: $\psi\in C\sb b(\R\times[0,\tau])$,
443: for some $\tau>0$.
444: This is accomplished in Section~\ref{sect-cb}.
445:
446: In Section~\ref{sect-regularized},
447: we construct
448: approximate solutions
449: $\psi\sb\epsilon\in C\sb b(\R,H^1(\R))$
450: that are solutions to a regularized problem
451: (the $\delta$-function substituted by its smooth approximations
452: $\rho\sb\epsilon$, $\epsilon>0$).
453: On one hand, the approximate solutions have their
454: energy and charge conserved.
455: On the other hand,
456: we will show in Section~\ref{sect-convergence}
457: that the approximate solutions converge to $\psi(x,t)$
458: uniformly for $\abs{x}\le R$, $0\le t\le\tau$.
459:
460: In Section~\ref{sect-wp},
461: we use the uniform convergence of approximate solutions
462: to conclude that $\psi\in L\sp\infty([0,\tau],H^1(\R))$
463: and moreover that $\psi$ could be extended to all $t\ge 0$.
464: Then we show that
465: the energy and the charge are conserved.
466: We will use these conservations
467: to extend the solution $\psi(x,t)$ for $t\in\R$.
468: Then we prove that $\psi\in C(\R,H^1(\R))$.
469:
470: In Section~\ref{sect-holder},
471: we study the H\"older continuity in time,
472: showing that $\psi\in C\sp{(1/4)}(\R\times\R)$.
473:
474:
475: \section{Local well-posedness in $C\sb b$}
476: \label{sect-cb}
477:
478: \begin{lemma}\label{lemma-a-priori}
479: A priori bound
480: (\ref{a-priori-h1}) follows from (\ref{u-is-such}) and
481: the energy and charge conservation
482: (\ref{star}).
483: \end{lemma}
484:
485: \begin{proof}
486: %\footnote{AC: changed the proof}
487: Let $A\in\R$, $B>0$
488: be constants from (\ref{u-is-such}),
489: and let $\psi\in H\sp 1(\R)$.
490: To estimate $\norm{\psi}\sb{H\sp 1}$ in terms of the values
491: of $Q(\psi)$ and $\mathscr{H}(\psi)$,
492: we need to control the possibly negative contribution of $U(\psi)$
493: into $\mathscr{H}(\psi)$.
494: We achieve this control
495: by using the inequality
496: \begin{equation}\label{psi-0}
497: B\abs{\psi(0)}^2
498: \le
499: B\Big[\int\sb\R\hat\psi(k)\frac{dk}{2\pi}\Big]^2
500: \le
501: B\int\sb\R \big(B^2+{\textstyle\frac{k^2}{4}}\big)
502: \abs{\hat\psi(k)}^2\frac{dk}{2\pi}
503: \cdot
504: \int\sb R\frac{dk}{2\pi\big(B^2+\frac{k^2}{4}\big)}
505: =B^2\norm{\psi}\sb{L\sp 2}^2+{\textstyle\frac{1}{4}}\norm{\psi'}\sb{L\sp 2}^2.
506: \end{equation}
507: This allows us to write
508: \begin{equation}
509: \label{q-h-0}
510: \mathscr{H}(\psi)
511: \ge
512: {\textstyle\frac 1 2}
513: \norm{\psi'}^2\sb{L\sp 2}
514: +A-B\abs{\psi(0)}^2
515: \ge
516: {\textstyle\frac 1 4}
517: \norm{\psi'}^2\sb{L\sp 2}
518: +A-B^2\norm{\psi}\sb{L\sp 2}^2
519: =
520: {\textstyle\frac 1 4}\norm{\psi}\sb{H\sp 1}^2+A
521: -(B^2+{\textstyle\frac{1}{4}})\norm{\psi}^2\sb{L\sp 2}.
522: \end{equation}
523: The first inequality follows from (\ref{u-is-such}),
524: while the second one holds due to the bound (\ref{psi-0}).
525: We rewrite (\ref{q-h-0}) as the bound on $\norm{\psi}\sb{H\sp 1}^2$:
526: \begin{equation}
527: \label{q-h}
528: \norm{\psi}\sb{H\sp 1}^2
529: \le
530: (8B^2+2)Q(\psi)+4\mathscr{H}(\psi)-4A.
531: \end{equation}
532: When we take into account the energy and charge conservation (\ref{star}),
533: the inequality (\ref{q-h})
534: leads to the bound (\ref{a-priori-h1})
535: with
536: \begin{equation}\label{def-Lambda}
537: \Lambda(\phi)
538: =\sqrt{(8B^2+2) Q(\phi)+4\mathscr{H}(\phi)-4A}.
539: \end{equation}
540: \end{proof}
541:
542: \begin{lemma}\label{lemma-u-u}
543: Let us assume that
544: Theorem~\ref{theorem-gwp}
545: is true for the nonlinearities $U$ that satisfy
546: the following additional condition:
547: %\footnote{AC:
548: %{\it For $k=0,1,2$ there exist $U\sb k<\infty$...} moved inside (\ref{u-p-bound})}
549: \begin{equation}\label{u-p-bound}
550: {\it For}
551: \quad
552: k=0,\,1,\,2
553: \quad
554: {\it there\ exist}
555: \quad
556: U\sb k<\infty
557: \quad
558: {\it so\ that}
559: \quad
560: \sup\sb{z\in\C}\abs{\nabla^k U(z)}\le U\sb k.
561: %\qquad
562: %k=0,\,1,\,2.
563: \end{equation}
564: Then Theorem~\ref{theorem-gwp}
565: is also true without this additional condition.
566: \end{lemma}
567:
568: \begin{proof}
569: Fix a nonlinearity $U$
570: that does not necessarily satisfy (\ref{u-p-bound}).
571: For a particular initial data $\phi\in H\sp 1(\R)$
572: in Theorem~\ref{theorem-gwp},
573: we choose $\widetilde{U}(z)\in C\sp 2(\C)$
574: so that
575: $\widetilde{U}(z)=\widetilde{U}(\abs{z})$ for $z\in\C$
576: and
577: $\widetilde{U}(z)=U(z)$ for $\abs{z}\le\Lambda(\phi)$,
578: where $\Lambda(\phi)$ is defined by
579: (\ref{def-Lambda}).
580: We can choose $\widetilde{U}$
581: so that it satisfies (\ref{u-is-such})
582: with the same $A$, $B$ as $U$ does,
583: and also satisfies the uniform bounds
584: \[
585: %U\sb k:=
586: \sup\sb{z\in\C}\abs{\na^k\widetilde{U}(z)}<\infty,
587: \qquad
588: k=0,1,2.
589: \]
590: By the assumption of the Lemma,
591: Theorem~\ref{theorem-gwp}
592: is true for the nonlinearity
593: $\widetilde F=-\nabla\widetilde{U}$
594: instead of $F=-\nabla U$.
595: Hence, there is a unique solution
596: $\psi(x,t)\in L\sp\infty(\R,H\sp 1)\cap C\sb b(\R\times\R)$
597: to the equation
598: \[
599: i\dot\psi(x,t)
600: =-\psi''(x,t)
601: -\delta(x)\widetilde{F}(\psi(0,t)),
602: \]
603: with $\psi\at{t=0}=\phi$.
604: By Lemma~\ref{lemma-a-priori},
605: $\psi$ satisfies the a priori bound (\ref{a-priori-h1})
606: with $\Lambda(\phi)$ defined by (\ref{def-Lambda}).
607: This bound implies that
608: $\abs{\psi(0,t)}
609: \le \Lambda(\phi)$ for $t\in\R$.
610: Therefore, $\widetilde F(\psi(0,t))=F(\psi(0,t))$ for $t\in\R$,
611: and
612: $\psi(x,t)$
613: is also a solution to (\ref{snls})
614: with the nonlinearity $F=-\nabla U$.
615: \end{proof}
616:
617: From now on, we shall assume in the proof of
618: Theorem~\ref{theorem-gwp}
619: that the bounds (\ref{u-p-bound}) hold true.
620:
621:
622: \begin{lemma}
623: \label{lemma-l8-existence}
624: \begin{enumerate}
625: \item
626: Let $\phi\in H^1:=H^1(\R)$.
627: There exists
628: $\tau>0$
629: that depends only on $U\sb 2$ in (\ref{u-p-bound})
630: so that there is a unique solution
631: $\psi\in C\sb b(\R\times[0,\tau])$
632: to equation (\ref{snls})
633: with the initial data $\psi\at{t=0}=\phi$.
634: \item
635: The map $\phi\mapsto\psi$
636: is continuous from $H^1$ to
637: $C\sb b(\R\times[0,\tau])$.
638: \end{enumerate}
639: \end{lemma}
640:
641: \begin{proof}
642: Let us denote
643: the dynamical group for the free Schr\"odinger equation by
644: \begin{equation}\label{def-w0}
645: \eub{W}\sb t\phi(x)
646: =\frac{1}{\sqrt{2\pi t}}
647: \int\sb{\R} e^{i\frac{\abs{x-y}^2}{2t}}\phi(y)\,dy,
648: \qquad
649: x\in\R.
650: \end{equation}
651: For its Fourier transform, we have:
652: \begin{equation}
653: \label{fourier-transform}
654: \mathcal{F}\sb{x\to k}[\eub{W}\sb t\phi(x)](k)=e^{ik^2t}\hat\phi(k),
655: \qquad
656: k\in\R.
657: \end{equation}
658: Then the solution $\psi$ to (\ref{snls})
659: with the initial data $\psi\at{t=0}=\phi$
660: admits the Duhamel representation
661: \begin{equation}\label{duhamel}
662: \psi(x,t)
663: =\eub{W}\sb t\phi(x)
664: =
665: \eub{W}\sb t\phi(x)
666: +\eub{Z}\psi(x,t),
667: \end{equation}
668: where
669: \begin{equation}\label{def-z}
670: \eub{Z}\psi(x,t)
671: =-\int\sb{0}\sp{t}
672: \eub{W}\sb s\delta(x)F(\psi(0,t-s))\,ds
673: =
674: -\int\sb{0}\sp t
675: \frac{e^{i\frac{x^2}{2s}}}{\sqrt{2\pi s}}F(\psi(0,t-s))
676: \,ds.
677: \end{equation}
678: The Fourier representation (\ref{fourier-transform}) implies that
679: $
680: \eub{W}\sb t\phi(x)\in C\sb b(\R,H^1)
681: \subset C\sb b(\R\times\R)
682: $.
683: Further, we compute for
684: $\psi\sb 1$, $\psi\sb 2\in C\sb b(\R\times[0,\tau])$:
685: \[
686: \Abs{\eub{Z}\psi\sb 2(x,t)
687: -\eub{Z}\psi\sb 1(x,t)}
688: \le
689: \int\limits\sb{0}\sp t
690: \frac{
691: \abs{F(\psi\sb 2(0,t-s))-F(\psi\sb 1(0,t-s))}
692: }
693: {\sqrt{2\pi s}}
694: \,ds
695: \le
696: U\sb 2\sqrt{t}
697: \sup\sb{0\le s\le t} \abs{\psi\sb 2(s)-\psi\sb 1(s)},
698: \]
699: where we used (\ref{u-p-bound})
700: with $k=2$.
701: For definiteness, we set
702: \begin{equation}\label{def-tau-0}
703: \tau=\frac{1}{4U\sb 2^2}.
704: \end{equation}
705: Then the map $\psi \mapsto \eub{W}\sb{t}\phi+\eub{Z}\psi$
706: is contracting in the space $C\sb b(\R\times [0,\tau])$.
707: It follows that
708: equation (\ref{duhamel}) admits a unique solution
709: $\psi\in C\sb b(\R\times[0,\tau])$,
710: proving the first part of the theorem.
711: The second part of the theorem also follows by contraction.
712: \end{proof}
713:
714:
715:
716: \section{Regularized equation}
717: \label{sect-regularized}
718: We proved that there is a unique solution
719: $\psi(x,t)
720: \in C([0,\tau]\times\R)$.
721: Now we are going to prove that
722: $\psi\in L\sp\infty(\R\sb{+},H^1)$
723: and moreover that
724: $\norm{\psi(t)}\sb{H\sp 1}$
725: is bounded uniformly in time.
726:
727: \bigskip
728:
729: Let us fix a family of functions
730: $\rho\sb\epsilon(x)$
731: approximating
732: the Dirac $\delta$-function.
733: We pick $\rho\sb 1(x)\in C\sp\infty\sb 0[-1,1]$,
734: nonnegative,
735: and such that $\int\sb{\R}\rho\sb 1(x)\,dx=1$,
736: and define
737: \begin{equation}\label{def-rho-epsilon}
738: \rho\sb\epsilon(x)=\frac{1}{\epsilon}\rho\sb 1\left(\frac{x}{\epsilon}\right),
739: \qquad
740: \epsilon\in(0,1),
741: \end{equation}
742: so that
743: \[
744: \supp\rho\sb\epsilon(x)\subseteq [-\epsilon,\epsilon],
745: \qquad\rho\sb\epsilon(x)\ge 0,
746: \qquad
747: \int\sb{\R}
748: \rho\sb\epsilon(x)\,dx=1.
749: \]
750: Consider the smoothed equation
751: with the ``mean field interaction''
752: \begin{equation}\label{snls-rho}
753: i\dot\psi(x,t)
754: =-\Delta\psi(x,t)-\rho\sb\epsilon(x)
755: F(\langle\rho\sb\epsilon,\psi(t)\rangle),
756: \qquad
757: \end{equation}
758: where
759: \[
760: \langle\rho\sb\epsilon,\psi(t)\rangle
761: =\langle\rho\sb\epsilon(\cdot),\psi(\cdot,t)\rangle
762: =\int\sb{\R}\rho\sb\epsilon(x)\psi(x,t)\,dx.
763: \]
764: Clearly,
765: equation (\ref{snls-rho})
766: is the Hamiltonian equation,
767: with the Hamilton functional
768: \begin{equation}
769: \mathscr{H}\sb\epsilon(\psi)=\int\frac{\abs{\nabla\psi}^2}{2}\,dx
770: +U(\langle\rho\sb\epsilon,\psi\rangle).
771: \end{equation}
772: The Hamiltonian form of (\ref{snls-rho}) is (cf. (\ref{Ham}))
773: \begin{equation}\label{Hameps}
774: \dot\Psi\sb\epsilon=J D \mathscr{H}\sb\epsilon(\Psi\sb\epsilon).
775: \end{equation}
776: The solution $\psi\sb\epsilon$
777: to (\ref{snls-rho})
778: with the initial data $\psi\sb\epsilon\at{t=0}=\phi$
779: admits the Duhamel representation
780: \begin{equation}\label{Duep}
781: \psi\sb\epsilon(x,t)
782: =\eub{W}\sb t\phi(x)
783: +\eub{Z}\sb\epsilon\psi\sb\epsilon(x,t),
784: \end{equation}
785: where
786: \begin{equation}\label{def-z-epsilon}
787: \eub{Z}\sb\epsilon\psi\sb\epsilon(x,t)
788: =
789: -\int\sb{0}\sp{t}
790: \eub{W}\sb s
791: \rho\sb\epsilon(x)
792: F(\langle\rho\sb\epsilon,\psi\sb\epsilon(t-s)\rangle)
793: \,ds.
794: \end{equation}
795:
796: \begin{lemma}[Local well-posedness]
797: \label{lemma-local-existence}
798: \begin{enumerate}
799: \item
800: For any $\epsilon\in (0,1)$,
801: there exists $\tau\sb\epsilon>0$
802: that depends on $\epsilon$ and on $U\sb 2$
803: from (\ref{u-p-bound})
804: so that there is a unique solution
805: $\psi\sb\epsilon\in C\sb b([0,\tau\sb\epsilon],H^1)$
806: to equation (\ref{snls-rho})
807: with $\psi\sb\epsilon\at{t=0}=\phi$.
808: \item
809: For each $t\le\tau\sb\epsilon$,
810: the map
811: $\eub{U}\sb\epsilon(t):
812: \phi=\psi\sb\epsilon(0)\mapsto \psi\sb\epsilon(t)$
813: is continuous in $H^1$.
814:
815: \item
816: The values of the functionals
817: $\mathscr{H}\sb\epsilon$ and $Q$
818: on solutions to (\ref{snls-rho})
819: are conserved in time.
820: \end{enumerate}
821: \end{lemma}
822:
823: \begin{proof}
824: \begin{enumerate}
825: \item
826: %We are going to prove that there exists $\tau\sb\epsilon>0$
827: %that only depends on $\mathscr{H}(\phi)$ and $\epsilon$
828: %so that the mapping
829: %$\psi\mapsto \eub{W}\sb t\phi+\eub{Z}\sb\epsilon\psi$
830: %is contracting in the norm $C\sb b([0,\tau\sb\epsilon],H^1(\R))$.
831: %It suffices to show that the map $\psi\mapsto \eub{Z}\sb\epsilon\psi$
832: %is contracting in the norm $C\sb b([0,\tau\sb\epsilon],H^1(\R))$
833: %if $\tau\sb\epsilon>0$ is small enough.
834: For $\psi\sb 1$, $\psi\sb 2\in C\sb b([0,\tau\sb\epsilon],H^1)$,
835: we compute:
836: \begin{eqnarray*}
837: &&\norm{\eub{Z}\sb\epsilon\psi\sb{2}(\cdot,t)
838: -\eub{Z}\sb\epsilon\psi\sb{1}(\cdot,t)}\sb{H\sp 1}
839: \\
840: &&=\norm{
841: \int\sb{0}\sp{t}
842: \eub{W}\sb s
843: \rho\sb\epsilon
844: F(\langle\rho\sb\epsilon,\psi\sb{2}(t-s)\rangle)
845: -F(\langle\rho\sb\epsilon,\psi\sb{1}(t-s)\rangle)
846: \,ds
847: }\sb{H\sp 1}
848: \\
849: &&\le
850: \int\sb{0}\sp{t}
851: \Norm{
852: \eub{W}\sb s
853: \rho\sb\epsilon
854: }\sb{H\sp 1}
855: \abs{
856: F(\langle\rho\sb\epsilon,\psi\sb{2}(t-s)\rangle)
857: -F(\langle\rho\sb\epsilon,\psi\sb{1}(t-s)\rangle)
858: }
859: \,ds.
860: \end{eqnarray*}
861: The first factor under the integral sign
862: is bounded uniformly for $0<s\le t$:
863: \begin{eqnarray}
864: \nonumber
865: \norm{
866: \eub{W}\sb s
867: \rho\sb\epsilon
868: }\sb{H\sp 1\sb x}
869: =\frac{1}{\sqrt{2\pi}}
870: \Norm{
871: \sqrt{1+k^2}
872: e^{ik^2 s/2}
873: \widehat{\rho\sb\epsilon}(k)
874: }\sb{L\sp 2\sb k}
875: =
876: \Norm{
877: \rho\sb\epsilon
878: }\sb{H\sp 1}.
879: \nonumber
880: \end{eqnarray}
881: Taking this into account, we get:
882: \begin{eqnarray*}
883: \norm{\eub{Z}\sb\epsilon\psi\sb{2}(\cdot,t)
884: -\eub{Z}\sb\epsilon\psi\sb{1}(\cdot,t)}\sb{H\sp 1}
885: &\le&
886: \norm{\rho\sb\epsilon}\sb{H\sp 1}
887: \int\sb{0}\sp{t}
888: \abs{
889: F(\langle\rho\sb\epsilon,\psi\sb{2}(t-s)\rangle)
890: -F(\langle\rho\sb\epsilon,\psi\sb{1}(t-s)\rangle)
891: }
892: \,ds
893: \\
894: &\le&
895: t U\sb 2\norm{\rho\sb\epsilon}\sb{H\sp 1}
896: \sup\sb{s\in[0,t]}
897: \abs{
898: \langle\rho\sb\epsilon,\psi\sb{2}(s)-\psi\sb{1}(s)\rangle}.
899: \end{eqnarray*}
900: Therefore, the map $\psi\mapsto\eub{W}\sb t\phi+\eub{Z}\sb\epsilon\psi$
901: is contracting if we choose, for definiteness,
902: \begin{equation}\label{def-tau-e}
903: \tau\sb\epsilon=\frac{1}{4 U\sb 2\norm{\rho\sb\epsilon}\sb{H\sp 1}}.
904: \end{equation}
905:
906: \item
907: The continuity of the mapping
908: $\eub{U}\sb\epsilon(t)$
909: also follows from the contraction argument.
910:
911: \item
912: It suffices to prove
913: the conservation of the values of
914: $\mathscr{H}\sb\epsilon(\psi\sb\epsilon(t))$
915: and $Q(\psi\sb\epsilon(t))$
916: for $\phi\in H^2:=H^2(\R)$ since the functionals
917: are continuous on $H^1$.
918: For $\phi\in H^2$,
919: the corresponding solution belongs to the space
920: $C\sb b([0,\tau\sb\epsilon],H^2)$
921: by
922: the Duhamel representation (\ref{Duep}).
923: Then
924: the energy and charge conservation follows
925: by the Hamiltonian structure (\ref{Hameps}).
926: Namely, the differentiation of the Hamilton functional
927: gives by the chain rule,
928: \begin{equation}\label{eneps}
929: \displaystyle
930: \frac d{dt}\mathscr{H}\sb\epsilon(\Psi\sb\epsilon(t))=
931: \langle
932: D\mathscr{H}\sb\epsilon(\Psi\sb\epsilon(t)),\dot\Psi\sb\epsilon(t)
933: \rangle
934: =
935: \langle
936: D\mathscr{H}\sb\epsilon(\Psi\sb\epsilon(t)),JD\mathscr{H}\sb\epsilon(\Psi\sb\epsilon(t))
937: \rangle
938: =0
939: \end{equation}
940: since the Fr\'echet derivative
941: $D\mathscr{H}\sb\epsilon(\Psi\sb\epsilon(t))
942: =-\Delta\Psi\sb\epsilon(\cdot,t)
943: -\rho\sb\epsilon(\cdot)F(\langle\rho\sb\epsilon,\Psi\sb\epsilon(t)) \rangle)
944: $
945: belongs to $L^2(\R)$ for $t\in[0,\tau\sb\epsilon]$.
946: Similarly, the charge conservation
947: follows by the differentiation,
948: \begin{eqnarray}\label{qeps}
949: \displaystyle
950: \frac d{dt}Q (\Psi\sb\epsilon(t))&=&
951: \langle
952: D Q(\Psi\sb\epsilon(t)),\dot\Psi\sb\epsilon(t)
953: \rangle
954: =
955: \langle
956: D Q(\Psi\sb\epsilon(t)),J D\mathscr{H}\sb\epsilon(\Psi\sb\epsilon(t))
957: \rangle
958: \nonumber\\
959: &=&
960: \langle
961: \Psi\sb\epsilon(x,t),J\Delta\Psi\sb\epsilon(x,t)\rangle
962: -
963: \langle\Psi\sb\epsilon(x,t),J\rho\sb\epsilon(x)
964: F(\langle\rho\sb\epsilon,\Psi\sb\epsilon(t)\rangle)
965: \rangle.
966: \end{eqnarray}
967: Here
968: $\Psi\sb\epsilon(x,t),J\Delta\Psi\sb\epsilon(x,t)\rangle
969: =\na\Psi\sb\epsilon(x,t),J\na\Psi\sb\epsilon(x,t)\rangle=0$,
970: and also
971: \begin{eqnarray}
972: \langle\Psi\sb\epsilon(x,t),J\rho\sb\epsilon(x)
973: F(\langle\rho\sb\epsilon,\Psi\sb\epsilon(t)\rangle)\rangle
974: &=&
975: \int\Psi\sb\epsilon(x,t)
976: \cdot[J\rho\sb\epsilon(x)F(\langle\rho\sb\epsilon,\Psi\sb\epsilon(t)\rangle)]
977: \,dx\nonumber\\
978: &=&
979: \langle\rho\sb\epsilon,\Psi\sb\epsilon(t)\rangle
980: \cdot[J F(\langle\rho\sb\epsilon,\Psi\sb\epsilon(t)\rangle)]
981: =0.
982: \end{eqnarray}
983: Here ``$\cdot$'' stands for the real scalar product in $\R^2$,
984: and $Z\cdot[J F(Z)]=0$ for $Z\in\R^2$
985: since
986: $F(Z)=a(\abs{Z})Z$ with $a(\abs{Z})\in\R$ by (\ref{def-a}).
987: \end{enumerate}
988: \end{proof}
989:
990: \begin{corollary}[Global well-posedness]
991: \label{corollary-global}
992: \begin{enumerate}
993: \item
994: For any $\epsilon>0$, $\epsilon\le 1$,
995: there exists a unique solution
996: $\psi\sb\epsilon\in C(\R,H^1)$
997: to equation (\ref{snls-rho})
998: with $\psi\sb\epsilon\at{t=0}=\phi$.
999:
1000: The $H\sp 1$-norm of $\psi\sb\epsilon$
1001: is bounded uniformly in time:
1002: \begin{equation}\label{eps}
1003: %\sup\limits\sb{t\in\R}\norm{\psi\sb\epsilon(t)}\sb{H\sp 1}^2
1004: %\le
1005: %M^2(\phi):=\norm{\phi}\sb{H\sp 1}^2 +4U\sb 0,
1006: %\qquad
1007: \sup\limits\sb{t\in\R}\norm{\psi\sb\epsilon(t)}\sb{H\sp 1}
1008: \le\Lambda\sb\epsilon(\phi),
1009: \qquad
1010: t\in\R,
1011: \end{equation}
1012: where
1013: \begin{equation}
1014: \Lambda\sb\epsilon(\phi)
1015: =
1016: \sqrt{(8B^2+2)Q(\phi)
1017: +4\mathscr{H}\sb\epsilon(\phi)-4A}.
1018: \end{equation}
1019: \item
1020: For each $t\ge 0$,
1021: the map
1022: $\eub{U}\sb\epsilon(t):
1023: \psi\sb\epsilon(0)\mapsto \psi\sb\epsilon(t)$
1024: is continuous in $H^1$.
1025: \end{enumerate}
1026: \end{corollary}
1027:
1028: \begin{proof}
1029: \begin{enumerate}
1030: \item
1031: The existence and uniqueness of the solution
1032: $\psi\sb\epsilon\in C\sb b([0,\tau\sb\epsilon],H^1)$ follow
1033: from Lemma~\ref{lemma-local-existence} ({\it i}).
1034: The bound on the value of the $H\sp 1$-norm
1035: of $\psi\sb\epsilon(t)$
1036: is obtained as in Lemma~\ref{lemma-a-priori}.
1037: %\begin{eqnarray}\label{M}
1038: %&&\norm{\psi\sb\epsilon(t)}\sb{H\sp 1}^2
1039: %=2 Q(\psi\sb\epsilon(t))+
1040: %2\mathscr{H}\sb\epsilon(\psi\sb\epsilon(t))
1041: %-2U(\langle\rho\sb\epsilon,\psi\sb\epsilon(t)\rangle)
1042: %\nonumber\\
1043: %&&\le 2 Q(\psi\sb\epsilon(t))+
1044: %2\mathscr{H}\sb\epsilon(\psi\sb\epsilon(t)) +2 U\sb 0
1045: %\le 2 Q(\phi)+2\mathscr{H}\sb\epsilon(\phi)+2U\sb 0
1046: %\nonumber\\
1047: %&&\le\norm{\phi}\sb{H\sp 1}^2+4 U\sb 0,
1048: %\end{eqnarray}
1049: %where we used the first bound from (\ref{u-p-bound}) (twice).
1050: Namely, noting that
1051: \[
1052: U(\langle\rho,\psi\sb\epsilon\rangle)
1053: \ge
1054: A-B\langle\rho,\psi\sb\epsilon\rangle^2
1055: \ge
1056: A-B\sup\sb{x\in\R}\abs{\psi\sb\epsilon}^2
1057: \ge
1058: A
1059: -B^2\norm{\psi}\sb{L\sp 2}^2
1060: -\frac 1 4\norm{\psi'}\sb{L\sp 2}^2
1061: \]
1062: and using the energy and charge conservation
1063: proved in Lemma~\ref{lemma-local-existence} ({\it iii}),
1064: we conclude that
1065: \[
1066: (2B^2+{\textstyle\frac 1 2})Q(\phi)
1067: +\mathscr{H}\sb\epsilon(\phi)
1068: =
1069: (2B^2+{\textstyle\frac 1 2})Q(\psi\sb\epsilon)
1070: +\mathscr{H}\sb\epsilon(\psi\sb\epsilon)
1071: \ge
1072: A+{\textstyle\frac 1 4}
1073: \norm{\psi\sb\epsilon}\sb{H\sp 1}^2,
1074: \]
1075: so that
1076: \begin{equation}\label{M1}
1077: \norm{\psi\sb\epsilon}\sb{H\sp 1}^2
1078: \le
1079: (8B^2+2)Q(\phi)
1080: +4\mathscr{H}\sb\epsilon(\phi)-4A.
1081: \end{equation}
1082: By (\ref{def-tau-e}),
1083: the time span $\tau\sb\epsilon$ depends only on
1084: $\norm{\rho\sb\epsilon}\sb{H\sp 1}$ and $U\sb 2$.
1085: Hence, the bound (\ref{eps}) allows us to
1086: extend the solution to
1087: $t\in[\tau\sb\epsilon,2\tau\sb\epsilon]$.
1088: The bound (\ref{eps})
1089: for $t\in[0,2\tau\sb\epsilon]$
1090: follows from (\ref{M1})
1091: by the energy and charge conservation
1092: proved in Lemma~\ref{lemma-local-existence} ({\it iii}).
1093: We conclude by induction that the solution exists
1094: and the bound (\ref{eps}) holds
1095: for all $t\in\R$.
1096:
1097: \item
1098: The continuity of the mapping
1099: $\eub{U}\sb\epsilon(t):
1100: \psi\sb\epsilon(0)\mapsto \psi\sb\epsilon(t)$
1101: for all $t\ge 0$ follows from its continuity for small times
1102: by dividing the interval $[0,t]$ into small time intervals.
1103: \end{enumerate}
1104: \end{proof}
1105:
1106: \section{Convergence of regularized solutions}
1107: \label{sect-convergence}
1108:
1109: \begin{lemma}
1110: \label{lemma-uniform}
1111: Let $\tau$
1112: and $\psi\in C\sb b(\R\times[0,\tau])$
1113: be as in Lemma~\ref{lemma-l8-existence},
1114: and let $\psi\sb\epsilon\in C(\R\sb{+},H^1)$
1115: be as in Corollary~\ref{corollary-global}.
1116: Then for any finite $R>0$
1117: \begin{equation}\label{W}
1118: \psi\sb\epsilon(x,t)
1119: \mathop{\rightrightarrows}\limits\sb{\epsilon\to 0}
1120: \psi (x,t),
1121: \qquad
1122: \abs{x}\le R,
1123: \quad
1124: 0\le t\le\tau.
1125: \end{equation}
1126: \end{lemma}
1127:
1128: \begin{proof}
1129: We have
1130: \begin{eqnarray}
1131: \psi\sb\epsilon(x,t)
1132: =\eub{W}\sb t\phi(x)
1133: +\int\sb{0}\sp t
1134: \eub{W}\sb s
1135: \rho\sb\epsilon(x)
1136: F(\langle\rho\sb\epsilon,\psi\sb\epsilon(t-s)\rangle)
1137: \,ds,
1138: \\
1139: \psi(x,t)
1140: =\eub{W}\sb t\phi(x)
1141: +\int\sb{0}\sp t
1142: \eub{W}\sb s
1143: \delta(x)
1144: F(\psi(0,t-s))
1145: \,ds.
1146: \end{eqnarray}
1147: Taking the difference of these equations
1148: and regrouping the terms,
1149: we can write:
1150: \begin{eqnarray}
1151: \psi\sb\epsilon(x,t)-\psi(x,t)
1152: =\int\sb{0}\sp t
1153: \eub{W}\sb s
1154: \rho\sb\epsilon(x)
1155: \left(
1156: F(\langle\rho\sb\epsilon,\psi\sb\epsilon(t-s)\rangle)
1157: -
1158: F(\psi(0,t-s))
1159: \right)\,ds
1160: \nonumber
1161: \\
1162: +\int\sb{0}\sp{t}
1163: [\eub{W}\sb s
1164: \rho\sb\epsilon(x)-\eub{W}\sb s\delta(x)
1165: ]
1166: F(\psi(0,t-s))
1167: \,ds.
1168: \label{lhs}
1169: \end{eqnarray}
1170: Let us analyze the first term
1171: in the right-hand side of (\ref{lhs}).
1172: It is bounded by
1173: \begin{eqnarray}
1174: \Abs{\int\sb{0}\sp t
1175: \frac{e^{i\frac{(x-y)^2}{2s}}}{\sqrt{2\pi s}}
1176: \rho\sb\epsilon(y)\,dy\,ds}
1177: \sup\sb{0\le s\le t}
1178: \abs{F(\langle\rho\sb\epsilon,\psi\sb\epsilon(s)\rangle)
1179: -
1180: F(\psi(0,s))}
1181: \nonumber
1182: \\
1183: \le
1184: \Abs{\int\sb{0}\sp t
1185: \frac{ds}{\sqrt{2\pi s}}
1186: }
1187: U\sb 2
1188: \sup\sb{\abs{x}\le\epsilon,\,0\le s\le t}
1189: \abs{\psi\sb\epsilon(x,s)-\psi(x,s)}
1190: \nonumber
1191: \\
1192: \le
1193: \sqrt{\frac{2t}{\pi}}U\sb 2
1194: \sup\sb{\abs{x}\le\epsilon,\,0\le s\le t}
1195: \abs{\psi\sb\epsilon(x,s)-\psi(x,s)}
1196: \nonumber
1197: \\
1198: \le
1199: \frac 1 2
1200: \sup\sb{\abs{x}\le\epsilon,\,0\le s\le t}
1201: \abs{\psi\sb\epsilon(x,s)-\psi(x,s)},
1202: \label{smaller-than-half}
1203: \end{eqnarray}
1204: where in the last inequality we used (\ref{def-tau-e}).
1205: Setting
1206: $M\sb{R,\tau}=\sup\sb{\abs{x}\le R,\,0\le t\le\tau}
1207: \abs{\psi\sb\epsilon(x,t)-\psi(x,t)}$,
1208: we can rewrite (\ref{lhs}) as
1209: \[
1210: M\sb{R,\tau}
1211: \le
1212: \frac 1 2 M\sb{R,\tau}
1213: +
1214: \sup\sb{\abs{x}\le R,\,0\le t\le\tau}
1215: \int\sb{0}\sp{t}
1216: [\eub{W}\sb s
1217: \rho\sb\epsilon(x)-\eub{W}\sb s\delta(x)
1218: ]
1219: F(\psi(0,t-s))
1220: \,ds.
1221: \]
1222: Therefore,
1223: \begin{equation}\label{sup-delta-psi}
1224: M\sb{R,\tau}
1225: \le
1226: 2\sup\sb{\abs{x}\le R,\,0\le t\le\tau}
1227: \int\sb{0}\sp{t}
1228: \int
1229: \frac{e^{i\frac{(x-y)^2}{2s}}}{\sqrt{2\pi s}}
1230: [
1231: \rho\sb\epsilon(y)-\delta(y)
1232: ]\,dy\,F(\psi(0,t-s))ds.
1233: \end{equation}
1234: We claim that the right-hand side tends to zero as
1235: $\epsilon\to 0$.
1236: To prove this, we split the integral
1237: into two pieces:
1238: \begin{eqnarray}
1239: I\sb 1(\delta,\epsilon)
1240: =\int\sb{\delta}\sp{t}\int
1241: \frac{e^{i\frac{(x-y)^2}{2s}}}{\sqrt{2\pi s}}
1242: [
1243: \rho\sb\epsilon(y)-\delta(y)
1244: ]\,dy\,F(\psi(0,t-s))ds,
1245: \label{second-guy}
1246: \\
1247: I\sb 2(\delta,\epsilon)
1248: =\int\sb{0}\sp \delta\int
1249: \frac{e^{i\frac{(x-y)^2}{2s}}}{\sqrt{2\pi s}}
1250: [
1251: \rho\sb\epsilon(y)-\delta(y)
1252: ]\,dy\,F(\psi(0,t-s))ds,
1253: \label{third-guy}
1254: \end{eqnarray}
1255: where $\delta\in(0,t)$ is yet to be chosen.
1256: Let us analyze the term (\ref{second-guy}):
1257: \begin{equation}
1258: \abs{I\sb 1(\delta,\epsilon)}
1259: \le
1260: C U\sb 0
1261: \sup\sb{s\ge\delta,\abs{x}\le R}
1262: \Abs{
1263: \int\sb{\abs{y}<\epsilon}
1264: \frac{e^{i\frac{(x-y)^2}{2s}}}{\sqrt{2\pi s}}
1265: [
1266: \rho\sb\epsilon(y)-\delta(y)
1267: ]
1268: \,dy}.
1269: \end{equation}
1270: Since $s\ge\delta>0$ and $\abs{x}\le R$,
1271: the function
1272: $\frac{e^{i\frac{(x-y)^2}{2s}}}{\sqrt{2\pi s}}$
1273: is Lipschitz in $y\in [-\epsilon,\epsilon]$, uniformly in all the parameters.
1274: Therefore,
1275: \begin{equation}
1276: \int\sb\R
1277: \frac{e^{i\frac{(x-y)^2}{2s}}}{\sqrt{2\pi s}}
1278: [
1279: \rho\sb\epsilon(y)-\delta(y)
1280: ]
1281: \,dy
1282: \to 0,
1283: \qquad
1284: \epsilon\to 0,
1285: \end{equation}
1286: uniformly in the parameters.
1287: We conclude that
1288: \begin{equation}\label{i-1}
1289: \lim\sb{\epsilon\to 0}
1290: I\sb 1(\delta,\epsilon)=0,
1291: \end{equation}
1292: for any fixed $\delta>0$.
1293: We then bound (\ref{third-guy})
1294: uniformly by
1295: \[
1296: I\sb 2(\delta,\epsilon)
1297: \le C U\sb 0
1298: \int(\rho\sb\epsilon(y)+\delta(y))\,dy
1299: \int\sb{0}\sp \delta\frac{ds}{\sqrt{s}}
1300: \le C\sqrt{\delta},
1301: \]
1302: with $C$ independent of $\epsilon$.
1303: Now apparently
1304: the right-hand side of (\ref{sup-delta-psi})
1305: tends to zero as $\epsilon\to 0$.
1306: \end{proof}
1307:
1308: \section{Well-posedness in energy space}
1309: \label{sect-wp}
1310:
1311: \begin{lemma}[Local well-posedness]
1312: \label{lemma-h1}
1313: There is a unique solution
1314: $\psi\in L^\infty([0,\tau],H^1(\R))\cap C\sb b(\R\times[0,\tau])$
1315: to equation (\ref{snls}) with $\psi\at{t=0}=\phi$,
1316: where $\tau$ is as in (\ref{def-tau-0}).
1317: \end{lemma}
1318:
1319: \begin{proof}
1320: The unique solution $\psi\in C\sb b(\R\times[0,\tau])$
1321: is constructed in Lemma~\ref{lemma-l8-existence}.
1322: According to (\ref{eps}) and (\ref{W}),
1323: \begin{equation}\label{mono-h}
1324: \norm{\psi(t)}\sb{H\sp 1}
1325: \le\liminf \sb{\epsilon\to 0}\norm{\psi\sb\epsilon(t)}\sb{H\sp 1}
1326: \le \Lambda(\phi),
1327: \qquad
1328: 0\le t\le\tau.
1329: \end{equation}
1330: \end{proof}
1331:
1332: \begin{lemma}
1333: The values of the functionals
1334: $\mathscr{H}$ and $Q$
1335: are conserved in time for $t\in[0,\tau]$.
1336: \end{lemma}
1337:
1338: \begin{proof}
1339: The convergence
1340: (\ref{W}) and the bounds (\ref{eps}) imply that
1341: \begin{equation}\label{charge-conservation}
1342: Q(\psi(t))
1343: =
1344: \frac 1 2
1345: \norm{\psi(t)}\sb{L\sp 2}^2
1346: \le
1347: \frac 1 2
1348: \lim\sb{\epsilon\to 0}\norm{\psi\sb\epsilon(t)}\sb{L\sp 2}^2
1349: =Q(\phi),
1350: \end{equation}
1351: where we used the conservation of $Q$
1352: for the approximate solutions $\psi\sb\epsilon$
1353: (Lemma~\ref{lemma-local-existence}).
1354: The same argument
1355: applied to the initial data $\psi\at{t=t\sb 0}$
1356: with any $t\sb 0\in (0,\tau)$
1357: and combined with the uniqueness of the solution,
1358: allows to conclude that
1359: $Q(\psi(t))$ is monotonically non-increasing
1360: when time changes from $0$ to $\tau$.
1361: Instead,
1362: solving the Schr\"odinger equation backwards in time
1363: and using the uniqueness of solution,
1364: we can as well conclude that
1365: $Q(\psi(t))$ is monotonically non-decreasing
1366: when time changes from $0$ to $\tau$.
1367: This proves that
1368: $Q(\psi(t))=\const$ for $t\in[0,\tau]$.
1369:
1370: To prove the conservation of $\mathscr{H}(\psi(t))$,
1371: we will need the relation
1372: \begin{equation}\label{lim-u-good}
1373: \lim\sb{\epsilon\to 0}
1374: U(\langle\rho\sb\epsilon,\psi\sb\epsilon\rangle)
1375: =U(\psi(0,t)).
1376: \end{equation}
1377: This relation
1378: follows from continuity of the potential $U$
1379: and from
1380: \begin{equation}\label{lim-is-lim}
1381: \lim\sb{\epsilon\to 0}
1382: \langle\rho\sb\epsilon,\psi\sb\epsilon(t)\rangle
1383: =
1384: \lim\sb{\epsilon\to 0}
1385: \langle\rho\sb\epsilon,(\psi\sb\epsilon(t)-\psi(t))\rangle
1386: +
1387: \lim\sb{\epsilon\to 0}
1388: \langle\rho\sb\epsilon,\psi(t)\rangle
1389: =\psi(0,t),
1390: \end{equation}
1391: where
1392: $\lim\sb{\epsilon\to 0}
1393: \langle\rho\sb\epsilon,(\psi\sb\epsilon(t)-\psi(t))\rangle
1394: =0$
1395: since $\psi\sb\epsilon$
1396: approaches $\psi$ uniformly for $0\le t\le\tau$
1397: and $\abs{x}\le R$ (including $x=0$),
1398: while
1399: $\lim\sb{\epsilon\to 0}
1400: \langle\rho\sb\epsilon,\psi(t)\rangle
1401: =\psi(0,t)
1402: $
1403: since $\psi$
1404: is continuous in $x$
1405: (due to the finiteness of $H\sp 1$-norm
1406: of $\psi$ that follows from (\ref{mono-h})).
1407: We have:
1408: \[
1409: \mathscr{H}(\psi(t))
1410: =
1411: \frac{\norm{\na \psi(x,t)}\sb{L^2}^2}{2}
1412: +U(\psi(0,t))
1413: \le\lim\sb{\epsilon\to 0}
1414: \left\{
1415: \frac{\norm{\na\psi\sb\epsilon(x,t)}\sb{L^2}^2}{2}
1416: +U(\langle\rho\sb\epsilon,\psi\sb\epsilon\rangle)
1417: \right\}
1418: =\mathscr{H}(\phi),
1419: \]
1420: where
1421: we used the relation (\ref{lim-u-good})
1422: and (\ref{W}).
1423: We also used the conservation of
1424: the values of the functional $\mathscr{H}\sb\epsilon$
1425: for the approximate solutions $\psi\sb\epsilon$
1426: (see Lemma~\ref{lemma-local-existence}).
1427: Proceeding just as with $Q(\psi(t))$ above,
1428: we conclude that $\mathscr{H}(\psi(t))=\const$
1429: for $0\le t\le\tau$.
1430: \end{proof}
1431:
1432:
1433: \begin{corollary}[Global well-posedness]
1434: \label{corollary-global-energy}
1435: There is a unique solution
1436: $\psi\in L^\infty(\R,H^1(\R))\cap C\sb b(\R\times\R)$
1437: to equation (\ref{snls}) with $\psi\at{t=0}=\phi$.
1438: The values of the functionals
1439: $\mathscr{H}$ and $Q$
1440: are conserved in time.
1441: \end{corollary}
1442:
1443: \begin{proof}
1444: The solution
1445: $\psi\in L\sp\infty([0,\tau],H\sp 1)$
1446: constructed in Lemma~\ref{lemma-h1}
1447: exists for $0\le t\le\tau$,
1448: where the time span $\tau$
1449: defined in (\ref{def-tau-0}) depends only on $U\sb 2$ from (\ref{u-p-bound}).
1450: Hence, the bound (\ref{a-priori-h1})
1451: at $t=\tau$
1452: allows us to
1453: extend the solution $\psi$ constructed in Lemma~\ref{lemma-h1}
1454: to the time interval $[\tau,2\tau]$.
1455: We proceed by induction.
1456: \end{proof}
1457:
1458: For the conclusion of Theorem~\ref{theorem-gwp},
1459: it remains to prove that $\psi\in C(\R,H\sp 1(\R))$.
1460: This follows from the next two lemmas.
1461:
1462: \begin{lemma}\label{lemma-h1-weak}
1463: $\psi\in C(\R, H^1\sb{weak}(\R))$.
1464: \end{lemma}
1465:
1466: \begin{proof}
1467: Fix $f\in H^{-1}(\R)$ and pick any $\delta>0$.
1468: Since $H\sp 1$ is dense in $H\sp{-1}$,
1469: there exists
1470: $g\in H\sp 1(\R)$ such that
1471: \begin{equation}\label{qq}
1472: \norm{f-g}\sb{H\sp{-1}}
1473: <\frac{\delta}{4\Lambda(\phi)},
1474: \end{equation}
1475: where $\Lambda(\phi)$ given by (\ref{def-Lambda})
1476: is the a priori bound on
1477: $\norm{\psi(t)}\sb{H\sp{1}}$
1478: proved in Lemma~\ref{lemma-a-priori}
1479: on the grounds of the energy and the charge conservation
1480: for $\psi(t)$.
1481: Then
1482: \begin{eqnarray}
1483: &&
1484: \abs{\langle f,\psi(t)-\psi(t\sb 0)\rangle}
1485: \le
1486: \abs{\langle f-g, \psi(t)-\psi(t\sb 0)\rangle}
1487: +\abs{\langle g, \psi(t)-\psi(t\sb 0)\rangle}
1488: \\
1489: &&
1490: \le
1491: \norm{f-g}\sb{H\sp{-1}}
1492: \big(\norm{\psi(t)}\sb{H\sp{1}}+\norm{\psi(t\sb 0)}\sb{H\sp{1}}\big)
1493: +\norm{g}\sb{H\sp 1}
1494: \norm{\psi(t)-\psi(t\sb 0)}\sb{H\sp{-1}}.
1495: \label{rrr}
1496: \end{eqnarray}
1497: By (\ref{qq}),
1498: the first term
1499: in the right-hand side of (\ref{rrr})
1500: is bounded by
1501: $\delta/2$.
1502: By Corollary~\ref{corollary-global-energy},
1503: we have $\psi\in L\sp\infty(\R,H\sp 1(\R))$,
1504: and equation (\ref{snls}) yields $\psi\in C(\R,H\sp{-1}(\R))$.
1505: Hence, the second term
1506: in the right-hand side of (\ref{rrr})
1507: becomes smaller than
1508: $\delta/2$
1509: if $t$ is sufficiently close to $t\sb 0$.
1510: Since $\delta>0$ was arbitrary,
1511: this proves that
1512: $\lim\sb{t\to t\sb 0}\langle f,\psi(t)-\psi(t\sb 0)\rangle=0$.
1513: \end{proof}
1514:
1515: \begin{proposition}
1516: \label{prop-c-h-1}
1517: $\psi\in C(\R,H\sp 1(\R))$.
1518: \end{proposition}
1519:
1520: \begin{proof}
1521: Let us fix $t\sb 0\in\R$
1522: and compute
1523: \begin{equation}\label{diff-h1}
1524: \lim\sb{t\to t\sb 0}\norm{\psi(t)-\psi(t\sb 0)}\sb{H\sp 1}^2
1525: =
1526: \lim\sb{t\to t\sb 0}
1527: \left(
1528: \norm{\psi(t)}\sb{H\sp 1}^2
1529: -2\langle\psi(t),\psi(t\sb 0)\rangle\sb{H\sp 1}
1530: +\norm{\psi(t\sb 0)}\sb{H\sp 1}^2
1531: \right).
1532: \end{equation}
1533: The relation
1534: \[
1535: \norm{\psi(t)}\sb{H^1}^2
1536: =
1537: 2\big(Q(\psi(t))+H(\psi(t))\big)-2U(\psi(0,t)),
1538: \]
1539: together with the conservation of the energy and charge
1540: and
1541: the continuity of $\psi(0,t)$ for $t\in\R$
1542: (see Corollary~\ref{corollary-global-energy}),
1543: shows that
1544: \[
1545: \lim\sb{t\to t\sb 0}
1546: \norm{\psi(t)}\sb{H^1}^2
1547: =
1548: \norm{\psi(t\sb 0)}\sb{H^1}^2.
1549: \]
1550: By Lemma~\ref{lemma-h1-weak},
1551: $\lim\sb{t\to t\sb 0}
1552: \langle\psi(t),\psi(t\sb 0)\rangle\sb{H\sp 1}
1553: =\langle\psi(t\sb 0),\psi(t\sb 0)\rangle\sb{H\sp 1}$.
1554: This shows that
1555: the right-hand side of (\ref{diff-h1}) is equal to zero.
1556: \end{proof}
1557:
1558: Now Theorem~\ref{theorem-gwp} is proved.
1559:
1560: \section{H\"older regularity of solution}
1561: \label{sect-holder}
1562:
1563: In this section, we prove Theorem~\ref{theorem-holder}.
1564:
1565: \begin{lemma}
1566: \label{lemma-w0}
1567: If $\phi\in H^1$, then
1568: $\eub{W}\sb{(\cdot)}\phi(x)\in C\sp{(1/4)}[0,\tau]$, uniformly in $x\in\R$.
1569: \end{lemma}
1570:
1571: \begin{proof}
1572: Let $t$, $t'\in[0,\tau]$.
1573: We have by the Cauchy-Schwarz inequality:
1574: \begin{equations}
1575: %\begin{eqnarray}
1576: \abs{
1577: \eub{W}\sb{t'}\phi(x)
1578: -\eub{W}\sb t\phi(x)
1579: }
1580: \le
1581: C\Abs{
1582: \int
1583: e^{-i k x}\Big(e^{i\frac{t' k^2}{2}}
1584: -e^{i\frac{tk^2}{2}}\Big)
1585: \hat\phi(k)\,dk
1586: }
1587: \nonumber
1588: \\
1589: \le
1590: C\int
1591: \min(1,\abs{t'-t}k^2)
1592: \abs{\hat\phi(k)}\,dk
1593: \le
1594: C\left[
1595: \int\sb\R
1596: \frac{\min(1,\abs{t'-t}k^2)^{2}}{1+k^2}
1597: \,dk
1598: \right]^{\frac 1 2}
1599: \norm{\phi}\sb{H\sp 1}.
1600: \nonumber
1601: %\end{eqnarray}
1602: \end{equations}
1603: We bound the last integral
1604: as follows:
1605: \[
1606: \int\sb\R
1607: \frac{\min(1,\abs{t'-t}k^2)^{2}}{1+k^2}
1608: \,dk
1609: \le
1610: \int\limits\sb{\abs{k}<\abs{t'-t}^{-\frac 1 2}}
1611: \frac{\abs{t'-t}^2 k^4}{1+k^2}
1612: \,dk
1613: +
1614: \int\limits\sb{\abs{k}>\abs{t'-t}^{-\frac 1 2}}
1615: \frac{dk}{1+k^2}
1616: \le\const\abs{t'-t}^{\frac 1 2}.
1617: \]
1618: \end{proof}
1619:
1620: \begin{lemma}[Regularity of $\psi(0,t)$]
1621: \label{lemma-t}
1622: The unique solution $\psi\in C\sb b(\R\times[0,\tau])$
1623: to equation (\ref{snls})
1624: with the initial data
1625: $\psi\at{t=0}=\phi$
1626: constructed in Lemma~\ref{lemma-l8-existence}
1627: satisfies
1628: \[
1629: \psi(0,\cdot)\in C\sp{(1/4)}[0,\tau].
1630: \]
1631: \end{lemma}
1632:
1633: \begin{proof}
1634: Due to Lemma~\ref{lemma-w0},
1635: it suffices to consider the regularity of
1636: $\eub{Z}\psi(0,t)$.
1637: For
1638: any $t,\,t'\in[0,\tau]$,
1639: $t'<t$,
1640: we have:
1641: \begin{equation}\label{z-z-0}
1642: \eub{Z}\psi(0,t')-\eub{Z}\psi(0,t)
1643: =
1644: \int\sb{0}\sp t
1645: \left[
1646: \frac{F(\psi(0,s))}{\sqrt{2\pi(t'-s)}}
1647: -\frac{F(\psi(0,s))}{\sqrt{2\pi(t-s)}}
1648: \right]
1649: \,ds
1650: +\int\sb{t}\sp{t'}
1651: \frac{F(\psi(0,s))}{\sqrt{2\pi(t'-s)}}\,ds.
1652: \end{equation}
1653: The first integral in the right-hand side of (\ref{z-z-0})
1654: is bounded by
1655: \[
1656: C\sb 1\int\sb{0}\sp t
1657: \Abs{
1658: \frac{1}{\sqrt{t'-s}}
1659: -\frac{1}{\sqrt{t-s}}
1660: }\,ds
1661: \le
1662: C\sb 2\abs{t'-t}\sp{1/2}.
1663: \]
1664: The second integral in the right-hand side of (\ref{z-z-0})
1665: is also bounded by $C\abs{t'-t}\sp{1/2}$.
1666: \end{proof}
1667:
1668: \begin{lemma}\label{lemma-holder}
1669: $\psi(x,\cdot)\in C\sp{(1/4)}(\R)$,
1670: uniformly in $x\in\R$.
1671: \end{lemma}
1672:
1673: \begin{proof}
1674: We have the relation
1675: \begin{equation}\label{t-zero}
1676: \psi(x,t)
1677: =\eub{W}\sb{t-t\sb 0}\psi(x,t\sb 0)
1678: +
1679: \int\sb{0}\sp{t-t\sb 0}
1680: \frac{e^{i\frac{x^2}{2s}}}{\sqrt{2\pi s}}F(\psi(0,t-s))\,ds.
1681: \end{equation}
1682: By Lemma~\ref{lemma-w0},
1683: the first term in the right-hand side of (\ref{t-zero}),
1684: considered as a function of time,
1685: belongs to
1686: $C\sp{(1/4)}(\R)$
1687: (uniformly in $x\in\R$).
1688: The second term
1689: in the right-hand side of (\ref{t-zero})
1690: is bounded by $\const\abs{t-t\sb 0}\sp{1/2}$.
1691: This proves that
1692: $\psi(x,\cdot)\in C\sp{(1/4)}(\R)$,
1693: uniformly in $x$.
1694: \end{proof}
1695:
1696: It remains to mention that
1697: the H\"older continuity in $x$
1698: follows from the inclusion
1699: $H\sp 1(\R)\subset C\sp{(1/4)}(\R)$.
1700: Theorem~\ref{theorem-holder} is proved.
1701:
1702: \section*{Acknowledgments}
1703: The authors are grateful to David Stuart
1704: for helpful remarks.
1705:
1706: \bibliographystyle{amsalpha}
1707: \bibliography{shd-local,all,berestycki,books,bourgain,cazenave,christ,comech,cuccagna,delort,esteban,grillakis,hormander,komech,melrose,pego,pelinovsky,phong,roudenko,seeger,smith,shatah,soffer,sogge,stein,strauss,sugim
1708: oto,symes,tataru,uhlmann,weinstein,zworski}
1709:
1710: \end{document}
1711:
1712:
1713: \def\cprime{$'$} \def\cprime{$'$} \def\cprime{$'$}
1714: \def\cydot{\leavevmode\raise.4ex\hbox{.}} \def\cprime{$'$}
1715: \def\cydot{\leavevmode\raise.4ex\hbox{.}} \def\cprime{$'$} \def\cprime{$'$}
1716: \def\cprime{$'$} \def\cprime{$'$}
1717: \def\polhk#1{\setbox0=\hbox{#1}{\ooalign{\hidewidth
1718: \lower1.5ex\hbox{`}\hidewidth\crcr\unhbox0}}} \def\cprime{$'$}
1719: \def\cprime{$'$} \def\cprime{$'$}
1720: \providecommand{\bysame}{\leavevmode\hbox to3em{\hrulefill}\thinspace}
1721: \providecommand{\MR}{\relax\ifhmode\unskip\space\fi MR }
1722: % \MRhref is called by the amsart/book/proc definition of \MR.
1723: \providecommand{\MRhref}[2]{%
1724: \href{http://www.ams.org/mathscinet-getitem?mr=#1}{#2}
1725: }
1726: \providecommand{\href}[2]{#2}
1727: \begin{thebibliography}{KK06b}
1728:
1729:
1730: \bibitem[KK06a]{ubk-cr}
1731: Alexander Komech and Andrew Komech,
1732: \emph{On global attraction to solitary waves for the {K}lein-{G}ordon
1733: equation coupled to nonlinear oscillator}, C. R., Math., Acad. Sci. Paris
1734: \textbf{343} (2006), 111--114.
1735:
1736: \bibitem[KK06b]{ubk-arma}
1737: \bysame,
1738: \emph{Global attractor for a nonlinear
1739: oscillator coupled to the {K}lein-{G}ordon field}, Arch. Ration. Mech. Anal.
1740: (2006), to appear. Max-Planck Institute for Mathematics in the Sciences,
1741: Preprint Nr. 121/2005, {\tt\small
1742: http://www.mis.mpg.de/preprints/2005/prepr2005\_121.html\,}.
1743:
1744: \end{thebibliography}
1745:
1746:
1747: %\bibliographystyle{amsalpha}
1748: %\bibliography{shd-local,all,berestycki,books,bourgain,cazenave,christ,comech,cuccagna,delort,esteban,grillakis,hormander,komech,melrose,pego,pelinovsky,phong,roudenko,seeger,smith,shatah,soffer,sogge,stein,strauss,sugimoto,symes,tataru,uhlmann,weinstein,zworski}
1749:
1750: \end{document}
1751:
1752: