1: %documentstyle[preprint,prl,aps]{revtex}

2: \documentclass[12pt]{article}

3: \usepackage{graphicx,amssymb,amsmath,amsthm}

4: %\usepackage{time} %for displaying of the time

5: \usepackage{cite}

6: \newtheorem{theorem}{Theorem}

7: \newtheorem{definition}{Definition}

8: \newtheorem{proposition}{Proposition}

9: \newtheorem{property}{Property}

10: \newtheorem{corollary}{Corollary}

11: \newcommand{\Li}{[L_1, L_2]}

12: \newcommand{\Lp}{[L_1^{'}, L_2^{'}]}

13: \newcommand{\be}{\begin{equation}}

14: \newcommand{\ee}{\end{equation}}

15: \newcommand{\bea}{\begin{align}}

16: \newcommand{\eea}{\end{align}}

17: \newcommand{\ben}{\begin{eqnarray}}

18: \newcommand{\een}{\end{eqnarray}}

19: 

20: \newcommand{\spann}{{\mbox{\rm{span}}}}

21: 

22: \newcommand{\til}{\tilde}

23: \newcommand{\ov}{\overline}

24: \newcommand{\ra}{\rangle}

25: \newcommand{\la}{\langle}

26: \newcommand{\ii}{\hat{I}}

27: 

28: \newcommand{\sumn}{\sum_{n=1}^N}

29: 

30: 

31: \newcommand{\spa}{,\,}

32: \newcommand{\arexna}{\imath\frac{a n\pi x }{L}}

33: 

34: \newcommand{\subz}{n \in \N}

35: \newcommand{\sumz}{\sum_{\subz}}

36: \def\N{\mathbb{N}}

37: \newcommand{\h}{{\cal{H}}}

38: \newcommand{\W}{{\cal{W}}}

39: 

40: 

41: \setlength\unitlength{1cm}

42: \setlength\topmargin{-2cm}

43: \setlength\oddsidemargin{-0.1in}

44: \setlength\textwidth{17cm}

45: \setlength\textheight{23cm}

46: 

47: \title{On the truncation of the harmonic oscillator wavepacket } 

48: 

49: \author{L. Rebollo-Neira and S. Jain\\

50: Aston University,\\

51: Birmingham B4 7ET, United Kingdom}

52: 

53: \date{}

54: \begin{document}

55: \maketitle

56: \baselineskip = 1.7 \baselineskip

57: 

58: \begin{abstract}

59: We present an interesting result regarding the 

60: implication of truncating the wavepacket of the 

61: harmonic oscillator. We show that

62:  disregarding the  non-significant tails of 

63: a function which is the superposition of 

64: eigenfunctions of the harmonic oscillator 

65: has a remarkable consequence. Namely, there exit 

66: infinitely many different superpositions giving rise to 

67:  the same function on the interval. Uniqueness, in the 

68: case of a wavepacket, is restored by a postulate of  

69: quantum mechanics. 

70: 

71: PACS: 03.65.-w, 03.65.Ca

72: \end{abstract}

73: \section{Introduction}

74: We analyse the effect of truncating the wavepacket of the 

75: harmonic oscillator in the light of the {\em frame theory}. 

76: Such a theory, developed in 1952 by  Duffin and Shaffer 

77: in the context of harmonic analysis \cite{da} has been 

78: applied, for over fifteen years, to construct 

79: coherent states. More specifically, we 

80: should mention affine coherent states, 

81: also called wavelets, 

82: and  Weyl-Heisenberg coherent states, 

83: also  known as Gabor frames 

84: \cite{do,wa,ka1,ali3,ali1,do2,han,alib,rewf,rs}.

85: We recall in the 

86: next paragraph the general definition of frames and a

87: few properties, which is all what we need for the 

88: purpose of the present effort. For a complete treatment 

89: of frames we refer to \cite{yo,ftf,c2}.  

90: 

91: Given a Hilbert space $\h$, a family 

92: $\{\phi_n\}_{n\in \N}$ in $\h$ 

93: is called a {\em{frame}} for $\h$ if 

94: for every $f \in \h$ 

95: there exists a pair of constants

96: $0 < A \le B <\infty $ such that

97: \be

98: A \la  f , f \ra  \le  \sum_{\subz} | \la \phi_n , f \ra|^2 \le

99: B \la  f , f \ra.

100: \label{fc}

101: \ee

102: The constants $A$ and $B$ are called the

103: {\em {frame bounds}} and (\ref{fc}) the 

104: {\em frame condition}. 

105: From its definition it is clear that a frame

106: is a complete set,  since

107: the relations $ \la \phi_n, f\ra=0 \spa  \subz$ imply

108: that $f\equiv 0$. The removal of an element

109: from a frame leaves either a frame or an incomplete set.

110: A frame that ceases to be complete if

111: an arbitrary element $\phi_n$ is removed, is called

112: {\em{exact}}. This last property implies that only 

113: exact frames are bases;

114: in the general case

115: the  family $\{\phi_n\}_{n\in \N}$ may be

116:  over complete.

117: When the condition $A=B$ holds, the frame is said to be

118: a {\em{tight}} frame. Assuming that all the elements 

119: $\phi_n,\,  \subz $ are normalised to unity,

120: a tight frame is an orthogonal basis if and only if $A=B=1$.

121: Notice that this implies that a tight frame with

122: frame bounds $A=B=1$ is an orthogonal basis only if

123: the elements are normalised to unity, otherwise it

124: is a redundant frame. 

125: 

126: In this paper we introduce a class of redundant tight frames 

127: which are trivially obtained 

128: by redefining the functions of an orthonormal basis  

129: to be zero outside an interval. Such a restriction allows 

130: us to use the truncated functions to represent a given function  

131: vanishing outside the identical interval. This has 

132: a remarkable consequence, namely, the 

133: coefficients of the corresponding linear span are {\em{not unique}}. 

134: We discuss this important consequence in relation to the truncation of 

135: the wavepacket of an harmonic oscillator. We show that, 

136: by disregarding the

137: non-significant tails of a function which is  the

138: superposition of the harmonic oscillator eigenfunctions 

139: one creates a {\em{null space}}. As a consequence, the 

140: restriction of the wavepacket to the reduced interval can be 

141: realized by infinitely many different coefficients 

142: giving rise to the same function on the interval. This 

143: is, certainly, a striking result. 

144: Nevertheless, uniqueness can be 

145: restored by means of one of the postulates of 

146:  quantum mechanics. 

147: 

148: The paper is organised as follows: In Section II we 

149:  introduce the construction of tight frames for 

150: the Hilbert space of functions vanishing 

151: outside an interval 

152: by simple truncation of orthonormal basis functions. 

153: In Section III we apply these results to analyse the effect of 

154: truncating the wavepacket of the harmonic oscillator. The 

155: conclusions are drawn in Section IV. 

156: 

157: \section{Building tight frames from orthonormal bases}

158: The next proposition shows that, for the space of

159: square integrable functions vanishing outside an interval, 

160: one can construct tight frames by simple restriction of 

161: orthonormal functions to the corresponding interval. 

162: \begin{proposition}

163: \label{p1}

164: Let $\{\psi_n\}_{\subz}$ be an orthonormal basis for $L^2 \Li $, i.e.,

165: \be

166: \la \psi_m, \psi_n \ra =\int_{L_1}^{L_2} \psi_m^\ast(x)\psi_n(x)\, dx =

167: \delta_{m,n}

168: \ee 

169: and ${\chi}_{\Lp}$ the characteristic function for the 

170: interval ${\Lp}$, i.e, 

171: $${\chi}_{\Lp}(x)= \left\{\begin{array}{cll} 1

172:  \, \, & {\mbox{if}} \,\, x \in \Lp \\

173:                        0 \,\,& {\mbox{otherwise.}}

174: \end{array} \right.$$ 

175: Functions $\{\psi'_n\}_{\subz}$, obtained as

176: $\psi'_n(x) ={\chi}_{\Lp}(x)\psi_n(x) \spa \subz$, with

177:  $\Lp \subset \Li$, constitute a tight frame for the

178: subspace ${\cal{W}}$ of square integrable functions

179: vanishing outside $\Lp$.

180: \end{proposition}

181: \begin{proof}

182: On the one hand, since $\{\psi_n\}_{\subz}$ is an

183: orthonormal basis of $L^2\Li$, for all $f \in {L^2\Li}$ we have

184: \be

185: ||f||^2= \la f,f \ra=

186: \sumz \la f , \psi_n \ra \la  \psi_n,  f \ra.

187: \ee

188: On the other hand, for all $f \in {\cal{W}}$ it is true

189: that ${\chi}_{\Lp} f= f$ and we further have:

190: $$\sumz \la f , \psi_n' \ra \la  \psi_n', f \ra =

191:  \sumz \la f {\chi}_{\Lp}, \psi_n \ra \la \psi_n,

192: {\chi}_{\Lp} f \ra

193:   = ||f||^2,$$

194: which proves that $\{\psi_n\}_{\subz}$ is a redundant tight

195: frame for ${\cal{W}}$, since the elements $\psi_n'$ are not normalised 

196: to unity.

197: \end{proof}

198: %\begin{corollary}

199: %\label{fou}

200: %The set ${\chi}_{[-L\,\,L]} \frac{{1}}{\sqrt{2L}}e^{\arexna}$

201: %with $a\in (0\,,\,1)$

202: %is a redundant tight frame for the subspace of square

203: %integrable functions vanishing outside $[-L\,,\,L]$.

204: %The corresponding frame bound is $A=a^{-1}$.

205: %\end{corollary}

206: %\begin{proof}It is a direct consequence of the previous

207: %proposition,

208: %as $\frac{ \sqrt{a}}{\sqrt{2L}}e^{\arexna},\, \subz$ is an orthonormal

209: %basis for $\ii_{L^2[-\frac{L}{a}\,,\,\frac{L}{a}]}.$

210: %\end{proof}

211: We prove next that, through the finite subset of 

212: frames elements $\psi_1',\ldots,\psi_N'$

213: constructed as indicated in Proposition~\ref{p1} 

214: we can construct the orthogonal 

215: projection of $f$ onto 

216: ${\cal{S}}=\spann\{\psi_1,\ldots,\psi_N\}$, restricted to 

217: $\Lp$.

218: \begin{proposition}

219: Let $\psi_1',\ldots,\psi_N'$ be as defined in 

220: Proposition~\ref{p1}, and for each $f \in {\cal{W}}$ let us define  

221: a function $f^N$ as:

222: \be

223: \label{opp}

224: f^N= \sum_{i=1}^N \psi_i' \la \psi'_i, f \ra.

225: \ee

226: The function given in \eqref{opp} satisfies: 

227: $$f^N= \left\{\begin{array}{cll} \hat{P}_{\cal{S}}f(x)

228:  \, \, & {\mbox{if}} \,\, x \in \Lp \\

229:                    0 \,\,& {\mbox{otherwise,}}

230: \end{array} \right.$$

231: where $\hat{P}_{\cal{S}}$ stands for the orthogonal projector 

232:  operator onto ${\cal{S}}$.

233: \end{proposition}

234: \begin{proof}

235: Since $\{\psi_1,\ldots,\psi_N\}$ is an orthonormal set for

236: $L^2 \Li$, the orthogonal projection of $f \in L^2 \Li$  onto 

237: ${\cal{S}}$ is given as

238: \be

239: \hat{P}_{\cal{S}}f = \sum_{i=1}^N \psi_i \la \psi_i, f\ra. 

240: \ee

241: For $f \in {\cal{W}}$ it holds  that

242: $$\sum_{i=1}^N \psi_i \la \psi_i, f\ra= \sum_{i=1}^N \psi_i \la \psi_i', f\ra.$$Then 

243: \be

244: \hat{P}_{\cal{S}}f = \sum_{i=1}^N \psi_i \la \psi_i', f\ra,

245: \ee

246: and the proof follows by multiplication of both sides of the equation 

247: by ${\chi}_{\Lp}$. 

248: \end{proof}

249: A convenient property of tight frames is that 

250:  the coefficients of a linear span in a tight frame superposition

251: are obtained by inner products with the frame functions.

252: In our context this implies that  

253: the coefficients of an orthonormal 

254: expansion and the ones to span a function

255: in $\W$ by means of the frame in Proposition~\ref{p1}

256:  are computed in an equivalent manner.

257: The essential difference is that, as discussed below,

258: the coefficients in the frame superposition are 

259: {\em{not unique}}.

260:  

261: Let us consider the frame constructed in Proposition \ref{p1} and 

262: let us define ${\cal{S}'}=\spann\{\psi'_1,\ldots,\psi'_N\}$. For every 

263: $f\in {\cal{S}'} \subset {\cal{W}}$ we have: 

264: \be

265: \label{fd}

266: f=\sum_{n=1}^{N} 

267: c_n \psi'_n, \,\,\,\,\,\text{with}\,\,\,\,\,c_n=\la \psi_n', f \ra.

268: \ee

269: Since $f \in \W$ it is true that $c_n=\la \psi_n', f \ra= \la \psi_n, f\ra$ 

270: and the equivalence with the orthonormal case follows. However, the 

271: redundancy of the frame implies that the coefficients in \eqref{fd} are 

272: not unique. Indeed, since a redundant frame is linearly dependent,  

273: the following situation can occur:

274: \be

275: \label{ld}

276: 0=\sum_{n=1}^{N} c'_n \psi'_n,\,\,\,\,\,\text{for}\,\,\,\,\,\sum_{n=1}^N |c'_n|^2 \ne 0.

277: \ee

278: Taking inner product both sides of the equation on the left 

279: with each 

280:  $\psi'_m$   we have

281: \be 

282: 0=\sum_{n=1}^{N} c'_n \la \psi'_m , \psi'_n \ra,

283: \ee

284: that we can recast as 

285: \be

286: \label{ld2}

287: 0= G \vec{c'}

288: \ee

289: where $G$ is a matrix of elements 

290: $g_{m,n}= \la \psi'_m , \psi'_n \ra \,,n,m= 1,\ldots,N$ and

291: $\vec{c'}$ a vector, the component of which are 

292: the coefficients $c'_n,\, n= 1,\ldots,N$. 

293: Equations \eqref{ld} and \eqref{ld2} imply that the 

294: general form for \eqref{fd}  is

295: \be

296: \label{eq}

297: f=\sumn c_n \psi'_n + \sumn c'_n \psi'_n,

298: \ee

299: with $c_n, \,n=1,\ldots,N$ given in \eqref{fd} and 

300: $c'_n,\, n=1,\ldots,N$ the components of a vector 

301: $\vec{c'} \in$ null($G$). 

302: 

303: It is appropriate to stress the significance of 

304: Proposition~\ref{p1} when the interval $\Li$ is 

305: actually the whole real line. Then, for numerical 

306: calculations one is obliged 

307: to work on a finite domain. An important consequence of this 

308: fact will be discussed in the next section.

309: 

310: \section{On the truncation of the harmonic Oscillator 

311: wavepacket}

312: We show here that the results of the previous section 

313: are relevant to the analysis of the truncation of the

314: wavepacket of the harmonic oscillator. To this end 

315: we simulate two different situations, which are specially 

316: devised to illustrate the phenomenon we wish to 

317: discuss. 

318: 

319: Let us consider that a normalised to unity function 

320: $\Psi(x) \in \h$ is generated as linear 

321: superposition of the harmonic oscillator eigenfunctions, 

322: i.e., 

323: \be

324: \Psi(x)=

325: \sum_{n=1}^{N} c_n\frac{e^{-0.5x^2} H_n(x)}

326: {\sqrt{2^{n-1}(n-1)!\sqrt{\pi}}}, \,\,\,\,\,\,n,m=1,\ldots,N,

327: \label{wp}

328: \ee

329: where we have written 

330: $\psi_n(x) =\frac{e^{-0.5x^2}H_n(x)}{\sqrt{2^{n-1}(n-1)!\sqrt{\pi}}}$ in terms of the 

331:  Hermite polynomial $H_n(x)$ . The 

332: coefficients $c_n,\, n=1,\ldots,N$ in (\ref{wp}) 

333: are simulated according to the equation:

334: \be

335: \label{cn}

336: c_n=\frac{e^{-0.0032(n-80)^2}}{\sqrt{\sum_{n=1}^N e^{-0.0064(n-80)^2}}}.  

337: \ee

338: We consider $N=160$. 

339: The left graph of Figure~\ref{wp1} depicts 

340: the corresponding function for $x\in[-1,30]$. 

341: Since the values $\Psi(x)$ are 

342: very small outside this interval ($\Psi(-1)=8.7 \times 10^{-11}$ 

343: and $\Psi(30)= 2.7 \times 10^{-92}$) one does not commit a significant 

344: error by calculating expectation values using this domain. In order to 

345: illustrate this fact let us calculate $\ov{x}$ and $\ov{x^2}$ 

346: as

347: \ben

348: \ov{x}&=& \int_{-40}^{40} |\Psi(x)|^2 x\,dx = 12.56967570231863 

349: \nonumber \\

350: \ov{x}&= &\int_{-1}^{30} |\Psi(x)|^2 x\,dx = 12.56967570231863

351: \een

352: \ben

353: \ov{x^2}&= &\int_{-40}^{40} |\Psi(x)|^2 x^2\,dx = 158.4912423027778

354: \nonumber \\

355: \ov{x^2}&= &\int_{-1}^{30} |\Psi(x)|^2 x^2\,dx = 158.4912423027777.

356: \een

357: The difference in the values of  $\ov{x}$ 

358: cannot be observed in the given format and the values of $\ov{x^2}$

359: differ only in the last of the 16 digits.

360: Thus one could conclude that, 

361: for the purpose of computing expectation values, neglecting the   

362: tails of the distribution outside the interval $x\in[-1,30]$  is not harmful. 

363: However, the assumption that $|\Psi(x)|=0$ for 

364: $x \notin [-1,30]$ has a tremendous consequence: 

365: the coefficients of the  superposition (\ref{wp}) 

366: are thereby not unique.

367: In order to illustrate this we compute the vectors 

368: in the null space of 

369: matrix $G$ of elements 

370: \be

371: g_{m,n}= \int_{-1}^{30} H_m(x)H_n(x)\frac{e^{-x^2}}

372: {\sqrt{2^{m-1}(m-1)! 2^{n-1}(n-1)!}\pi}\,dx.

373: \ee

374: We use just one of the  eigenvectors spanning null$(G)$, 

375: say the vector $\vec{c'}$,  to construct the coefficients 

376: $c''_n=c_n + c'_n, \,n=1,\ldots,160$, with $c_n$ as in \eqref{cn} 

377: and $c'_n, n=1,\ldots,160$ the components of $\vec{c'} \in$ null$(G)$.

378: The right graph in Figure~\ref{wp1} plots the coefficients $c''_n$. 

379: We use now these coefficients to construct the function $\Psi''(x)$.  

380: By calculating 

381: \be

382: ||\Psi'' - \Psi|| =\sqrt{\int_{-1}^{30}|\Psi''(x) - \Psi(x)|^2\,dx}=

383:  1.4198\times 10^{-18}

384: \ee

385: we do not see any significant difference in the functions. 

386: However $||\vec{c''}-\vec{c}||=||\vec{c'}||=1$, which clearly shows that 

387: the  function are considering can be generated in

388: many different ways on the interval $[-1,30]$. Even considering the 

389: interval $[-10,40]$ null($G$) is still not empty. It would be empty for 

390: $[-15,40]$, though, but only  if the maximum 

391: number of states $N=160$ is maintained fixed.  

392: 

393: It should be noted that, although the `critical' interval 

394: depends on the number $N$ and a smaller value of $N$ would 

395: decrease the length of the critical interval, 

396: there is room for adjustments. 

397: Indeed, by considering zero the coefficients which do not 

398: intervene in the original superposition,

399: but allowing the transformation to have larger dimension, the critical interval is enlarged. 

400: Notice that this opens the possibility of producing the 

401: identical function by means of states that were not 

402:  present in the original superposition. The next example illustrates 

403: this situation. 

404: 

405: Consider that the coefficients of the  superposition 

406: (\ref{wp}) are now  simulated as 

407: \be

408: \label{cn2}

409: c_n=\frac{e^{-n}}{\sqrt{\sum_{n=1}^{20}e^{-2n}}}, \,\,\,\,\,\,\, n=1,\ldots,20

410: \ee

411: The corresponding function $\Psi(x)$ is plotted in the left graph of 

412:  Figure~\ref{wp2}. In this case the 

413: critical interval yielding lack of uniqueness is included in the 

414: main support of $\Psi(x)$. Hence, the restriction to such an interval 

415: is not possible. Now, increasing the value of $N$ to 130 for instance, 

416: and considering 

417: $c_n=0,\,n=31,\ldots,130$, the  function $\Psi(x)$ does not change. 

418: Nevertheless,  

419: we have a $130 \times 130$ matrix  $G$  constructed by  

420: extending the interval to one containing the most significant support of 

421:  $\Psi(x)$ (in this case $[-7\,,\,10]$). 

422: Taking one of the vectors in null$(G)$ we

423: construct the coefficients $\vec{c''}$ 

424: by an equivalent process as in the 

425: previous example. With this coefficients, 

426: plotted in  the right 

427: graph of Figure~\ref{wp2}, we 

428: construct the  graph on the left of Figure~\ref{wp2}. 

429: Notice that, the fact that coefficients $c''_n$ have significant values

430: for $n=31,\ldots,130$, implies that  $\Psi(x)$  

431: can be realized by states which were not  present in the  

432: original superposition. Nevertheless, the average energy is the same. 

433: Indeed,  

434: \be

435: \ov{E}= \la \psi'', \hat{H} \psi'' \ra=\sum_{n=1}^{N} \sum_{m=1}^{N} 

436: (c_n^{\ast} +{c'_n}^{\ast})\la \psi'_n, \hat{H} (c_m +c'_m) \psi'_m \ra

437: \ee

438: and, since by hypothesis $\sum_{m=1}^{N} c'_m \psi'_m,=0$, 

439: it follows that

440: \be

441: \ov{E}= \sum_{n=1}^{N} \sum_{m=1}^{N} c_n^{\ast} \la \psi'_n, \hat{H}

442: c_m \psi'_m \ra =  \la \psi',  \hat{H} \psi' \ra.

443: \ee

444: This result stresses the point that was 

445:  made initially. Namely, that without normalising the 

446: coefficients, {\em there  could be infinitely  many 

447: different ways of realizing the harmonic

448: oscillator wavepacket on a finite interval}. 

449: However, according to quantum mechanics

450:  each $|c''_n|^2$ represents the probability of 

451: finding the harmonic oscillator in the state $n$. Hence, to 

452:  be able to maintain this interpretation we must 

453:  impose the normalisation 

454: condition on the coefficients $c''_n$. As a consequence, we 

455: cannot use a vector $\vec{c'}$ of arbitrary norm. In fact, 

456: in order to construct a normalised vector 

457:  we need to consider $\vec{c''}= \vec{c}+D \vec{c'}$,

458:  with $\vec{c'}\in$ null$(G)$

459: and $D$ a constant to be determined by the condition 

460: $||\vec{c''}||^2=1$. Writing this condition explicitly 

461: we have

462: \be

463: ||\vec{c''}||^2= ||\vec{c}||^2  + D\la \vec{c},\vec{c'} \ra + 

464: D^\ast\la \vec{c'},\vec{c} \ra 

465: +|D|^2||\vec{c'}||^2 =1,

466: \ee

467: and, since $\vec{c'}$ and $\vec{c}$ are orthogonal to

468: each other we further have

469: \be

470: ||\vec{c''}||^2= ||\vec{c}||^2 + |D|^2||\vec{c'}||^2=1.

471: \ee

472: The value of $||\vec{c}||^2$ is fixed by the function 

473: $\Psi(x)$, in our case $||\vec{c}||^2=1$. 

474: Thus, the only possible solution to the above

475: equation is $D=0$.  This leads to  the conclusion that 

476:  amongst all the possible superposition 

477: giving rise to the  same 

478: function $\Psi(x)$ on the interval,

479: there is only {\em one} which is  consistent

480: with the physical 

481: significance of quantum mechanics. 

482: 

483: \section{Conclusions}

484: An interesting result, arising by limiting the domain of a

485: normalised function which is the superposition

486: of the harmonic oscillator eigenfunctions,

487: has been discussed. It was shown that,

488: when restricted to a finite interval,

489: such a function can be realized in many different ways.

490: Although the mean values of the

491: physical quantities are not affected in any significant

492: manner,  they can be the result of infinitely many

493: different combinations of eigenfunctions. Uniqueness is

494: restored by endowing the function with the significance of a

495: wavepacket. It was proved that, in that case, 

496: there is only {\em one} set of coefficients that can 

497: fulfil the normalisation condition.

498: 

499: \begin{figure}[h!]

500: \begin{center}

501: \mbox{\includegraphics[angle=0,width=6.5cm]{psi1.eps}

502: \includegraphics[angle=0,width=6.5cm]{coep1.eps}}

503: \caption{\small{The graph on the left 

504: represents the function $\Psi$ 

505: constructed by using the coefficients given in \eqref{cn} 

506: and also the coefficients $c''_n$ plotted

507: in the right graph.}}

508: \label{wp1}

509: \end{center}

510: \end{figure}

511: 

512: \begin{figure}[h!]

513: \begin{center}

514: \mbox{\includegraphics[angle=0,width=6.5cm]{psi2.eps}

515: \includegraphics[angle=0,width=6.5cm]{coep2.eps}}

516: \caption{\small{The graph on the left represents the function $\Psi$

517: constructed by using the coefficients given in \eqref{cn2}

518: and also the coefficients $c''_n$ plotted in the right graph.}}

519: \label{wp2}

520: \end{center}

521: \end{figure}

522: 

523: \begin{thebibliography}{99}

524: %\begin{references}

525: \bibitem{da} R. J. Duffin, A. C. Shaffer, ``A Class of Nonharmonic Fourier

526:  Series,''  Trans. Amer. Math. Soc. {\bf {72}}, 341-366 (1952).

527: \bibitem{do} I. Daubechies, A. Grossmann, Y. Meyer, ``Painless nonorthogonal

528: expansions'', Journal of Mathematical Physics, {\bf{27}}, 1271-1283 (1986).

529: \bibitem{wa} C. Heil, D. Walnut, ``Continuous and discrete wavelet transforms,''

530:  SIAM Rev. {\bf{31,}} 628-666 (1989).

531: \bibitem{ka1} G. Kaiser,

532:  {\em Quantum Physics, Relativity and Complex Space time:

533: Towards a New Synthesis }  (North-Holland, Amsterdam, 1990).

534: \bibitem{ali3}S. T. Ali, J. P. Antoine, J. P. Gazeau,

535: ``Square integrability of group representation on homogeneous spaces.

536: II. Coherent and quasi-coherent states. The case of the Poincar\'e group,''

537:   Ann. Inst. Henri Poincar\'e {\bf {55,}} 857-890 (1991).

538: \bibitem{ali1} S. T. Ali, J. P. Antoine, J. P. Gazeau, ``Continuous

539: Frames in Hilbert Space,''  Annals of Physics {\bf {222,}} 1-37 (1993).

540: \bibitem{do2} I. Daubechies, ``The Wavelets Transform, Time Frequency

541: Localization  and Signal Analysis,'' IEEE Trans. Inform. Theory

542: {\bf 36}, 961-1005 (1990).

543: \bibitem{han} D. Han, D. R. Larson,

544: {\em Frames, Bases and Group Representations},

545: (Memoirs of the  American Mathematical Society, Number 697, 2000)

546: \bibitem{alib} S. T. Ali, J. P. Antoine, J. P. Gazeau,

547: ``Coherent States, Wavelets and their generalizations'', 

548: (Springer, London, 2000).

549: \bibitem{rewf}  L. Rebollo-Neira, A. Plastino,

550:  ``The cross Wigner distribution as a generator of frames on the

551: the Euclidean plane from frames on the real line'',

552: Journal of Physics A: Mathematical and General,

553:  Vol(33, 15), 3053-3061, (2000).

554: \bibitem{rs}L. Rebollo-Neira,``Frames in two dimensions arising

555: from wavelet transforms'', Proceedings of the Royal

556: Society, series A, {\bf{457}}, 2013, 2079-2091. (2001).

557: \bibitem{yo} R. M. Young, {\em An introduction to Nonharmonic Fourier

558:  Series} (Academic Press, New York, 1980).

559: \bibitem{ftf} K. Gr\"ochening,

560: {\em Foundations of time-frequency analysis},

561: (Birkh\"auser, Boston, 2000)

562: \bibitem{c2}O. Christensen,  {\em An Introduction to

563: Frames and Riesz Bases},  (Birkh\"auser, Berlin, 2002)

564: \end{thebibliography}

565: %\end{references}

566: \end{document}

567: