1: %&LaTeX

2: \documentclass[12pt]{iopart}

3: \usepackage[]{indentfirst}

4: \usepackage{amssymb}

5: 

6: 

7: \begin{document}

8: 

9: 

10: 

11: \begin{titlepage}

12: 

13: 

14: \title{Quantum Singularity of Levi-Civita Spacetimes}

15: 

16: \author{D A Konkowski\dag, T M  Helliwell\ddag,   and C Wieland\ddag}

17: 

18: \address{\dag Department of Mathematics, U.S. Naval Academy, Annapolis,

19: Maryland, 21402 U.S.A.}

20: 

21: \address{\ddag Department of Physics, Harvey Mudd College,

22: Claremont, California 91711 USA}

23: 

24: \eads{\mailto{dak@usna.edu}, \mailto{T\_Helliwell@HMC.edu}}

25: 

26: 

27: 

28: \begin{abstract}

29:         Quantum singularities in general relativistic spacetimes are

30:         determined by the behavior of quantum test particles. A static

31:         spacetime is quantum mechanically singular if the spatial

32:         portion of the wave operator is not essentially self-adjoint.

33:         Here Weyl's limit point-limit circle criterion is used to

34:         determine whether a wave operator is essentially self-adjoint.

35:         This test is then applied to scalar wave packets in

36:         Levi-Civita spacetimes to help elucidate the physical

37:         properties of the spacetimes in terms of their metric parameters.

38: \end{abstract}

39: 

40: \pacs{04.20 Dw, O4.62.+v, 03.65 Db}

41: 

42: \maketitle

43: 

44: \end{titlepage}

45: 

46:  \section{Introduction}

47:  In classical general relativity, singularities are not part of the

48: spacetime;

49:  they are boundary points indicated by incomplete geodesics or

50:  incomplete curves of bounded acceleration in a maximal

51:  spacetime (see, e.g. \cite{HE, ES}). So at least for timelike

52:  and null geodesics, this incompleteness can be considered as the

53:  abrupt ending of classical particle paths.

54: 

55:  \par What happens if, instead of classical particles, one uses quantum

56:  mechanical particles to identify singularities? This is the question G.

57:  Horowitz and D. Marolf \cite{HM} set out to answer following earlier

58:  work by R. Wald \cite{Wald}. They call a spacetime quantum

59:  mechanically nonsingular if the evolution of a test scalar wave

60:  packet in the spacetime, representing a quantum mechanical particle,

61:  is uniquely determined by the initial wave

62:  packet, without having to place arbitrary boundary conditions at the

63:  classical singularity. Their construction is restricted to

64:  static spacetimes.

65: 

66:  \par Mathematically, this evolution is related to properties of a

67:  quantum mechanical operator. A static spacetime is quantum mechanically

68:  singular \cite{HM} if the spatial portion of the Klein-Gordon wave

69: operator is not

70:  essentially self-adjoint \cite{RS} \cite{Rich}. If this operator is not

71:  essentially self-adjoint then the evolution of a test scalar wave

72:  packet is not determined uniquely by the initial wave packet;

73:  boundary conditions at the classical singularity are needed to ``pick

74:  out'' the correct wave function and thus one needs to add information

75:  that is not already present in the wave operator, spacetime metric,

76:  and manifold. This will be described further in the paragraphs below.

77: 

78:  \par Horowitz and Marolf restrict the wave operator to the Klein-Gordon

79:  operator, but it has been shown \cite{HKA} that Maxwell and Dirac operators

80:  may also be used with equivalent generic results. Throughout this paper

81:  we will restrict the analysis to Klein-Gordon operators.

82: 

83:  \par A relativistic quantum particle with mass $M$ can be described

84:  by a positive frequency solution to the Klein-Gordon equation

85: 

86:  \begin{equation}

87:      \frac{\partial ^{2} \Psi}{\partial t^{2}} = -A \Psi

88:      \label{eq:1}

89:  \end{equation}

90: 

91: 

92: \noindent in a static spacetime \cite{HM}. The spatial Klein-Gordon

93: operator $A$

94: is

95: 

96: \begin{equation}

97:     A \equiv -V D^{i}(V D_{i}) + V^{2} M^{2}

98:     \label{eq:2}

99: \end{equation}

100: 

101: \noindent where $V^{2} = - \xi_{\nu}\xi^{\nu}$ (here $\xi^{0}$ is the

102: timelike

103: Killing field) and $D_{i}$ is the spatial covariant derivative on the

104: static slice $\Sigma$. The appropriate Hilbert space $H$ is

105: $\mathcal{L}^{2}(\Sigma)$, the space of square integrable functions on

106: $\Sigma$.

107: This choice agrees with that of Horowitz and Marolf \cite{HM}.  The

108: volume element used to define $H$ is $V^{-1}$ times the natural volume

109: element on $\Sigma$. We do

110: not choose the first Sobolev norm $H^{1}$ proposed by Ishibashi and

111: Hosoya \cite{IH}; that choice is related to Dirichlet boundary

112: conditions at the singularity \cite{Wald}. If we initially define

113: the domain of $A$ to be $C_{0}^{\infty}$, $A$ is a real positive

114: symmetric operator and self-adjoint extensions always exist \cite{RS}.

115: If there is one unique self-adjoint extension $A_{E}$, then $A$ is

116: essentially self-adjoint \cite{RS}. In this case the Klein-Gordon

117: equation for a free relativistic particle takes the form \cite{HM}

118: 

119: \begin{equation}

120:     i \frac{\partial \Psi}{\partial t} = (A_{E})^{1/2} \Psi

121:     \label{eq:3}

122: \end{equation}

123: 

124: \noindent with

125: 

126: \begin{equation}

127:    \Psi (t) = \exp ( -i t (A_{E}^{1/2}) \Psi (0).

128:     \label{eq:4}

129: \end{equation}

130: 

131: \noindent Equations (\ref{eq:3}) and (\ref{eq:4}) are ambiguous if $A$ is

132: not essentially

133: self-adjoint. This fact led Horowitz and Marolf \cite{HM} to define

134: quantum-mechanically singular spacetimes as those in which $A$ is not

135: essentially self-adjoint.

136: 

137: \par In their paper Horowitz and Marolf test several classically

138: singular spacetimes to determine whether they are quantum mechanically

139: singular as well. They find that Reissner-Nordstr\"{o}m, negative mass

140: Schwarzschild, and the $2D$ cone remain singular when probed by

141: quantum scalar test particles; however, certain orbifolds, extreme

142: Kaluza-Klein black holes, the $D=5$ fundamental string, and a few other

143: examples are nonsingular. Helliwell and Konkowski \cite{HK} and

144: Helliwell, Konkowski and Arndt \cite{HKA} show that a broad class of

145: quasiregular spacetimes (spacetimes with quasiregular singularities)

146: are quantum mechanically singular. This work was an extension of

147: earlier work by Kay and Studer \cite{KS} which showed that the $2D$ cone and

148: a $4D$ idealized cosmic string are not essentially self-adjoint.

149: 

150: \par The work of Horowitz and Marolf \cite{HM} and subsequent work by

151: Helliwell, Konkowski and Arndt \cite{HKA} test for essential

152: self-adjointness of the spatial

153: operator $A$ by using a von Neumann criterion \cite{VN, weyl} . This method

154: involves studying solutions to

155: 

156: \begin{equation}

157:     A \Psi = \pm i \Psi

158:     \label{eq:5}

159: \end{equation}

160: 

161: \noindent and finding the number of solutions that belong to

162: $\mathcal{L}^{2}(\Sigma)$

163: for each sign of $i$. This determines the von Neumann deficiency indices

164: which indicate whether the operator is essentially self-adjoint or

165: whether it has self-adjoint extensions, and how many self-adjoint

166: extensions it has \footnote{For an

167: introduction to the mathematics of self-adjoint operators and von

168: Neumann deficiency indices, see \cite{HKA} which is based on \cite{RS, IH,

169: BFV}} . If

170: the deficiency indices are $(0,0)$, so that no solutions are square

171: integrable, then the operator is essentially self-adjoint and so has

172: a unique self-adjoint extension. The wave behavior described by the

173: operator is uniquely determined for all time by the operator, the

174: spacetime metric, and the manifold. No additional information in the

175: form of boundary conditions need be added.

176: 

177: \par In this paper we use an alternative test for quantum

178: singularity based upon the limit point-limit circle criterion for

179: essential self-adjointness of operators. That is, we use a theorem of Weyl's

180: \cite{RS} to relate the essential self-adjointness of the operator via

181: the von Neumann deficiency indices to the ``potential'' which

182: determines the behavior of the scalar wave packet. The effect is

183: determined by the limit point-limit circle criterion

184: which we will discuss in Section 3. After developing this technique in

185: the context of determining the quantum singularity nature of a general

186: spacetime, we

187: study the so-called Levi-Civita spacetimes in particular.

188: 

189: \par Levi-Civita spacetimes are static, primarily cylindrically symmetric,

190: spacetimes that are classically singular at ``$r=0$'' unless the

191: metric is Minkowski or Minkowski in accelerated coordinates. They are

192: used to model infinite line masses and idealized cosmic strings.

193: There has been some controversy in the literature about the physical

194: relevance of certain parameters; this will be reviewed in Section 2.

195: The quantum singularity or nonsingularity of $r=0$ will be studied in

196: Section 4 (following the mathematical background in Section 3).  Finally

197: in Section 5

198: we use the quantum singularity results

199: for certain parameter values to glean what information we can about the

200: physical

201: interpretation of the corresponding spacetimes.

202: 

203: \section{Levi-Civita Spacetimes}

204: 

205: The metric in Levi-Civita spacetimes \cite{LC} has the form

206: 

207: \begin{equation}

208:     ds^{2} = r^{4 \sigma}dt^{2} - r^{8 \sigma^{2} - 4\sigma}(dr^{2}+

209:     dz^{2}) - \frac{r^{2 - 4 \sigma}}{C^{2}} d\theta^{2}

210:     \label{eq:6}

211: \end{equation}

212: 

213: \noindent where $\sigma$ and $C$ are real numbers ($C>0$).  It is

214: straightforward to show that if $\sigma$ is replaced by $1/4\sigma$

215: in the metric, the spacetime is unchanged, although a coordinate

216: transformation is required to make this obvious.  Therefore we

217: need only study the range $-\frac{1}{2}\le \sigma\le\frac{1}{2}$.

218: Neither of the parameters $\sigma$ and $C$ can

219: be removed by a coordinate transformation \cite{bonnor,

220: HSTW, HRS}.  Computation of the Kretschmann scalar

221: 

222: \begin{equation}

223:     R_{\mu \nu \sigma \tau}R^{\mu \nu \sigma \tau} = \frac{64

224:     \sigma^{2} (2\sigma -1)^{2}}{(4 \sigma^{2} -2\sigma +1)^{3} r^{4}}

225:     \label{eq:7}

226: \end{equation}

227: 

228: \noindent shows that $r=0$ is a scalar curvature singularity for all

229: $\sigma $

230: except $\sigma = 0, 1/2$.

231: 

232: \par If $\sigma = 0$, the metric is

233: 

234: \begin{equation}

235:     ds^{2} = -dt^{2} + dr^{2} +dz^{2} + \frac{r^{2}}{C^{2}} d \theta^{2}.

236:     \label{eq:8}

237: \end{equation}

238: 

239: \noindent If $C=1$, this is simply Minkowski spacetime in

240: cylindrical coordinates. It thus seems reasonable to choose the

241: coordinate ranges

242: 

243: \begin{equation}

244:     -\infty < t < \infty,\,  -\infty < z < \infty,\,

245:     0 < r < \infty, \,  0  \le \theta \le 2 \pi,

246:     \label{eq:9}

247: \end{equation}

248: 

249: \noindent with $0$ and $2\pi$ identified. If $C \not= 1$, equation

250: \ref{eq:8} is the

251: metric for an idealized cosmic string. There is a quasiregular

252: (``disclination'') singularity at $r=0$ (see, e.g., \cite{HKA} for a

253: discussion). This is a

254: topological singularity, not a curvature singularity. It is thus

255: reasonable to identify the parameter $C$ in the general metric,

256: equation (6), with a topological property of the spacetime, that is,

257: the deficit angle.

258: 

259: \par The sigma parameter is less easily interpreted \cite{bonnor,

260: HSTW, HRS}. For certain values one can interpret these

261: spacetimes as those of infinite line masses. This

262: interpretation is clearest for $|\sigma|\ll 1$. The most physically

263: realistic range is $0<\sigma<1/4$,  where $\sigma$

264: represents the mass per unit length and timelike circular orbits exist.

265: As $\sigma$ increases from 1/4 to 1/2 the interpretation is more

266: difficult \cite{bonnor, HSTW, HRS}. In this range, the Kretschmann

267: scalar decreases with increasing $\sigma$, which seems to suggest that the

268: gravitational field is becoming weaker. But as several authors comment

269: \cite{bonnor, HSTW, HRS}, the Kretschmann scalar may not be a good

270: indicator of gravitational

271: strength; the acceleration of test particles, increasing with

272: increasing $\sigma$ in the interval $(1/4, 1/2)$, may be a better

273: indicator.  Interior solutions matching to exterior Levi-Civita spacetimes

274: exist for the entire range $0<\sigma<1/2$. Bonnor \cite{bonnor}

275: proposes to interpret Levi-Civita as the spacetime generated by a

276: cylinder whose radius increases with increasing $\sigma$ and tends to

277: infinity as $\sigma$ tends to $1/2$. Thus the entire range

278: $0<\sigma<1/2$ can be taken to represent an ``infinite line mass''.

279: 

280: 

281: \par The $\sigma = 1/2$ case is more problematic. Bonnor's analysis

282: suggests (but does not prove) that when $\sigma = 1/2$ the cylinder

283: becomes a plane, an interpretation first put forward by Gautreau and

284: Hoffmann \cite{GH} based on different considerations. In fact, the $\sigma

285: =1/2$ metric

286: 

287: \begin{equation}

288:     ds^{2} = -r^{2} dt^{2} + dr^{2} + dz^{2} + \frac{1}{C^{2}}

289:     d\theta^{2}

290:     \label{eq:10}

291: \end{equation}

292: 

293: \noindent is flat. We can transform to Minkowski coordinates

294: 

295: \begin{equation}

296:     ds^{2} = -d{\bar t}^{2} + d{\bar x}^{2} + d{\bar y}^{2} + d{\bar

297:     z}^{2}

298:     \label{eq:11}

299: \end{equation}

300: 

301: \noindent if we let

302: ${\bar t} = r \sinh t$, ${\bar x} = r \cosh t$, ${\bar y} = \theta/C$,

303: and ${\bar z} = z$. Here we can obviously let the ${\bar y}$ coordinate range

304: from $-\infty$ to $\infty$. This is flat spacetime described from the

305: view of an accelerating frame of reference. This seems to support the

306: interpretation of the $\sigma = 1/2$ case as a planar source producing

307: flat spacetime described by a uniformly accelerating observer

308: \cite{bonnor}. In other words, one can interpret it as the spacetime of a

309: gravitational field produced by an

310: infinite planar sheet of positive mass density.

311: 

312: \par On the other hand, one can interpret the metric with $\sigma = -1/2$

313: as the

314: gravitational field produced by an infinite sheet of negative mass

315: density (an idea also first proposed by Gautreau and Hoffmann

316: \cite{GH}). However, this case is also problematic \cite{bonnor, HSTW,

317: HRS}. The metric

318: 

319: \begin{equation}

320:     ds^{2} = - \frac{1}{r^{2}} dt^{2} + r^{2} (dr^{2} + dz^{2} +

321:     \frac{1}{C^{2}} d \theta^{2})

322:        \label{eq:12}

323: \end{equation}

324: 

325: \noindent is not flat and it has a curvature singularity at $r=0$. It

326: has planar symmetry with one additional extra Killing vector beyond

327: those in the general

328: Levi-Civita spacetimes. The interpretation of this

329: spacetime as one caused by an infinite plane of negative mass density

330: comes from the fact that test particles are repelled from the $r=0$

331: plane and the fact that ``the Gaussian curvature of spacelike

332: `eigensurfaces' is

333: zero'' \cite{GH}. This is a reasonable interpretation; the

334: ``problem'' comes from interpreting both $\sigma = 1/2$ and $\sigma

335: =-1/2$ as planes of infinite (positive/negative) mass density as one

336: is singularity-free and the other has a curvature singularity

337: \cite{HSTW, HRS}.

338: 

339: \par We will now study the quantum singularity properties of these

340: spacetimes for various parameter values in the following sections. We

341: begin by introducing the mathematics necessary to discuss essentially

342: self-adjoint operators and quantum singularities.

343: 

344: \section{Weyl's Limit Point - Limit Circle Criterion}

345: 

346: \par  A particularly convenient way to establish essential

347: self-adjointness in the spatial operator of the Klein-Gordon equation

348: is to use the concepts of limit circle and limit point behavior.\footnote

349: {This section is based on {\bf Appendix to X.1} in Reed and

350: Simon \cite{RS}}  The

351: approach is as follows.  The Klein-Gordon equation for Levi-Civita

352: spacetimes can be separated in the coordinates $t, r, \theta, z$.

353: Only the radial equation is non-trivial.  With changes in both

354: dependent and independent variables, the radial equation can be

355: written as a one-dimensional Schr\"{o}dinger equation

356: 

357: \begin{equation}

358:     H\Psi(x) = E\Psi(x)

359:     \label{eq:13}

360: \end{equation}

361: 

362: \noindent where $x \in (0,\infty )$ and the operator  $H = - d^{2}/dx^{2}

363: + V(x)$.

364: 

365: \newtheorem{Definition}{Definition}

366: \begin{Definition} The potential $V(x)$ is in the limit circle case

367: at $x = 0$  if for some, and therefore for all $E$, {\it all}

368: solutions of equation (\ref{eq:13}) are square integrable at zero.  If

369: $V(x)$  is not

370: in the limit circle case, it is in the limit point case.

371: \end{Definition}

372: 

373: \par  A similar definition pertains to $x=\infty$.  The potential

374: $V(x)$ is in the limit circle case at $x=\infty$ if all solutions of

375: equation (\ref{eq:13}) are square integrable at infinity; otherwise,

376: $V(x)$ is in

377: the limit point case at infinity.

378: 

379: \par There are of course two linearly independent solutions of the

380: Schr\"{o}dinger equation for given $E$.  If $V(x)$ is in the limit circle

381: case at zero, both solutions are $\mathcal{L}^{2}$ at zero, so all linear

382: combinations are $\mathcal{L}^{2}$ as well.  We would therefore need a

383: boundary

384: condition at $x=0$ to establish a unique solution.  If $V(x)$ is in

385: the limit {\it point} case, the $\mathcal{L}^{2}$ requirement eliminates

386: one of

387: the solutions, leaving a unique solution without the need of

388: establishing a boundary condition at $x=0$. This is the whole idea of

389: testing for quantum singularities; there is no singularity if the

390: solution is unique, as it is in the limit point case.  The critical

391: theorem is due to Weyl \cite{RS}.

392: 

393: \newtheorem{theorem}{Theorem}

394: \begin{theorem}[The Weyl limit point-limit circle criterion.] If

395: $V(x)$ is a continuous real-valued function on $(0, \infty)$, then

396: $H = - d^{2}/dx^{2} + V(x)$ is essentially self-adjoint on

397: $C_{0}^{\infty}(0, \infty)$ if and only if $V(x)$ is in the limit

398: point case at both zero and infinity.

399: \end{theorem}

400: 

401: \par The following theorem can be used to establish the limit circle-limit

402: point behavior at infinity \cite{RS}.

403: 

404: \newtheorem{theorem1}[theorem]{Theorem}

405: \begin{theorem1}[Theorem X.8 of Reed and Simon \cite{RS}.] If $V(x)$ is

406: continuous and real-valued on $(0, \infty)$, then $V(x)$ is in the

407: limit point case at infinity if there exists a {\em positive}

408: differentiable function $M(x)$ so that

409: \begin{enumerate}

410:     \item[(i)] $V(x) \ge - M(x)$

411:     \item[(ii)] $\int_{1}^{\infty} [M(x)]^{-1/2} dx = \infty$

412:     \item[(iii)] $M'(x)/M^{3/2}(x)$ is bounded near $\infty$.

413: \end{enumerate}

414: Then $V(x)$ is in the limit point case (complete) at $\infty$.

415: \end{theorem1}

416: 

417: A sufficient choice of the $M(x)$ function for our purposes is the

418: power law function $M(x) = c x^{2}$ where $c > 0$. Then {\it (ii)} and

419: {\it (iii)} are satisfied, so if $V(x) \ge -c x^{2}$, $V(x)$ is in the

420: limit point case at infinity.

421: 

422: \par A theorem useful near zero is the following.

423: 

424: \newtheorem{theorem2}[theorem]{Theorem}

425: \begin{theorem2} [Theorem X.10 of Reed and Simon \cite{RS}.]

426: Let $V(x)$ be continuous and {\it positive} near zero.

427: If $V(x)\ge\frac{3}{4} x^{-2}$ near zero then $V(x)$ is in the limit

428: point case.  If for some $\epsilon > 0$,

429: $V(x)\le(\frac{3}{4}-\epsilon)x^{-2}$ near zero, then $V(x)$ is in the

430: limit circle case.

431: \end{theorem2}

432: 

433: \noindent These results can now be used to help test for quantum

434: singularities in the Levi-Civita spacetimes.

435: 

436: \section{Limit Point - Limit Circle properties of Levi-Civita Spacetimes}

437: 

438: \par The Klein-Gordon equation for a scalar particle of mass $M$ is

439: 

440: \begin{equation}

441:      \Box \Phi \equiv g^{\mu \nu} \Phi_{\,\mu \nu} +

442:      \frac{1}{\sqrt{g}}(\sqrt{g}g^{\mu \nu}),_{\nu} \Phi_{\,\mu} = M^{2}

443:      \Phi ;

444:      \label{eq:14}

445: \end{equation}

446: 

447: \noindent for the Levi-Civita metrics it is

448: 

449: \begin{equation}

450:     \fl -r^{-4 \sigma} \Phi,_{tt} + r^{8 \sigma^{2} + 4 \sigma}

451:     (\Phi,_{rr} + \Phi,_{zz}) +C^{2} r^{4 \sigma - 2}

452:     \Phi,_{\theta\theta} +

453:     r^{-8 \sigma^{2} + 4\sigma - 1} \Phi,_{r} = M^{2} \Phi.

454:     \label{eq:15}

455: \end{equation}

456: 

457: \noindent The equation separates using $\Phi = e^{-i \omega

458: t}e^{ikz}e^{im \theta} R(r)$, where $\omega, k$ are continuous and $m=0,

459: \pm 1, \pm 2, \cdots$. The resulting radial equation is

460: 

461: \begin{equation}

462:     \frac {d^{2} R}{d r^{2}} + \frac{1}{r} \frac{d R}{d r} + \left[

463:     \omega^{2} r^{8 \sigma^{2}- 8 \sigma} - k^{2} - M^{2} r^{8

464:     \sigma^{2} - 4 \sigma}-m^{2} C^{2} r^{8 \sigma^{2} - 2} \right]

465:     R = 0.

466:     \label{eq:16}

467: \end{equation}

468: 

469: \noindent As described in Section 1, square integrability is judged by

470: evaluating

471: the integral

472: 

473: \begin{equation}

474:     I = \int dr \sqrt{g_{3}/g_{00}} R^{*} R = (1/C) \int dr  r^{8

475:     \sigma^{2} - 8 \sigma +1} R^{*} R

476:     \label{eq:17}

477: \end{equation}

478: 

479: \noindent where $g_{3}$ is the negative determinant of the 3-space portion of

480: the metric tensor \cite{Wald, HM}.  The resulting integral is

481: invariant under coordinate transformations.

482: 

483: \par To test for limit point-limit circle behavior one must

484: rewrite the radial equation in the one-dimensional

485: Schr\"{o}dinger-equation form

486: 

487: \begin{equation}

488:     \frac{d^{2} \Psi (x)}{dx^{2}} + \left[E - V(x) \right] \Psi = 0,

489:     \label{eq:18}

490: \end{equation}

491: 

492: \noindent for which square integrability is judged by the integral $I =

493: \int dx

494: \Psi^{*} \Psi$.  Except for the special case $\sigma = 1/2$, which we

495: return to later, the

496: appropriate transformations of dependent and independent variables are

497: $R = (1/\sqrt{x}) \Psi$ and $r = ( \alpha C x^{2})^{1/(2 \alpha)}$

498: where $\alpha \equiv (2 \sigma - 1)^{2}$.

499: 

500: \par Equation (16) transforms to equation (18) if we let $E = C

501: \omega^{2}/ \alpha$ and

502: \begin{eqnarray}

503:     V(x) & = & (C k^{2}/ \alpha) ( \alpha C x^{2})^{(- \alpha + 1)/

504:     \alpha} + (C M^{2}/ \alpha) ( \alpha C x^{2})^{2 \sigma/

505:     \alpha} \nonumber\\

506:       & + & (m^{2} C^{3}/ \alpha) (\alpha C x^{2})^{(4 \sigma - 1)/

507:     \alpha} - 1/(4 x^{2}).

508:     \label{eq:19}

509: \end{eqnarray}

510: 

511: \noindent In the limit $x \rightarrow \infty$ the final term vanishes,

512: leaving

513: a non-negative $V(x)$. Therefore the solutions are limit point at

514: infinity for all $\sigma$ ($\sigma \not= 1/2$), since as $x \rightarrow

515: \infty$, $V(x) > - c x^{2}$ for any positive constant $c$ \cite{RS,

516: Rich}.

517: 

518: \par It remains to test the solutions in the limit $x \rightarrow 0$.

519: This we do by identifying the important terms in $V(x)$. The final term

520: diverges as $x^{-2}$. The first term in $V(x)$ diverges less quickly for

521: any finite  $\sigma$. The second term $\sim x^{4 \sigma/ \alpha}$,

522: whose exponent has the minimum value $-1/2$ at $\sigma = -1/2$. The

523: third term $\sim x^{2( 4 \sigma - 1)/ \alpha}$, whose exponent has the

524: minimum value -2 at $\sigma = 0$.

525: 

526: \par Therefore unless $\sigma = 0$ (or $\sigma = 1/2$, which we

527: treat later), the equation has the form

528: 

529: \begin{equation}

530:     \frac{d^{2} \Psi}{dx^{2}} + \frac{1}{4 x^{2}} \Psi = 0

531:     \label{eq:20}

532: \end{equation}

533: 

534: \noindent as $x \rightarrow 0$.  By Theorem 3 the solutions

535: should exhibit limit circle behavior; in fact, one solution is $\Psi_{1} =

536: x^{1/2}$

537: and the other is $\Psi_{2} = \Psi_{1} \int^{x} dx/

538: \Psi_{1}^{2} = x^{1/2} \ln (x)$. Both are $\mathcal{L}^{2}$

539: near $x = 0$, so are indeed limit circle.

540: 

541: \par If $\sigma =0$, the equation has the form

542: \begin{equation}

543:     \frac{d^{2} \Psi}{d x^{2}} - \frac{m^{2} C^{2} - 1/4}{x^{2}}

544:     \Psi = 0

545:     \label{eq:21}

546: \end{equation}

547: 

548: \noindent as $x \rightarrow 0$. One solution is $\Psi_{1} = x^{1/2 + |m| C}$;

549: the other is $\Psi_{2} = x^{1/2} \ln (x)$ (if $m = 0$), or $

550: \Psi_{2} = x^{1/2 - |m| C}$ (if $m \not= 0$). Both solutions are

551: limit circle if $|m| C < 1$, and limit point if $|m| C \ge 1$.

552: 

553: \par In the special case $\sigma = 0,\, C = 1$, the spacetime is

554: Minkowski with no classical singularity at $r =0$. In that case $x =

555: r$ and $R_{1} = 1, R_{2} = \ln (r)$. The solution $R_{2} = \ln (r)$ is

556: unacceptable in this case because it diverges at $r =0$ even though $r

557: =0$ is a

558: regular hypersurface in the spacetime.  \footnote {We assumed at the

559: outset that

560: the domain of the spatial operator consists

561: of smooth functions with compact support away from the origin.

562: For Minkowski space the points $r = 0$ are arbitrary, so it is

563: unphysical to permit a function that diverges at  $r = 0$  and nowhere else.}

564: If $\sigma = 0,\, C \neq 1$ the spacetime

565: corresponds to a linear idealized cosmic string with a quasiregular

566: singularity at $r = 0$.

567: 

568: \par For the special case $\sigma = 1/2$ the radial Klein-Gordon

569: equation is

570: 

571: \begin{equation}

572:     \frac{d^{2} R}{dr^{2}} + \frac{1}{r} \frac{d R}{dr} +\left[ -(

573:     k^{2} + M^{2} + m^{2} C^{2}) + \omega^{2}/ r^{2} \right] R = 0.

574:     \label{eq:22}

575: \end{equation}

576: 

577: \noindent The transformation $R(r) = \Psi (x)$ and $r = \exp (C x)$ puts the

578: equation in the form

579: 

580: \begin{equation}

581:     \frac{d^{2} \Psi}{dx^{2}} + \left[ E - V(x) \right] \Psi = 0

582:     \label{eq:23}

583: \end{equation}

584: 

585: \noindent where $E = C^{2} \omega^{2}$ and

586: 

587: \begin{equation}

588:     V(x) = C^{2} ( k^{2} + M^{2} + m^{2} C^{2}) \exp (2Cx)

589:     \label{eq:24}

590: \end{equation}

591: 

592: \noindent with  $\int dr \sqrt{g_{3}/g_{00}} R^{*} R = \int dx \Psi^{*} \Psi$.

593: 

594: \par In the transformation $r = 0$ corresponds to $x = - \infty$, while

595: $r = \infty $ corresponds to $x = \infty$. So in this case we must

596: determine the limit point-limit circle behavior at $x = \pm \infty $.

597: In this case $V(x) > -c x^{2}$ at both $\pm \infty $, for any

598: positive $c$. Therefore the solutions are limit point at both ends.

599: It is interesting that the $\sigma = 1/2$ case is a special case

600: which must be studied on the interval $(-\infty, \infty)$.  This result seems

601: to support Bonnor's description of the $\sigma = 1/2$ case as a

602: cylinder whose radius tends to $\infty$ as $\sigma$ tends to $1/2$.

603: 

604: 

605: \section{Conclusions}

606: 

607: \par In the preceding section we showed that the Klein-Gordon operator

608: is limit point at infinity for the whole class of Levi-Civita

609: spacetimes. If $\sigma$ is neither zero nor one-half, the

610: Klein-Gordon operator is limit circle at $r = 0$ and thus not

611: essentially self-adjoint,  so all $\sigma \neq

612: 0, \sigma \neq 1/2$ Levi-Civita spacetimes are quantum mechanically

613: singular as well as being classically singular.

614: 

615: \par If $\sigma = 0$ and $C = 1$, the spacetime is simply Minkowski

616: space. One of the two solutions of the radial Klein-Gordon equation

617: can be rejected because it diverges at a regular point ($r=0$) of the

618: spacetime. The operator is therefore limit point at $r=0$, and

619: so Minkowski spacetime is quantum

620: mechanically nonsingular (a well known fact, repeated here for

621: completeness).

622: 

623: 

624: par If $\sigma = 0$ and $C \neq 1$, the spacetime is the conical

625: spacetime corresponding to an idealized cosmic string. The operator

626: is limit circle if $|m| C < 1$ and limit point if $|m| C \ge 1$.

627: Therefore the cosmic string spacetimes are quantum mechanically

628: singular for azimuthal quantum number $m$ such that $|m| C < 1$ and

629: nonsingular if  $|m| C \ge 1$. If arbitrary values of $m$ are allowed,

630: these

631: spacetimes are quantum mechanically singular in agreement with earlier

632: results \cite {HK, HKA, KS}. Recall that these spacetimes are also

633: classically singular with a quasiregular (``disclination'')

634: singularity at $r=0$.

635: 

636: \par If $\sigma = 1/2$ the classical spacetime is flat and without a

637: classical singularity. We have shown that this spacetime is quantum

638: mechanically nonsingular as well for massive scalar particles obeying the

639: Klein-Gordon equation, since the operator is limit point at both ends

640: and thus essentially self-adjoint.

641: 

642: 

643: \par Thus for the Levi-Civita spacetimes, all that are classically

644: singular are also quantum mechanically singular, and all that are

645: classically nonsingular ($\sigma = 0,\, C = 1$, and $\sigma = 1/2$) are

646: also quantum mechanically nonsingular.  The classically and

647: quantum-mechanically nonsingular spacetimes correspond to isolated

648: values of $\sigma$, so that (for example) even though the spacetime

649: $\sigma = 0,\, C = 1$ is nonsingular, the spacetimes with  $\sigma

650: \to 0,\, C = 1$ are singular.  The only discrepency between

651: classical and quantum singularities are for the $\sigma =0,\, C \neq 1$

652: modes with $|m| C \ge 1$, which are quantum mechanically nonsingular

653: in a classically singular spacetime. The physical reason is that the

654: wavefunction for large values of $m$ in a flat space with a quasiregular

655: singularity at $r=0$ are unable to detect the presence of the

656: singularity because of the repulsive centrifugal potential

657: $m^{2}C^{2}/x^{2}$ in equation (21).  The limit point/limit circle

658: criterion, together with the theorems that connect it to the effective

659: potential in a Schrodinger equation, provide physical insight into

660: when quantum singularities are prevented from occurring by potential

661: barriers.

662: 

663: 

664: \ack We thank David Clarke, Jack Harrison and Cassidi Reese for useful

665: conversations.

666: One of us (DAK) was partially funded by NSF grant PHY99-88607 to

667: the U.S. Naval Academy. She also thanks Queen Mary, University of London,

668: where some of this work was carried out.

669: 

670: \section*{References}

671: 

672: \begin{thebibliography}{99}

673: 

674:     \bibitem{HE} Hawking S W and Ellis G F R 1973 {\it The Large-Scale

675:     Structure of Spacetime} (Cambridge: Cambridge University Press)

676: 

677:     \bibitem{ES} Ellis G F R and Schmidt B G 1977 {\it Gen. Relativ. Grav}

678:     {\bf 8} 915

679: 

680:     \bibitem{HM} Horowitz G T and Marolf D 1995 {\it Phys. Rev.

681:     D} {\bf 52} 5670

682: 

683:     \bibitem{Wald} Wald R M 1980 {\it J. Math Phys.} {\bf 21} 2802

684: 

685:     \bibitem{RS} Reed M and Simon B 1972 {\it Functional Analysis}

686:     (New York: Academic Press)

687: 

688:     \nonum Reed M and Simon B 1972 {\it Fourier Analysis and

689:     Self-Adjointness} (New York: Academic Press)

690: 

691:     \bibitem{Rich} Richmyer R -D 1978 {\it Principles of Advanced

692:     Mathematical Physics, Vol. I} (New York: Springer)

693: 

694:     \bibitem{HKA} Helliwell T M, Konkowski D A and Arndt V 2003 {\it

695:     Gen. Relativ. Grav.} {\bf 35} 79

696: 

697:     \bibitem{IH} Ishibashi A and Hosoya A 1999 {\it Phys. Rev. D} {\bf

698:     60} 104028

699: 

700:     \bibitem{HK} Konkowski D A and Helliwell T M 2001 {\it Gen.

701:     Relativ. Grav.} {\bf 33} 1131

702: 

703:     \bibitem{KS} Kay B S and Studer U M 1991 {\it Commun. Math. Phys.}

704:     {\bf 139} 103

705: 

706:     \bibitem{VN} von Neumann J 1929 {\it Math. Ann.} {\bf 102} 49

707: 

708:     \bibitem{weyl} Weyl H  1910 {\it Math. Ann.} {\bf 68} 220

709: 

710:     \bibitem{BFV} Bonneau G, Faraut J and Valent G 2001 {\it Am. J.

711:     Phys.} {\bf 69} 322

712: 

713:     \bibitem{LC} Levi-Civita T 1919 {\it Rend. Acc. Lincei} {\bf 28}

714:     101

715: 

716:     \bibitem{bonnor} Bonnor W B ``The static cylinder in general relativity''

717:     in {\it On Einstein's Path}  ed. A. Harvey 1999 (New York:

718:     Springer) 113

719: 

720:     \bibitem{HRS} Herrera L, Ruifern\'{a}ndez J and Santos N O 2001 {\it Gen.

721:     Relativ. Grav.} {\bf 33} 515

722: 

723:     \bibitem{HSTW} Herrera L, Santos N O, Teixeira A F F, and Wang A Z

724:     2001 {\it Class. Quantum Grav.} {\bf 18} 3847

725: 

726:     \bibitem{GH} Gautreau R and Hoffmann R B 1969 {\it Nuovo Cimento B}

727:     {\bf 61} 411

728: 

729: 

730: 

731: 

732: 

733: \end{thebibliography}

734: 

735: 

736: 

737: \end{document}

738: 

739: 

740: 

741: