1: \documentstyle[12pt]{article}

2: 

3: %\usepackage{amsfonts,amsmath}

4: 

5: 

6: %\usepackage{latexsym}

7: %\usepackage{amsthm}

8: %\usepackage{amscd}

9: %\usepackage{epsfig}

10: \setlength{\textwidth}{430pt}

11: \setlength{\textheight}{46\baselineskip}

12: \setlength{\topmargin}{10pt}

13: \setlength{\oddsidemargin}{10pt}

14: %\usepackage{psfrag}

15: 

16:  \textwidth 150mm

17:  \textheight 230mm

18:  \setlength{\unitlength}{0.01in}

19:  \def\sepand{\rule{14cm}{0pt}\and}

20: 

21:  %\usepackage{epsf}

22: 

23: 

24:  %\input{psfig}

25: 

26:  \input eqmacros

27:  \input totalmacroe

28:  %\English

29: 

30:  \begin{document}

31:  \topmargin 0pt

32:  \oddsidemargin 5mm

33:  \headheight 0pt

34:  \topskip 0mm

35: 

36: 

37:  \addtolength{\baselineskip}{0.4\baselineskip}

38: 

39:  \pagestyle{empty}

40: 

41:  \vspace{0.1cm}

42: 

43: \hfill  

44: 

45: \hfill 

46: 

47: 

48: \vspace{1cm}

49: 

50: \begin{center}

51: 

52: 

53: {\Large \bf Noncommutative Waves have}

54: 

55: \medskip

56: 

57: {\Large\bf Infinite Propagation Speed}

58: 

59: %\vspace{1 truecm}

60: 

61: %\centerline{}

62: 

63: \vspace{1.2 truecm}

64: 

65: 

66: %\vspace{1.5cm}

67: 

68: %13 November 2000

69: 

70: 

71: 

72:  \vspace{0.7 truecm}

73: {\bf Bergfinnur Durhuus}\footnote{e-mail: durhuus@math.ku.dk} 

74: 

75: 

76: \vspace{0.5 truecm}

77: 

78: Matematisk Institut and MaPhySto\footnote{MaPhySto - A network in Mathematical Physics and Stochastics, funded by the Danish National Research Foundation.}, Universitetsparken 5

79: 

80: 2100 Copenhagen \O, Denmark

81: 

82:  \vspace{.8 truecm}

83:   

84: 

85: 

86: {\bf Thordur Jonsson}\footnote{e-mail:

87: thjons@raunvis.hi.is}\footnote{Permanent address: University of

88: Iceland, Dunhaga 3, 107 Reykjavik, Iceland}

89: 

90: \vspace{0.5 truecm}

91: 

92: Matematisk Institut, Universitetsparken 5

93: 

94: and

95: 

96: The Niels Bohr Institute, Blegdamsvej 17

97: 

98: 2100 Copenhagen \O, Denmark

99: 

100: 

101:  

102: 

103: 

104: % \vspace{.8 truecm}

105:   

106: 

107: 

108:  \vspace{1.5 truecm}

109: 

110: 

111: 

112:  \end{center}

113: 

114:  \noindent

115:  {\bf Abstract.} We prove the existence of global solutions to the

116: Cauchy problem for noncommutative 

117: nonlinear wave equations in arbitrary even spatial dimensions where the 

118: noncommutativity is only in the spatial directions.  We find that 

119: for existence there are

120: no conditions on the degree of the nonlinearity 

121: provided the potential is positive.

122: We furthermore prove that nonlinear noncommutative waves 

123: have infinite propagation speed, i.e.,

124: if the initial conditions at time $0$ have a compact support then for any

125: positive time the support of the solution can be arbitrarily large. 

126: 

127: 

128: 

129:  \vfill

130: 

131:  \newpage

132:  \pagestyle{plain}

133: 

134: \section{Introduction}  

135: Classical solutions to the static equations of field theories

136: in noncommutative spaces have been studied in some detail in recent years.

137: Most of these theories are defined in the Moyal plane, its higher 

138: dimensional generalizations or on the fuzzy sphere.  

139: Explicit solitonic solutions 

140: have been found in various gauge theories, see, e.g.,

141: \cite{g1,g2,g3,g4,g5} 

142: as well as in 

143: scalar field theories \cite{h1,h2,h3} at infinite noncommutativity  

144: where the existence theory for finite noncommutativity 

145: is now rather complete \cite{p1,p2,p3,p4} especially for the rotationally

146: invariant case.  For general

147: background and reviews of

148: noncommutative field theory we refer to \cite{r1,r2,r3}. 

149: 

150: Some time dependent solutions to noncommutative field equations 

151: have been found in gauge theories and sigma models, see, e.g., 

152: \cite{bak,td1,td2,td3, popov1,popov2,bieling}, 

153: and also by boosting static solutions in scalar 

154: field theories \cite{boost}.

155: The main purpose of this paper is to study the initial value problem for 

156: nonlinear wave equations in odd dimensional noncommutative Lorentzian 

157: spaces with 

158: the noncommutativity in the spatial directions only.

159: We prove the existence  

160: of global solutions for a large class of initial conditions provided 

161: the nonlinear term in the wave equation is the derivative of a positive

162: polynomial which vanishes to second order at zero.

163: We do not need to impose any conditions on the degree of the nonlinearity. 

164: This is in stark contrast with the classical case 

165: where the existence theory depends strongly on the dimension of space 

166: as well as the nonlinearity \cite{strauss}.

167: 

168: Our existence results are obtained by adapting the standard theory of

169: evolution

170: equations in Banach spaces, see, e.g., \cite{reed,reedsimon} and

171: references therein, to

172: the setting of  noncommutative wave equations.

173: The arguments presented in this paper can be generalized to 

174: nonpolynomial nonlinearities, 

175: e.g., entire functions.

176: 

177: We analyse the support properties of the solutions and show 

178: quite generally that the support is arbitrarily large 

179: in all spatial slices at 

180: positive times even though the initial data at time $0$ have compact 

181: supports.  Again

182: this is qualitatively different from classical nonlinear waves which

183: always have a propagation speed equal to that of the corresponding

184: linear equation \cite{strauss}. On the other hand, it should be noted that the noncommutative wave equation under consideration is not Lorentz invariant, since a specific choice of spatial slicing of Minkowski space is assumed. 

185: 

186: \section{Existence of Solutions}

187: Let $\nabla^2$ denote the Laplacian in $\bbR^{2d}$.  For simplicity we

188: choose the nonlinear term in the wave equation to be a polynomial $F$.

189: We assume $F(s)=V'(s)$ where $V(s) > 0$ for all

190: $s\neq 0$,

191: $V(0)=V'(0)=0$ but

192: $V''(0) >0$. 

193: % We can write $F(x)=\sum_jc_jx^j$ where $F(0)=0$.  

194: We study wave equations of the form

195: \beq{1}

196: (\pa_t^2-\nabla^2)\vp (t,x)+F_*(\vp )(t,x)=0,

197: \eeq

198: where $F_*(\vp )$ is defined by replacing the ordinary pointwise product

199: of functions by the standard star product for functions on $\bbR^{2d}$, i.e.,

200: \beq{2}

201: F_*(\vp )=\sum_{n=1}^{N}c_n\, \vp^{*n}\,,

202: \eeq

203: if $F(s)=\sum_{n=1}^{N}c_n\, s^n$ and $\vp^{*n}$ denotes the $n$'th $*$-power of $\vp$. 

204: The star product of two functions $f$ and $g$ is defined by

205: \beq{2x}\left.

206: (f*g)(t,x)=\exp\left(-\frac{i}{2}\Theta_{ij}{\pa\over\pa y_i}

207: {\pa \over\pa z_j}\right)f(t,y)g(t,z)\right|_{y=z=x},

208: \eeq

209: where $\Theta$ is a nondegenerate antisymmetric $2d\times 2d$ matrix.

210: A more enlightening formula for the star product, which clearly

211: exhibits its nonlocal character, is given by 

212: \beq{2xx}

213: (f*g)(t,x)={1\over (2\pi)^{2d}}\int f(t, x+\oh \Theta p)g(t, x+y)

214: e^{ip\cdot y}\,dpdy

215: \eeq

216: which can be obtained from \rf{2x} by Fourier transforming.

217: 

218: We are interested in real valued solutions $\vp$ to Eq.\ \rf{1}.

219: We will assume for simplicity that the

220: noncommutativity matrix $\Theta$ is a direct 

221: sum of $d$ $2\times 2$ antisymmetric matrices 

222: $$

223: \left(\begin{array}{rr}0&1\\-1&0\end{array}\right),

224: $$

225: times a positive noncommutativity

226: parameter, denoted $\theta$. Our results are, however, valid for arbitrary non-degenerate antisymmetric $\Theta$.

227: 

228: It is convenient to apply the Weyl 

229: quantization map to the wave equation and then, with our conventions, 

230: it takes the form

231: \beq{3}

232: \pa_t^2\phi (t)+2\theta^{-1}\sum_{k=1}^d[a^*_k,[a_k,\phi (t)]]+F(\phi(t) )=0,

233: \eeq

234: where $\phi (t)$ is a self-adjoint operator on $L^2(\bbR^d)$ and the operators 

235: $a_k^*$ and $a_k$ are $d$ independent pairs of raising and lowering operators 

236: whose explicit form can be taken to be  

237: \beq{4}

238: a_k={1\over\sqrt{2}}\left(\xi_k+{\pa\over\pa\xi_k}\right),

239: \eeq

240: where $\xi_k$, $k=1,\ldots ,d$, are the coordinates in $\bbR^d$.

241: The relation between a function $\vp$ on $\bbR ^{2d}$ 

242: and the corresponding Weyl operator

243: $\pi (\vp )= \phi$ on $L^2(\bbR ^d)$ is given by

244: \beq{5}

245: \phi ={1\over (2\pi)^d}\int \tilde{\vp}(w_1,\ldots ,w_{2d})

246: \exp\left(i\sum_{j=1}^d (w_{2j-1}\hat{\xi}_j + w_{2j}\hat{p}_j)\right)

247: \, dw,

248: \eeq

249: where $\hat{\xi_k}$ and $\hat{p}_k$ are the position and momentum

250: operators corresponding to the raising and lowering operators $a^*_k$

251: and $a_k$ and $\tilde{\vp }$ denotes the Fourier transform of $\vp$.  

252: The Weyl map is in fact an algebra isomorphism from $L^2(\bbR

253: ^{2d})$ (with the star product) 

254: to the space of Hilbert-Schmidt operators on $L^2(\bbR ^d)$ and takes

255: real valued functions to self-adjoint operators.

256: 

257: The collection of all Hilbert-Schmidt operators on $L^2(\bbR ^d)$ 

258: forms a Hilbert space, denoted 

259: $\cH ^2$, with inner product

260: \beq{6}

261: \br A,B\kt _2=\Tr (A^*B).

262: \eeq

263: We define a self-adjoint operator $\Delta$ on $\cH ^2$ by

264: \beq{6y}

265: \Delta\pi (\vp )=\pi (\nabla^2\vp )

266: \eeq

267: and it is easily checked that

268: \beq{6yy}

269: \Delta\phi =-2\sum_{k=1}^d[a^*_k,[a_k,\phi ]].

270: \eeq

271: Its domain is ${\cal D}(\Delta)=\{\pi(\vp):\;\vp,\nabla^2\vp\in L^2(\bbR^d)\} $.

272: 

273: It is convenient to write the wave equation as two first order

274: equations in $t$:

275: \bea

276: \pa_t\phi_1 & = & \phi_2\\

277: \pa_t\phi_2 & = & \Delta\phi_1-\theta F(\phi_1).

278: \eea

279: Here we have also rescaled the time variable by $\theta^{-1/2}$.  Writing

280: \beq{6xxzz}

281: \Phi =\left(\begin{array}{cc}\phi_1\\ \phi_2\end{array}\right),

282: \eeq

283: for ordered pairs of operators, and defining

284: \beq{7}

285: A=i\left(\begin{array}{cc}0 & I\\ \Delta & 0\end{array}\right),~~~~

286: J\left(\begin{array}{cc}\phi_1\\  \phi_2\end{array}\right)=

287: \left(\begin{array}{cc}0\\- F(\phi_1)\end{array}\right),

288: \eeq

289: we can write the wave equation on the form

290: \beq{8}

291: \pa_t \Phi =-iA\Phi +\theta J(\Phi),

292: \eeq

293: where $A$ is a linear operator acting on pairs of operators.

294: 

295: We now prove existence of solutions to \rf{8} using standard mehods

296: from the theory of nonlinear evolution equations which can also be

297: used to deal with ordinary nonlinear wave equations, see

298: \cite{reed,reedsimon}.  

299: First let us note

300: that we can replace the operator $\Delta$ in \rf{8} by $\Delta -m^2$,

301: $m>0$, by adding a compensating linear term to $F$ and the modified $F$ 

302: is still the derivative of a positive polynomial, provided $m$ is small enough.   Define a self-adjoint 

303: operator $B$ on $\cH ^2$ by

304: \beq{9}

305: B=(-\Delta +m^2)^{1/2}.

306: \eeq

307: Then $B\geq m$ and we define the space $\cH ^{1,2}$ as the

308: set of operators $\phi\in\cH ^2$ for which $B\phi\in\cH ^2$.

309: Let $\cH=\cH ^{1,2}\oplus\cH ^2$  and define an inner product 

310: $\langle\cdot\, , \cdot\rangle$ on $\cH$ by

311: \beq{10}

312: \langle\left(\begin{array}{cc}\phi_1\\ \phi_2\end{array}\right) ,

313: \left(\begin{array}{cc}\psi_1\\ \psi_2\end{array}\right)\rangle =

314: \br B\phi_1, B\psi_1\kt _2 +\br \phi_2 ,\psi _2\kt _2.

315: \eeq

316: With this inner product $\cH$ is a Hilbert space and the operator

317: \beq{33}

318: D=i\left(\begin{array}{cc}0 & I\\ -B^2 & 0\end{array}\right)

319: \eeq

320: is a self-adjoint operator on $\cH$ with domain ${\cal D}(D)= {\cal D}(\Delta)\oplus \cH^2$. We can now apply the following theorem. 

321: 

322: \medskip   

323: \noindent

324: {\bf Theorem.}  {\it Let $T$ be a self-adjoint operator with domain

325: $\cD (T)$ on a Hilbert space $H$ with norm $\|\cdot \|$ and $N:\cD

326: (T)\mapsto \cD (T)$ a mapping such that the following conditions hold

327: for any $\phi, \psi\in\cD (T)$:

328: \begin{enumerate}

329: 

330: \item[(i)] $\| N(\phi )\|\leq C_1(\|\phi\| )\|\phi \|$

331: 

332: \item[(ii)] $\| TN(\phi )\|\leq C_2(\|\phi\| )\|T\phi \|$

333: 

334: \item[(iii)] $\|N(\phi )-N(\psi )\|\leq C_3(\|\phi \|, \|\psi\|) \|\phi

335: -\psi\|$

336: 

337: \item[(iv)] $\|T(N(\phi )-N(\psi ))\|\leq C_4(\|\phi \|, \|\psi\|,

338: \|T\phi\| , \| T\psi\|) \|T\phi

339: -T\psi\|$

340: 

341: \end{enumerate}

342: where the constants $C_j$, $j=1,2,3,4$, are increasing functions of

343: the norms.  Then, for any $\phi_0\in\cD (T)$ there exists $t_0>0$ and

344: a unique

345: continuously differentiable family of vectors $\{\phi (t)\}_{0\leq t<

346: t_0}\subseteq\cD (T)$ such that

347: \beq{13}

348: \pa_t\phi (t)=-iT\phi (t) + N(\phi (t))

349: \eeq

350: and

351: \beq{14}

352: \phi (0)=\phi_0.

353: \eeq

354: If $\|\phi (t)\|$ is bounded by a constant independent of $t$ then one can

355: take $t_0=\infty$. 

356: 

357: }

358: \medskip

359: 

360: A proof of this theorem can be found in \cite{reedsimon}.  The basic

361: step is to convert the differential equation (together with the

362: initial conditions) into the integral equation

363: \beq{15}

364: \phi(t)=e^{-itT}\phi_0+\int_0^te^{-i(t-s)T}N(\phi (t))\,ds.

365: \eeq

366: It is quite straightforward to verify conditions $(i)$-$(iv)$

367: for $N=\theta J$, $T=D$ and the Hilbert space $\cH$ that we defined before.

368: The uniform bound on the norm of the solution is obtained from the

369: conservation of energy.

370: 

371: For illustration we verify $(i)$ and 

372: $(iv)$ leaving the others, which

373: are similar,  to the reader.

374: Let us denote the norm on the space of Hilbert-Schmidt operators by

375: $\|\cdot \|_2$.  Then we have for the norm $\|\cdot \|$ on $\cH$

376: \beq{15x}

377: \|\left(\begin{array}{cc}\phi_1\\ \phi_2\end{array}\right)\|^2=

378: \|B\phi_1\|_2^2+\|\phi_2\|_2^2

379: \eeq

380: and clearly

381: \beq{16}

382: \|J\left(\left(\begin{array}{cc}\phi_1\\

383: \phi_2\end{array}\right)\right)\|^2 =\|F(\phi_1)\|_2^2.

384: \eeq

385: For $(i)$ we therefore need to show that

386: \beq{17}

387: \|F(\phi_1)\|_2\leq C_1 (\|B\phi_1\|_2)\|B\phi_1\|_2.

388: \eeq

389: Since $F$ is a polynomial it suffices to verify the inequality \rf{17}

390: for the case $F(s)=s^n$ and this follows from 

391: \bea

392: \|\phi_1^n\|_2 & \leq & \|\phi_1\|^n_2\nonumber\\

393:                & \leq & \left(m^{-n}\| B\phi_1\|^{n-1}_2\right)

394:                \| B\phi_1\|_2.

395: \eea

396: 

397: For condition $(iv)$ it is again enough to consider the case when $F$ is

398: a monomial.   With $F$ as before we have

399: \beq{18}

400: \| DJ\left(\left(\begin{array}{cc}\phi_1\\

401: \phi_2\end{array}\right)\right)-DJ\left(\left(\begin{array}{cc}\psi_1\\

402: \psi_2\end{array}\right)\right)\|=\|B\phi_1^n-B\psi_1^n\|_2

403: \eeq

404: and

405: \beq{19}

406: \| D\left(\begin{array}{cc}\phi_1\\

407: \phi_2\end{array}\right)-D\left(\begin{array}{cc}\psi_1\\

408: \psi_2\end{array}\right)\|^2 = \|B(\phi_2-\psi_2)\|_2^2 +

409: \|B^2(\phi_1-\psi_1)\|_2^2

410: \eeq

411: so we need to show that

412: \beq{20}

413: \|B(\phi_1^n-\psi_1^n)\|_2\leq C \; \|B^2(\phi_1-\psi_1)\|_2,

414: \eeq

415: where the constant $C$ is an increasing function of the norms as

416: stated in the Theorem.

417: Let us define operators $A_k$ on $\cH ^{1,2}$ by

418: \beq{21}

419: A_k\phi=[a_k,\phi],

420: \eeq

421: for $k=1,\ldots , d$.  The operators $A_k$ are derivations and 

422: \beq{22}

423: -B^2+m^2=\Delta =-2\sum_{k=1}^d A_k^*A_k.

424: \eeq

425: We can write

426: \beq{23}

427: \phi_1^n-\psi_1^n=\sum_{i=0}^{n-1}\phi_1^{n-1-i}(\phi_1-\psi_1)\psi_1^i

428: \eeq

429: so in order to establish \rf{20} it suffices to estimate

430: \beq{24}

431: \|B(\phi_1^i(\phi_1-\psi_1)\psi_1^j)\|_2^2  = 

432: m^2\|\phi_1^i(\phi_1-\psi_1)\psi_1^j)\|_2^2 +2\sum_{k=1}^d

433: \|A_k(\phi_1^i(\phi_1-\psi_1)\psi_1^j)\|_2^2

434: \eeq

435: with $i+j=n-1$.

436: The first term on the right hand side above is bounded by

437: $$

438: m^{-2} \|\phi_1\|_2^{2i}\|\psi_1\|_2^{2j}\|B^2(\phi_1-\psi_1)\|_2^2

439: $$

440: so it suffices to consider the second term.  We have

441: \bea

442: \lefteqn{\|A_k(\phi_1^i(\phi_1-\psi_1)\psi_1^j)\|_2  \leq 

443: i\,\|A_k\phi_1\|\,\|\phi_1\|^{i-1}\|\phi_1-\psi_1\|_2\|\psi_1\|^j }

444: \nonumber\\ &&

445:  +\|\phi_1\|^{i}\|A_k(\phi_1-\psi_1)\|_2\|\psi_1\|^j

446: +j\,\|\phi_1\|^{i}\|\phi_1-\psi_1\|_2\|\,\|A_k\psi_1\|\,\|\psi_1\|^{j-1}. 

447: \eea

448: Since 

449: \beq{26}

450: \|A_k\chi\|_2\leq \|B\chi\|_2

451: \eeq

452: for $k=1,\ldots ,d$ and any $\chi$ in the 

453: domain of $B$ the desired result follows easily by

454: summing over $k$.

455: 

456: It follows from the differentiability of the solution 

457: \beq{e0}

458: \Phi (t)=\left(\begin{array}{cc}\phi_1 (t)\\

459: \pa_t\phi_1(t) \end{array}\right)

460: \eeq

461: and the equation of

462: motion that the energy

463: \beq{e1}

464: E(t)={\rm Tr}\left( \sum_{k=1}^d [a_k,\phi_1 (t)][\phi_1 (t),a^*_k]+\oh 

465: (\pa_t\phi_1 (t))^2+\theta V(\phi_1 (t))\right)

466: \eeq

467: is independent of $t$. Since $V(x)\geq\oh m^2x^2$ for real $x$ it 

468: follows that, for $\varepsilon >0$ sufficiently small, we have

469: \beq{e3}

470: E(t)\geq \varepsilon \|\Phi (t) \|^2 

471: \eeq

472: provided $\Phi(t)$ is self-adjoint, which is the case if the initial conditions are self-adjoint.

473: Thus, we have proven that {\it all solutions to Eq.(\ref{1}) with real initial 

474: data $\vp (0,\cdot)\in {\cal D}(\nabla^2)$ and $\partial_t\vp(0,\cdot)\in L^2(\bbR^{2d})$ are global.}

475: 

476: We remark that the estimates above for the noncommutative case are in general

477: not valid for the corresponding classical (commutative) nonlinear equations.  

478: For example, the Sobolev inequality

479: needed to establish the conditions {\it (i)-(iv)} in the classical case is 

480: valid for arbitrary polynomials $F$ if $d=1$, whereas for $d\geq 2$ it only holds for linear $F$. 

481: %\beq{27}

482: %\|f^n\|_{L^2}\leq C\, \|(-\nabla^2+m^2)^{\oh }f\|_{L^2}

483: %\eeq

484: %is only valid for $n\leq 2d/(2d-2) $.   For other values of $n$ the

485: %$L^2$-norm of $f^n$ can be infinite but 

486: %$\|(-\nabla^2+m^2)^{\oh }f\|_{L^2}$ finite.

487: Star powers of functions are

488: more regular than ordinary powers of functions.  Roughly speaking, the local 

489: singularities of the star product of two functions are no worse than the

490: singularities of the individual functions.   We will discuss this in

491: more detail in the following section.  It follows from this

492: enhanced regularity that the existence

493: problem for nonommutative nonlinear waves is mathematically similar to the

494: existence problem for nonlinear waves in the case where space has been

495: replaced by a discrete lattice.  

496: 

497: \section{Infinite Propagation Speed}

498: In this section we show that in general 

499: the diameter of the support of a solution 

500: to the

501: noncommutative wave equation increases in time with

502: infinite speed.   By support we here mean the support of the function

503: which corresponds to the operator under the Weyl map.  

504: 

505: We first note that the solutions whose existence was established in

506: the preceding section can be written

507: \beq{28}

508: \phi (t)= (\cos tB)\phi (0) +(B^{-1}\sin tB )\pa_t\phi (0)-

509: \theta\int_0^t B^{-1}\sin((t-s)B)F(\phi (s))\,ds

510: \eeq

511: by using \rf{15} and the expression

512: \beq{29}

513: e^{-itD}=\left(\begin{array}{cc}\cos tB & B^{-1}\sin tB\\

514: -B\sin tB & \cos tB \end{array}\right) 

515: \eeq

516: for the linear time development operator acting on $\cH$.

517: Let $\vp(t,\cdot )$ be the function corresponding to $\phi (t)$ under the Weyl

518: map, let $g(t,\cdot )$ correspond to

519: \beq{30}

520: \chi (t) = (\cos tB)\phi (0) +(B^{-1}\sin tB )\pa_t \phi (0)

521: \eeq

522: and let $h(t,\cdot )$ correspond to

523: \beq{31}

524: \psi (t)= \theta\int_0^t B^{-1}\sin((t-s)B)F(\phi (s))\,ds .

525: \eeq

526: Then $g$ is a solution to the linear wave equation and 

527: the support of $g$ is contained in the ordinary causal future of

528: the union of the supports of $\vp(0, \cdot )$ and $\pa_t\vp(0, \cdot )$ by

529: the Huygen's principle for solutions to linear wave equations.    We are

530: therefore interested in studying the support properties of $h(t, \cdot )$.

531: 

532: We note that $\phi (t)\to \phi_0$ and hence also $F(\phi (t))\to

533: F(\phi_0)$ as $t\to 0$ where the convergence is in Hilbert-Schmidt norm.  

534: Since the operator

535: $ (tB)^{-1}\sin(t B)$ converges strongly to the identity operator $I$ 

536: as $t\to 0$ and is uniformly bounded in norm it follows

537: from Eq.\ \rf{31} that

538: \beq{32}

539: \| \psi (t)-\oh t^2 \theta F(\phi_0)\|_2 = o(t^2)

540: \eeq

541: as $t\to 0$.   It follows that the relevant support properties of $h(t,\cdot

542: )$ for small $t$ are determined by those of $F_*(\vp_0)$ where $\phi_0=\pi (\vp_0)$.

543: Suppose that $\vp_0$ is smooth with a compact support $S_0$ and assume that we

544: can show that the support of $F_*(\vp_0)$ is strictly larger

545: than $S_0$ in the sense that there is a closed ball $U$ disjoint from $S_0$ with

546: \beq{32xx}

547: \int_U |F_*(\vp_0)|^2\,dx\equiv A>0.

548: \eeq

549: Then, by Eq.\ \rf{32}, we have

550: \beq{32xy}

551: \int_U |h(t,x)|^2\, dx\geq \oh t^2\theta A+ o(t^2)

552: \eeq

553: and $h(t,\cdot )$ is nonzero on a subset of $U$ of positive measure for all 

554: $t$ small enough.  Evidently

555: this proves that the propagation speed is infinite for the initial values 

556: $\vp(0,\cdot )=\vp_0$ and $\pa_t\vp(0,\cdot )$ smooth with support in $S_0$.

557: 

558: It remains to demonstrate that the support of $F_*(\vp_0)$ is in general strictly larger than that of $\vp_0$. 

559: The support properties of star products of functions are 

560: simply reflected in the

561: algebra of star products of $\delta$-functions and their Fourier transforms.

562: We define the distributions $D(\alpha )$, $E(\beta )$ for $\alpha , \beta\in

563: \bbR^{2d}$ by

564: \beq{32xyz}

565: D(\alpha)=\pi^d \delta_\a~~{\rm and}~~E(\beta )=e^{-2i\Theta (\beta ,\cdot )}

566: \eeq

567: where $\delta_\alpha $ is delta function supported at $\a$ and

568: $\Theta$ is the skew symmetric quadratic form on $\bbR ^{2d}$ entering 

569: Eq.\ (\ref{2x})

570: \beq{uuu}

571: \Theta (\alpha,\beta)=\Theta_{ij}\alpha_i\beta_j\;.

572: \eeq

573:   By a straightforward computation, using \rf{2xx}, one

574: finds

575: \bea

576: D(\alpha )*D(\beta ) & = & e^{2i\Theta (\alpha ,\beta)} E(\alpha -\beta )\label{a1} \\

577: D(\alpha )*E(\beta ) & = &  e^{2i\Theta (\alpha ,\beta)} D(\alpha -\beta )  \label{a2} \\

578: E(\alpha )*D(\beta ) & = & e^{-2i\Theta (\alpha ,\beta)} D(\alpha +\beta )  \label{a3}  \\

579: E(\alpha )*E(\beta ) & = &  e^{-2i\Theta (\alpha ,\beta)} E(\alpha +\beta ).\label{a4}

580: \eea

581: The Weyl map can in fact be extended to distributions using \rf{2xx}

582: and one can show that the star product of two tempered distributions

583: is again a tempered distribution provided they are sufficiently regular or 

584: one of the two distributions has a compact support.

585: 

586: We can write

587: \beq{40}

588: f=\int f(\alpha )\delta_{\alpha }\,d\alpha 

589: \eeq

590: for a function $f$ on $\bbR^d$ so

591: \bea

592: (f * f) (x) & = & \pi^{-2d} \int f(\alpha )f(\beta )

593: (D(\alpha )*D(\beta ))(x)\,d\alpha d\beta  \nonumber \\

594:   & = &  \pi^{-2d} \int f(\alpha )f(\beta )

595:   e^{2i\Theta (\alpha ,\beta )}

596:   e^{2i \Theta (\alpha -\beta ,x)}\,d\alpha d\beta \nonumber\\

597:   & = & \pi^{-2d} \int f(\alpha +\beta )f(\beta ) e^{2i\Theta (\alpha ,\beta

598: )}\,d\beta \,  e^{2i\Theta (\alpha ,x)}\, d\alpha.

599: \eea

600: If $f$ is, say, continuous with compact support, it follows from the last

601: expression that $f*f$ is the Fourier transform of a function with compact

602: support and therefore an entire analytic function.  In particular, its

603: support is $\bbR^{2d}$ unless $f=0$.

604: A similar argument shows that all even star powers of $f$ are analytic if $f$ has

605: a compact support. 

606: 

607: Let us now discuss odd star powers and concentrate on the third star power

608: for simplicity.  We find as before that

609: \bea

610: (f*f*f)(x) & = & \pi^{-3d}\int f(\alpha )f(\beta )f(\gamma )(D(\alpha)

611: *D(\beta) *D(\gamma) )(x)\,d\alpha d\beta d\gamma \nonumber \\

612:    & = & \pi^{-3d}\int  f(\alpha )f(\beta )f(\gamma ) e^{2i\Theta (\beta

613: ,\gamma )}

614: e^{2i\Theta (\alpha ,\beta -\gamma )} D(x -\alpha +\beta -\gamma )\,

615: d\alpha d\beta d\gamma  .\nonumber\\

616: & &\label{3star}

617: \eea

618: from which it follows that the support of $f*f*f$ is contained in

619: \beq{uu}

620: \{ \alpha +\beta -\gamma : \alpha ,\beta ,\gamma\in {\rm supp}\, f\}.

621: \eeq

622: In particular, if the support of $f$ has diameter $R$, the support of

623: $f*f*f$ has diameter 

624: \beq{666p}

625: R'\leq 3R.

626: \eeq

627: In the case of, say, a quartic potential with a nonzero cubic term, i.e.,

628: \beq{yyy}

629: F(s)=V'(s)= as^3+bs^2+cs\,,

630: \eeq

631: where $b\neq 0$, we have therefore proven infinite propagation speed.   

632: 

633: For generic functions

634: $f$ we have an equality in \rf{666p}.  Here, we will not elaborate on the

635: detailed conditions under which this is valid.  It suffices to consider an

636: example which demonstrates infinite propagation speed in

637: the case

638: $b=0$ in Eq.\ \rf{yyy}.  Let

639: \beq{nnn}

640: f=D(\alpha )+D(-\alpha ),

641: \eeq

642: where $\alpha\in \bbR^{2d}$ has Euclidean norm $|\alpha |=R/2$ such that the

643: support of $f$ has diameter $R$.   It is easy to see that

644: \beq{qqq}

645: f*f*f= 3(D(\alpha)+D(-\alpha))+D(3\alpha)+D(-3\alpha)

646: \eeq

647: which has support diameter $R'=3R$.

648: Similarly, $f$ raised to the $(2n+1)$st star power

649: will contain delta functions at $\pm (2n+1)\alpha $.  

650: In order to get an example where $f$ is a smooth function, let $d^\pm_n$ be

651: two sequences of nonnegative smooth functions on $\bbR^{2d}$ 

652: with $d_n^\pm$ supported in a

653: ball of radius $1/n$ around $\pm \alpha$ and such that

654: \beq{iii}

655: \int d_n^\pm (x)\,dx =1.

656: \eeq

657: This ensures that

658: \beq{xxx}

659: f_n\equiv \pi^d(d_n^++d_n^-)

660: \eeq

661: is smooth and has a support diameter $R_n\leq R+2/n$.

662: Furthermore, 

663: \beq{q22}

664: f_n\to f=D(\alpha )+D(-\alpha )

665: \eeq

666: in the sense of distributions.  It is easily seen from \rf{3star} that

667: \beq{ttt}

668: f_n*f_n*f_n\to f*f*f

669: \eeq

670: and hence the support diameter of $f_n*f_n*f_n$ converges to $3R$ as

671: $n\to\infty$.   A similar argument allows one to determine the support 

672: of arbitrary odd star powers of compactly supported functions. 

673: 

674: 

675: 

676: \section{Discussion}

677: We have proved the existence of global solutions to the initial value

678: problem for noncommutative nonlinear wave equations whose non-linear term is the dervative of a positive interaction potential and with

679: noncommutativity in the spatial directions only.  The existence

680: theory is quite simple and independent of the nonlinearity and the (even)

681: dimension of space.

682:  

683: We have shown that the speed of propagation is infinite, meaning that

684: the support of the solution is arbitrarily large at positive times given

685: that the support of the function and its first time derivative are 

686: compact at time $0$.  The part of the

687: solution which travels at infinite speed 

688: is proportional to the noncommutativity parameter $\theta$.

689: 

690: We have discussed only real solutions, but the arguments can be extended 

691: to cover complex wave equations such as

692: \beq{sss}

693: \pa_t^2\phi (t)+2\theta^{-1}\sum_{k=1}^d[a^*_k,[a_k,\phi (t)]]+\phi (t) F(|\phi(t)|^2 )=0\,,

694: \eeq

695: in operator form, 

696: where $|\phi(t)|^2=\phi(t)^*\phi(t)$ and the function $F$ is the derivative 

697: of a polynomial $V(s)$ that is positive for $s>0$ and vanishes linearly at $s=0$.

698: 

699: We can use the formalism set up in this paper to study scattering of

700: noncommutative waves and this will be the subject of a forthcoming

701: publication.  It will be interesting to see whether this scattering

702: theory can be used to analyse the scattering of noncommutative

703: solitons  and whether the results on soliton scattering at large

704: $\theta$ obtained in \cite{h2,sc1,sc2} can be proven rigorously.

705: 

706: 

707: \bigskip

708: 

709: \noindent

710: {\bf Acknowledgements} 

711: 

712: This work is supported in

713: part by MatPhySto funded by the Danish National Research Foundation and 

714: by TMR grant no. HPRN-CT-1999-00161.  We are indebted to the

715: Mittag-Leffler Institute for hospitality and to B. Birnir and R. Nest

716: for helpful discussions.

717: 

718: 

719: \begin{thebibliography}{99}

720: 

721: \bibitem{g1}A.~P. Polychronakos, {\it Flux tube solutions in

722: noncommutative gauge theories}, Phys. Lett. B {\bf 495} (2000) 407

723: [hep-th/0007043]

724: 

725: \bibitem{g2}M. Aganagic, R. Gopakumar, S. Minwalla and A.

726: Strominger,

727: {\it Unstable solitons in noncommutative gauge theory},

728: JHEP {\bf 0104} (2001) 001 [hep-th/0009142]

729: 

730: \bibitem{g3}D. Bak, {\it Exact solutions of multi-vortices and false

731: vacuum bubbles in noncommutative Abelian-Higgs theories},

732: Phys. Lett. B {\bf 495} (2000) 251

733: [hep-th/0008204]

734: 

735: \bibitem{g4}D.~J. Gross and N.~A. Nekrasov, {\it

736: Solitons in noncommutative gauge theory},

737: JHEP {\bf 0103} (2001) 044

738: [hep-th/0010090]

739: 

740: \bibitem{g5}J.~A. Harvey, P. Kraus and F. Larsen, {\it Exact

741: noncommutative solitons}, JHEP {\bf 0012} (2000) 024 [hep-th/0010060]

742: 

743: \bibitem{h1}R. Gopakumar, S. Minwalla and A. Strominger, {\it

744: Noncommutative solitons}, JHEP {\bf 0005} 020 (2000) [hep-th/0003160]

745: 

746: \bibitem{h2}  R. Gopakumar, M. Headrick, M. Spradlin, {\it

747:    On Noncommutative Multi-solitons},  Commun. Math. Phys. {\bf 233} (2003)

748: 355-381 [hep-th/0103256]

749: 

750: \bibitem{h3} M. Spradlin and A. Volovich, {\it Noncommutative solitons

751: on K{\"a}hler manifolds}, JHEP {\bf 0203} (2002) 011

752:    [hep-th/0106180]

753: 

754: \bibitem{p1}B. Durhuus, T. Jonsson and R. Nest,

755: {\it Noncommutative scalar solitons: existence and nonexistence},

756: Phys. Lett. B {\bf 500}

757: (2001) 320 [hep-th/0011139]

758: 

759: \bibitem{p2}B. Durhuus, T. Jonsson and R. Nest,

760: {\it The existence and stability of noncommutative scalar solitons},

761: Commun. Math. Phys. {\bf 233} (2003) 49

762: [hep-th/0107121]

763: 

764: 

765: \bibitem{p3} B. Durhuus and T. Jonsson, {\it 

766: A note on noncommutative scalar multisolitons}, 

767: Phys. Lett. B {\bf 539} (2002) 277

768: [hep-th/0204096]

769: 

770: \bibitem{p4}P. Austing, T. Jonsson and L. Thorlacius, {\it 

771: Scalar Solitons on the Fuzzy Sphere},

772: JHEP {\bf 0210} (2002) 073 [hep-th/0206060]

773:                  

774: \bibitem{r1}A. Konechny and A. Schwarz, {\it Introduction to

775: (M)atrix

776: theory and noncommutative geometry}, Phys. Rep. {\bf 360} (2002) 354-465

777: [hep-th/0012145, hep-th/0107251]

778:  

779: \bibitem{r2}M.~R. Douglas and N.~A. Nekrasov, {\it

780: Noncommutative field theory},

781: Rev. Mod. Phys. {\bf 73} (2002) 977-1029

782: [hep-th/0106048]

783: 

784: \bibitem{r3}R.~J. Szabo, {\it Quantum field theory on noncommutative

785: spaces},  Phys. Rept. {\bf 378} (2003) 207-299

786: [hep-th/0109162]

787: 

788: \bibitem{bak}D. Bak, K. Lee, and T.-H. Park, {\it Noncommutative vertex

789: solitons}, Phys. Rev. D {\bf 63} (2001) 125010 [hep-th/0011099]

790: 

791: \bibitem{td1}A. Hashimoto and N. Itzhaki, {\it Travelling faster than

792: the speed of light in noncommutative geometry}, Phys. Rev. D {\bf 63}

793: (2001) 126004 [hep-th/0012093]

794: 

795: \bibitem{td2} L. Hadasz, U. Lindstrom, M. Rocek and R. von Unge, {\it

796: Time dependent solutions of noncommutative Chern-Simons theory coupled

797: to scalar fields}, UUITP-13-03, YITP-SB-03038, hep-th/0309015

798: 

799: \bibitem{td3}P.~A. Horvathy and P.~C. Stichel, {\it Moving vortices in

800: noncommutative gauge theory}, Phys. Lett. B {\bf 583} (2004) 353 

801: [hep-th/0311157]

802: 

803: \bibitem{popov1}O. Lechtenfeld and A.~D. Popov, {\it Noncommutative

804: multi-solitons in 2+1 dimensions}, JHEP {\bf 0111} (2001) 040

805: [hep-th/0106213]

806: 

807: \bibitem{popov2}O. Lechtenfeld and A.~D. Popov, {\it Scattering of noncommutative

808: solitons in 2+1 dimensions}, Phys. Lett. B {\bf 253} (2001) 178

809: [hep-th/0108118]

810: 

811: \bibitem{bieling}S. Bieling, {\it Interaction of noncommutative plane waves

812: in 2+1 dimensions}, J. Phys. A {\bf 35} (2002) 6281

813: [hep-th/0203269]

814: 

815: \bibitem{boost}D. Bak and K. Lee, {\it Elongation of moving

816: noncommutative solitons}, Phys. Lett. B {\bf 495} (2000) 231

817: [hep-th/0007107]

818: 

819: \bibitem{strauss}W. A. Strauss, {\it Nonlinear wave equations},

820: Regional Conference Series in Mathematics {\bf 73}, AMS, 1989.

821: 

822: 

823: \bibitem{reed}M. Reed, {\it Abstract nonlinear wave equations},

824: Lecture Notes in Math. {\bf 507}, Springer, 1976.

825: 

826: \bibitem{reedsimon}M. Reed and B. Simon, {\it Methods of modern

827: mathematical physics, II Fourier analysis, self-adjointness}, Academic

828: Press, 1975.

829: 

830: \bibitem{sc1} U. Lindstr{\"o}m, M. Rocek and R. von Unge, {\it Noncommutative

831: soliton scattering}, JHEP {\bf 0012} (2000) 004 [hep-th/0008108]

832: 

833: \bibitem{sc2}L. Hadasz, U. Lindstr{\"o}m, M. Rocek and R. von Unge, {\it

834: Noncommutative multisolitons: moduli spaces, quantization, finite $\theta$

835: effects and stability}, JHEP {\bf 0106} (2001) 040 [hep-th/0104017]

836: 

837: \end{thebibliography}

838: 

839: 

840: \end{document}

841: 

842: 

843: 

844: 

845: 

846: 

847: 

848: 

849: