1: \documentclass[12pt,preprint]{aastex}

2: \begin{document}

3: \author{Daniel J. Cross}

4: \affil{Department of Physics, Drexel University}

5: \affil{Philadelphia, PA 19104, USA}

6: \email{d.j.cross@drexel.edu}

7: \title{Comments on the Cooperstock-Tieu Galaxy Model}

8: \begin{abstract}

9: The recently proposed Cooperstock-Tieu galaxy model claims

10: to explain the flat rotation curves without dark matter.  The purpose

11: of this note is to show that this model is internally inconsistent and thus

12: cannot be considered a valid solution.  Moreover, by making the solution

13: consistent the ability to explain the flat rotation curves is lost.

14: \end{abstract}

15: 

16: \keywords{galaxies: kinematics and

17:   dynamics-gravitation-relativity-dark matter}

18: 

19: \section{Introduction}

20: Cooperstock and Tieu model a galaxy in general relativity as an axially

21: symmetric, pressure free dust cloud with metric

22: \begin{equation} \label{metric}

23: ds^2=-e^{w}(cdt-Nd\varphi)^2+r^2e^{-w}d\varphi^2+e^{v-w}(dr^2+dz^2)

24: \end{equation}

25: where $w$, $v$, and $N$ are functions only of $r$ and $z$, thus the

26: metric is stationary \citep{CT}.  They further assume their coordinates to be

27: co-moving with the galactic dust, thus 

28: \begin{equation}

29: u^\mu=e^{-w/2}\delta_t^\mu

30: \end{equation}

31: is the four-velocity.

32: 

33: By performing a local diagonalization of the metric they obtain a

34: relation between the metric and local angular velocity as

35: \begin{equation}

36: \omega=\frac{cNe^w}{r^2e^{-w}-N^2e^w}\approx\frac{cN}{r^2}\label{rot}

37: \end{equation}

38: where the approximate value is appropriate to the order they consider.  The $tt$ and $t\varphi$ Einstein equations are respectively

39: \begin{mathletters}

40: \begin{eqnarray}

41: r^{-2}\left({N_{,r}^2+N_{,z}^2}\right) &=& \frac{8\pi G \rho}{c^2}\label{dens}\\ 

42: N_{,rr}-\frac{1}{r}N_{,r}+N_{,zz}&=&0. \label{neqn}

43: \end{eqnarray}

44: \end{mathletters}

45: The second equation is equivalent to the formula

46: \begin{equation} \label{lap}

47: \nabla^2\Phi=0

48: \end{equation}

49: where they have defined\footnote{It should be noted that this $\Phi$ is not

50: the Newtonian gravitational potential.}

51: \begin{equation}

52: \Phi=\int\frac{N}{r}dr.

53: \end{equation}

54: 

55: Thus the observed rotation curve becomes a boundary condition for

56: the solution to Laplace's equation (\ref{lap}) which they take in the form

57: \begin{equation}

58: \Phi=\sum_nC_ne^{-k_n|z|}J_0(k_nr)

59: \end{equation}

60: where the $k_n$ are chosen for orthogonality over the radius of the

61: galaxy.  Once $N$ is found by fitting this function to the

62: obsersved rotation curve, they derive the density by

63: (\ref{dens}) and in this way they obtain an excellent fit to the data

64: while obtaining a density profile that accords with observation.

65: 

66: However, it has been pointed out \citep{c1,c2} that, since the solution

67: depends on $|z|$, equation (\ref{neqn}) is not satisfied, but

68: rather yields a singular contribution to the $z=0$ plane, which has the

69: properties of an exotic form of matter.  It may be wondered whether

70: this singular disk can be removed by choosing a different solution

71: form or by increasing the complexity of the model.  However, in the

72: following analysis we will show that this is not possible and that

73: model fails to accord with general relativity.

74: 

75: 

76: \section{Analysis}

77: Assuming this form of the metric, and without making any

78: approximations, the scalar of volume expansion $\Theta \equiv

79: u^\mu_{\phantom{\mu};\mu}$ vanishes (a semicolon denotes covariant

80: differentiation and a comma denotes partial differentiation).  Defining the space-projection tensor

81: \begin{equation}

82: h_{\mu\nu}=g_{\mu\nu}+u_\mu u_\nu

83: \end{equation}

84: the shear tensor is given by\footnote{$(\mu,\nu)$

85:   and $[\mu,\nu]$ denote symmetrization and antisymmetrizaion with respect to

86:   the enclosed indices, respectively.}

87: \begin{equation}

88: \sigma_{\mu\nu}=u_{(\alpha;\beta)}h^\alpha_{\phantom{\alpha}\mu}h^\beta_{\phantom{\alpha}\nu}

89: \end{equation}

90: and this vanishes as well\footnote{The vanishing of shear was also

91:   found by Bonner in his solution \citep{B}.}.  However, the vorticity tensor, given by

92: \begin{equation}

93: \omega_{\mu\nu}=u_{[\alpha;\beta]}h^\alpha_{\phantom{\alpha}\mu}h^\beta_{\phantom{\alpha}\nu}

94: \end{equation}

95: does not vanish, but has nonzero components

96: \begin{mathletters}

97: \begin{eqnarray}

98: \omega_{\varphi r} = -\omega_{r\varphi}&=& \case{1}{2}e^{w/2}N_{,r}\\

99: \omega_{\varphi z} = -\omega_{z\varphi}&=& \case{1}{2}e^{w/2}N_{,z}.

100: \end{eqnarray}

101: \end{mathletters}

102: Though the matter in this model does indeed rotate, the rotation is

103: rigid and thus cannot characterize a galaxy which is differentially

104: rotating.  It should be emphasized that since $\sigma_{\mu\nu}$ is a

105: tensor, this cannot be a coordinate effect.

106: 

107: With the present solution Raychaudhuri's equation \citep{CW} simplifies to

108: \begin{equation}

109: -\omega_{\mu\nu}\omega^{\mu\nu}=R_{tt}/g_{tt}

110: \end{equation}

111: and in fact reduces to the condition $\nabla^2w=0$.  This condition is

112: demanded in \cite{CT} on the grounds that the geodesic equation be

113: satisfied.  This is equivalent to saying that the geodesics must be

114: circular orbits about the $z$-axis, which should not hold in general.

115: Orbits in the $z=0$ plane should indeed be azimuthal, but we cannot expect this

116: behavior off that plane.  That Raychaudhuri's equation demands this

117: condition again reveals the rigidity of the rotation\footnote{Actually,

118: any co-moving coordinate system requires $g_{tt}=-1$ as the coordinate

119: points are in free fall and thus keep proper time \citep{W}, which here

120: requires $w\equiv 0$.}.

121: 

122: Now, if we seek solutions to (\ref{neqn}) that are symmetric about the

123: plane, and singularity free, then must require $N$ to be independent\footnote{We could choose $\cosh(z)$, but this would lead to

124:   an exponentially increasing matter density.} of $z$.  Thus (\ref{neqn}) has

125: the trivial solutions

126: \begin{equation}

127: \begin{array}{ccc}

128: N = A & \textrm{or} & N=Br^2

129: \end{array}

130: \end{equation}

131: where $A$ and $B$ are constants.  The first solution leads to zero

132: density and the second to a constant density under rigid rotation,

133: according to equations (\ref{rot}) and (\ref{dens}).  Thus it appears

134: that the physical origin of the singularity is in attempting to

135: describe, in co-moving coordinates, a non-rigidly rotating dust cloud,

136: which the metric (\ref{metric}) cannot.

137: 

138: Next, in order to solve the Einstein Equations Cooperstock and Tieu perform an

139: expansion of the metric in  $\sqrt{G}$ and conclude that the functions

140: $w$ and $v$ are of second order, but the function $N$, which couples to the

141: rotation, is of first order.  Strictly speaking, this expansion is not

142: well defined as the expansion parameter has dimensions.  We can form

143: the dimensionless parameter 

144: \begin{equation} \label{lambda}

145: \lambda=\sqrt{\frac{GM}{Lc^2}}

146: \end{equation}

147: where $M$ is some characteristic mass of the system and $L$ some

148: characteristic length (for example, the mass and radius of the

149: galactic core).  We then compare equations (\ref{neqn}) and (\ref{rot}):

150: \begin{eqnarray*}

151: \lambda^2r^{-2}\left(N_{,r}^2+N_{,z}^2\right) &=& \frac{8\pi

152:   G\rho}{c^2} \\

153: \lambda\frac{Nc}{r} &=& v 

154: \end{eqnarray*}

155: where the order has been shown explicitly.  Substituting $v$ for $N$

156: in the first equation yields the relation

157: \begin{equation}

158: v=\mathcal{O}\left(\sqrt{8\pi G \rho L^2}\right)

159: \end{equation}

160: where we have taken derivatives to be of order $1/L$.  This can be

161: rewritten as 

162: \begin{equation}

163: v=\mathcal{O}\left(c\lambda \sqrt{\frac{\rho L^3}{M}}\right)

164: \end{equation}

165: and since the quantity under the square root is of order unity, we

166: have

167: \begin{equation}

168: \frac{v}{c}=\mathcal{O}(\lambda)=\mathcal{O}\left(\sqrt{\frac{GM}{Lc^2}}\right)\label{order}

169: \end{equation}

170: which is expected from Newtonian theory and is the basis of the PPN

171: expansion.

172: 

173: Now, suppose we choose a coordinate system in which the galactic dust

174: has coordinate velocity $\omega/c$, so that the stress-energy tensor has

175: the form

176: \begin{equation}

177: T^{\mu\nu} \propto \left( \begin{array}{cc}

178: 1 & \displaystyle \frac{\omega}{c} \\

179: \\

180: \displaystyle \frac{\omega}{c} &  \displaystyle \left(\frac{\omega}{c}\right)^2

181: \end{array} \right)

182: \end{equation}

183: in the $t\varphi$-subspace.  The $t\varphi$-Einstein equation then has the

184: form

185: \begin{equation}

186: G^{t\varphi}=\frac{8\pi G}{c^2}T^{t\varphi}\propto \frac{8\pi

187:   G}{c^2}\frac{\omega}{c}

188: \end{equation}

189: up to a factor of order unity due to the constraint condition $u_\mu

190: u^\mu=-1$.  Thus, the right hand side of this equation begins at order

191: $\lambda^3/L^2$ according to (\ref{order}), whereas the left hand

192: side is of order $\lambda/L^2$ according to (\ref{neqn}), since $N$ is assumed to be of order $\lambda$.  Thus the assumption that $N$ is of

193: first order is inconsistent, while consistency requires that $N$ be of at

194: least third order\footnote{This will be demonstrated more explicitly in the next section.}.

195: 

196: Moreover, given the form of the stress-energy tensor above, suppose we

197: make a \emph{global} transformation $\varphi \to \varphi+\omega(r,z)t$

198: to the co-moving frame, so that all components of the new

199: stress-energy tensor vanish except the density. The new metric will

200: have the same form as the old metric, but for the differential $d\varphi'$ we have

201: \begin{equation}

202: d\varphi'=\omega dt + d\varphi +t(\omega_{,r}dr+\omega_{,z}dz)

203: \end{equation}

204: which necessarily introduces time-dependence into the new metric

205: unless $\omega$ is spatially-independent, that is, unless the rotation

206: of the matter is rigid.  Thus, contrary to the assumption of

207: Coopertock and Tieu, the metric (\ref{metric}) cannot both be

208: co-moving and time-independent.  This accords with zero value of the shear tensor above.

209: 

210: It can be seen in the following figure that a co-moving metric of a

211: differentially rotating system is time-dependent and possesses shear.  Here,

212: $r$-coordinate lines ``twist'' up in time relative to observers at

213: spatial infinity.  $\varphi=\omega t$ has been plotted, where

214: $\omega(r)$ is the fit to the Milky Way from \cite{CT}.

215: \begin{figure}[h]

216: \begin{center}

217: \epsscale{.5}

218: \plottwo{f1.eps}{f2.eps}

219: \caption{The twisting of $r$-coordinate lines for various $\varphi$ at

220: two different times as seen by observers at spatial infinity.}\label{fig1}

221: \end{center}

222: \end{figure}

223: 

224: \section{Non-co-moving Expansion}

225: In this section we carry out an expansion of the metric

226: (\ref{metric}) in a system of reference in which the galactic

227: dust has coordinate velocity 

228: \begin{equation}

229: \frac{u^2}{u^0}=\frac{d\varphi}{dct}=\frac{\omega}{c}

230: \end{equation}

231: which is physically the angular velocity measured by observers at

232: spatial infinity \citep{BW}.  In this direct approach we will show that

233: the angular momentum coupling is too weak to account

234: for the flat rotation curves.

235: 

236: We expand the metric (\ref{metric}) as

237: \begin{mathletters}

238: \begin{eqnarray}

239: g_{tt} &=& -1 - \stackrel{2}{w}-\stackrel{4}{w}+\mathcal{O}(\lambda^6) \\

240: g_{t\varphi} &=&

241: \stackrel{1}{N}+\stackrel{2}{w}\stackrel{1}{N}+\stackrel{3}{N} +\mathcal{O}(\lambda^5) \\

242: g_{\varphi\varphi} &=& r^2 -\stackrel{1}{N^2}-r^2\stackrel{2}{w}+\mathcal{O}(\lambda^4) \\

243: g_{rr} =g_{zz} &=& 1 +\stackrel{2}{w}-\stackrel{2}{v}+\mathcal{O}(\lambda^4)

244: \end{eqnarray}

245: \end{mathletters}

246: where the over-script indicates the order of the term in $\lambda$,

247: which is the same as defined in (\ref{lambda}). The presence of

248: only even terms in the $g_{\mu\mu}$ and odd in $g_{t\varphi}$ are for the

249: appropriate behavior under time-reversal \citep{W}.

250: 

251: The constraint condition $u_{\mu}u^{\mu}=-1$ requires that 

252: \begin{equation}

253: u^t=(g_{tt}+2\frac{\omega}{c}g_{t\varphi}+\frac{\omega^2}{c^2}g_{\varphi\varphi})^{-1/2}

254: \end{equation}

255: and thus the stress energy tensor has the expansion

256: \begin{mathletters}

257: \begin{eqnarray}

258: T^{tt} &=& \rho{c^2}\left( 1- \stackrel{2}{w} +\frac{r^2\omega^2}{c^2} \right) +\mathcal{O}(\lambda^4)\\

259: T^{t\varphi} &=& \rho{c^2}\frac{\omega}{c} +\mathcal{O}(\lambda^3)\\

260: T^{\varphi\varphi} &=& \rho{c^2}\frac{\omega^2}{c^2}+\mathcal{O}(\lambda^4)

261: \end{eqnarray}

262: \end{mathletters}

263: As seen in the previous section, $8\pi G/c^2$ increases the order by

264: two so that the right-hand side of the Einstein equations are of second order and higher, thus the $t\varphi$ equation to first order is

265: \begin{equation}

266: \stackrel{1}{N}_{,rr}-\frac{1}{r}\stackrel{1}{N}_{,r}+\stackrel{1}{N}_{,zz}=0

267: \end{equation}

268: so that the lowest order term of $N$ is sourceless.  We are free to

269: choose $\stackrel{1}{N}$ however we wish to make the solution simplest, thus we set

270: $\stackrel{1}{N}=0$.  With this selection the Einstein equations

271: through third order become

272: \begin{mathletters}

273: \begin{eqnarray}

274: -\nabla^2\stackrel{2}{w}+\case{1}{2}\left( \stackrel{2}{v}_{,zz}+

275: \stackrel{2}{v}_{,rr} \right) &=& \frac{8\pi G \rho}{c^2}\\

276: \stackrel{2}{v}_{,r}&=& 0\\

277: \stackrel{2}{v}_{,zz}+\stackrel{2}{v}_{,rr}&=& 0\\

278: \stackrel{2}{v}_{,z} &=& 0\\

279: -\case{1}{2}r^{-2}\left(

280: \stackrel{3}{N}_{,rr}-\frac{1}{r}\stackrel{3}{N}_{,r}+\stackrel{3}{N}_{,zz}

281: \right) &=& \frac{8\pi G \rho r\omega}{c^3}.

282: \end{eqnarray}

283: \end{mathletters}

284: We see that  $\stackrel{2}{v}$ must be a constant so that we have

285: \begin{mathletters}

286: \begin{eqnarray}

287: \nabla^2\stackrel{2}{w} &=& -\frac{8\pi G \rho}{c^2}\\

288: \stackrel{3}{N}_{,rr}-\frac{1}{r}\stackrel{3}{N}_{,r}+\stackrel{3}{N}_{,zz}

289: &=& -\frac{16\pi G \rho r^3\omega}{c^3}.

290: \end{eqnarray}

291: \end{mathletters}

292: Thus we see that the coupling to the angular momentum is of third

293: order, there is no longer a nonlinear term in the mass

294: density\footnote{Even when $G^{tt}$ is written to fourth order the

295:   nonlinearity due to $N$ is not present.}, and we can identify 

296: \begin{equation}

297: \stackrel{2}{w}=-\frac{2\Phi}{c^2}

298: \end{equation}

299: with the Newtonian gravitational potential.

300: 

301: If we analyze circular orbits on the plane the geodesic equation

302: demands that

303: \begin{equation}

304: \Gamma^\mu_{tt}+2\Gamma^\mu_{t\varphi}\frac{\omega}{c}+ \Gamma^\mu_{\varphi\varphi}\frac{\omega^2}{c^2}=0

305: \end{equation}

306: which to third order can be written for $\mu=r$ as

307: \begin{equation}

308: \frac{v^2}{r}=-\Phi_{,r}\left( 1+\frac{v^2}{c^2}\right)-\frac{vc}{r}\stackrel{3}{N}_{,r}

309: \end{equation}

310: which is recognized as the Newtonian centripetal equation plus second

311: order corrections\footnote{The presence of the $c$ in the last term

312:   effectively lowers the order by one.}.  Thus the matter essentially

313: moves according to the predictions of Newtonian gravitation with

314: corrections of order $v^2/c^2$, which cannot account for flattening of

315: the rotation curves without extra non-luminous matter.

316: 

317: Finally, we can compute the shear tensor, which has the non-zero

318: components

319: \begin{mathletters}

320: \begin{eqnarray}

321: \sigma_{tr}=\sigma_{rt} &=& -\frac{(u^t)^3r^2}{2c^2}\omega\omega_{,r}\\

322: \sigma_{tz}=\sigma_{zt} &=& -\frac{(u^t)^3r^2}{2c^2}\omega\omega_{,z}\\

323: \sigma_{\varphi r}=\sigma_{r\varphi} &=& \frac{(u^t)^3r^2}{2c}\omega_{,r}\\

324: \sigma_{\varphi z}=\sigma_{z\varphi} &=& \frac{(u^t)^3r^2}{2c}\omega_{,r}

325: \end{eqnarray}

326: \end{mathletters}

327: all of which vanish exactly when $\omega$ is constant.

328: 

329: \section{Conclusion}

330: It has been shown that the Cooperstock-Tieu galaxy model is

331: inconsistent as a general relativistic model and that a proper model

332: fails to account for the flatness of the rotation curves without the

333: dark-matter hypothesis.  This failure is due to the weakness of the

334: metric coupling to the angular momentum of the galaxy.  

335: 

336: However, the flat rotation curves seem to imply a large inertial induction

337: effect, where the rotating inner matter boosts the rotation of the outer

338: matter, leveling off the rotation curve, which is what the

339: Cooperstock-Tieu model attempts to describe within general

340: relativity.  Since their solution predicts a matter density well

341: within visible limits it is quite possible that their solution

342: represents an alternative, more Machian, gravitational theory where

343: inertial induction effects are much larger than in General Relativity.

344: 

345: \acknowledgements

346: 

347: The author is greatful to Prof. R. Gilmore for his encouragement and

348: many helpful discussions on this project.

349: 

350: \begin{thebibliography}{99}

351: \bibitem[Bardeen \& Wagoner(1971)]{BW} Bardeen, J. M. \& Wagoner, R. V. 1971, \apj, 167, 359

352: \bibitem[Bonner(1977)]{B} Bonner, W.B. 1977, J. Phys. A: Math. Gen. 10, 1673.

353: \bibitem[Ciufolini \& Wheeler(1955)]{CW} Ciufolini, I. \& Wheeler,

354:   J.A. 1955, Gravitation and Inertia, (Princeton, NJ: Princeton University Press)

355: \bibitem[Cooperstock \& Tieu(2005)]{CT} Cooperstock, F. I. \& Tieu,

356:   S. 2005, submitted (astro-ph/0507619)

357: \bibitem[Korzynski(2005)]{c1} Korzynski, M. 2005, preprint (astro-ph/0508377)

358: \bibitem[Vogt \& Letelier(2005)]{c2} Vogt, B. \& Letelier,

359:   P. S. 2005, preprint (astro-ph/0510750)

360: \bibitem[Weinberg(1972)]{W} Weinberg, S. 1972, Gravitation and

361:   Cosmology, (New York: Wiley and Sons)

362: 

363: \end{thebibliography}

364: 

365: \end{document}

366: