1:
2: %\documentclass{article}
3: \documentclass[eqsecnum,aps,nofootinbib]{revtex4}
4:
5: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
6:
7: %TCIDATA{OutputFilter=LATEX.DLL}
8: %TCIDATA{Version=4.00.0.2312}
9: %TCIDATA{LastRevised=Thursday, September 01, 2005 11:39:29}
10: %TCIDATA{<META NAME="GraphicsSave" CONTENT="32">}
11: %TCIDATA{Language=American English}
12:
13: %\input{tcilatex}
14:
15: %\usepackage{amssymb}
16: \usepackage{graphicx}
17: \usepackage{bm}
18: \usepackage{theorem}
19: \usepackage{here}
20:
21: \begin{document}
22:
23: \title{Cosmological Co-evolution of Yang-Mills Fields and Perfect Fluids}
24:
25: \date{\today }
26:
27: \author{John D. Barrow$^{1}$}
28: \author{Yoshida Jin$^{2}$}
29: \author{Kei-ichi Maeda$^{2,3,4}$}
30: %EndAName
31: \affiliation{$^{1}$DAMTP, Centre for Mathematical Sciences,
32: Cambridge University, Cambridge CB3 0WA, UK}
33: \affiliation{$^{2}$Department of Physics, Waseda University,
34: 3-4-1 Okubo, Shinjuku-ku,Tokyo 169-8555, Japan}
35: \affiliation{$^{3}$ Waseda Institute for Astrophysics, Waseda University,
36: Shinjuku, Tokyo 169-8555, Japan}
37: \affiliation{$^{4}$ Advanced Research Institute for Science and Engineering,
38: Waseda University, Shinjuku, Tokyo 169-8555, Japan}
39:
40:
41: \begin{abstract}
42: We study the co-evolution of Yang-Mills fields and perfect fluids in Bianchi
43: type I universes. We investigate numerically the evolution of the universe
44: and the Yang-Mills fields during the radiation and dust eras of a universe
45: that is almost isotropic.
46: The Yang-Mills field undergoes small
47: amplitude chaotic oscillations,
48: which are also displayed by the expansion scale factors of the universe.
49: The results of the numerical simulations are
50: interpreted analytically and compared with past studies of the cosmological
51: evolution of magnetic fields in radiation and dust universes. We find that,
52: whereas magnetic universes are strongly constrained by the microwave
53: background anisotropy, Yang-Mills universes are principally constrained by
54: primordial nucleosynthesis and the bound is comparatively weak, and
55: $\Omega _{\mathrm{YM}}<0.105\Omega _{\mathrm{rad}} \,.$
56: \end{abstract}
57:
58: \maketitle
59:
60:
61:
62: \section{Introduction}
63:
64: There is considerable interest in the generation and evolution of magnetic
65: fields in cosmological models \cite{sub}. This interest has a double focus.
66: There is the hope that an explanation might be found for the existence of
67: significant magnetic fields in galaxies. This may require seed fields to
68: originate very early in the history of the universe, or even form part of
69: the initial conditions. We have also discovered that the cosmological
70: evolution of anisotropic stresses, like electric and magnetic fields, has
71: interesting mathematical features that create unexpected physical
72: consequences. These features appear when a homogeneous and anisotropic
73: stress with a trace-free energy-momentum tensor is present during the
74: radiation era of the universe. The isotropic black-body radiation and the
75: anisotropic stresses evolve to first order in the same way, but there is a
76: greatly slowed (logarithmic) decay of the shear anisotropy caused by the
77: pressure anisotropy of the anisotropic fluid \cite{zeld},\cite{collins}. If
78: the evolution were to be linearised around the isotropic model then a zero
79: eigenvalue would exist. When the equation of state of the accompanying
80: perfect fluid changes from radiation to dust the evolution of the shear is
81: still dominated by the pressure anisotropy of the anisotropic stresses but
82: the zero eigenvalue disappears and the shear falls as a reduced power of the
83: cosmic time. In the study of these effects made in ref \cite{Barrow} we
84: considered the effects of anisotropic stresses whose trace-free anisotropic
85: pressure tensor $\pi _{ab}$ was proportional to a density $\mu _{A},$ so
86:
87: \begin{equation}
88: \pi _{ab}=C_{ab}\mu _{A}, \label{an}
89: \end{equation}%
90: with $C_{ab}$ constant and $C_{a}^{a}=0.$ Barrow and Maartens considered the
91: generalisation of these results to general inhomogeneous cosmological models
92: close to isotropy using the covariant formalism \cite{BM} and also some
93: applications to Kaluza-Klein cosmologies \cite{BMKK}. The key feature in the
94: evolution of almost-isotropic cosmological models containing perfect fluids
95: plus a magnetic field is that during the radiation era the ratio of the
96: shear to Hubble expansion rate falls only logarithmically in time. The
97: universe evolves towards an attractor state in which it is proportional to
98: the ratio of the magnetic and perfect-fluid densities. This means that the
99: ratio of these densities at the epoch of last-scattering determines the
100: large-angle temperature anisotropy of the cosmic microwave background (CMB).
101: This enables us to use observations of the CMB to place strong bounds on any
102: homogeneous cosmological magnetic field \cite{maglim}, or other anisotropic
103: stress defined by a constant $C_{ab},$ \cite{Barrow}. An important extension
104: of these studies is to consider anisotropic stresses with more general
105: pressure tensors, for example those with time-dependent $C_{ab}.$
106:
107: An important case with time-dependent $C_{ab}$, which generalises the
108: electromagnetic field, is that of a Yang-Mills stress. This has been studied
109: for pure Yang-Mills fields and no accompanying fluid by several authors in
110: simple Bianchi type I anisotropic universes \cite{YMtypeI, kun, blev}, in
111: all class A Bianchi type universes \cite{jin}, and in Kantowski-Sachs
112: universes \cite{YMKS}. These studies reveal that the Yang-Mills field
113: creates a form of chaotic behaviour during the evolution of the Yang-Mills
114: stress \cite{blev}. This is not surprising because such behaviour is present
115: in Minkowski space-time and is nothing to do with the cosmological evolution
116: or the spacetime curvature. It should not be confused with the chaotic
117: behaviour of general relativistic origin that is found in the 'Mixmaster'
118: universes of Bianchi types VIII and IX, even in vacuum\footnote{%
119: The most general cosmological analysis of the Yang-Mills chaos is that
120: carried out for all Ellis-MacCallum Class A Bianchi-type vacuum cosmologies by Jin
121: and Maeda \cite{jin} and they are able to consider the simultaneous presence
122: of the Yang-Mills chaos and Mixmaster chaos \cite{mix, bkl, jdbchaos1,
123: jdbchaos2, jdbchaos3}.}.
124: Therefore the Yang-Mills chaos occurs in the
125: simplest axisymmetric universes of Bianchi type I and is present even when
126: the anisotropy level is arbitrarily small. This raises the interesting question of
127: whether the chaotic evolution might leave an imprint in the temperature
128: anisotropy of the CMB when it decouples at a redshift $z_{rec}\approx 1000.$
129:
130: However, all of these earlier studies consider the evolution of
131: Einstein-Yang-Mills equations only for the case of a pure Yang-Mills stress
132: in an anisotropic cosmological model. Our experience \cite{Barrow} with the
133: behaviour of magnetic fields in anisotropic universes teaches us that it is
134: important to include the presence of a perfect fluid in order to find the
135: realistic evolution of the Yang-Mills stress during the radiation and
136: dust-dominated eras of the cosmological expansion. In this paper we consider
137: this generalisation by analysing the evolution of spatially homogeneous,
138: anisotropic universes of Bianchi type I containing both a perfect fluid and
139: an anisotropic Yang-Mills stress. By combining analytic and numerical
140: studies we determine the evolution of the shear and the Yang-Mills stress
141: during the early universe in order to discover if it is possible for it to
142: leave an observable signature in the CMB.
143:
144: In section 2 we set out the Einstein-Yang-Mills equations for the Bianchi
145: class A universes. In section 3 we specialise to consider in detail the
146: evolution for the Bianchi type I universe containing Yang-Mills fields and
147: perfect fluids, focussing on the physically significant cases of
148: pressureless dust and black-body radiation. In section 4 we describe the
149: numerical and analytical results and in section 5 compare them in detail
150: with the situation that results when a pure magnetic field replaces the
151: Yang-Mills field. We consider the observational bounds that can be placed on
152: the magnetic and Yang-Mills fields in the presence of perfect fluids using
153: the microwave background temperature anisotropy and the primordial
154: nucleosynthesis of helium-4 in section 6. We summarise our results in
155: section 7.
156:
157:
158: \section{The Einstein-Yang-Mills evolution equations}
159:
160: The Einstein equations to be solved for universes containing a perfect fluid
161: and a Yang-Mills (YM) field are
162: \begin{eqnarray}
163: G_{\mu \nu } &=&8\pi GT_{\mu \nu }, \\
164: T_{\mu \nu } &=&T_{\mathrm{YM}\mu \nu }+T_{\mathrm{m}\mu \nu }, \\
165: T_{\mathrm{m}\mu \nu } &=&\mbox{diag}
166: (-\mu _{\mathrm{m}},p_{\mathrm{m}},p_{\mathrm{m}},p_{\mathrm{m}}), \\
167: T_{\mathrm{YM}\mu \nu } &=&\frac{1}{g_{\mathrm{YM}}^{2}}\left[ F_{\mu
168: \lambda }^{(A)}F_{\ \nu }^{(A)\ \lambda }-\frac{1}{4}g_{\mu \nu }F_{\lambda
169: \sigma }^{(A)}F^{(A)\lambda \sigma }\right] ,
170: \end{eqnarray}%
171: where $F_{\mu \nu }^{(A)}$ is the field strength of YM field,
172: \textquotedblleft $A$\textquotedblright\ describes the components of the
173: internal SU(2) space, and $\mu _{\mathrm{m}}$ and $p_{\mathrm{m}}$ have
174: an equation of state
175:
176: \begin{center}
177: $p_{\mathrm{m}}=(\gamma -1)\mu _{\mathrm{m}}.$
178: \end{center}
179:
180: We shall assume that $T_{\mathrm{YM}\mu \nu }$ and $T_{\mathrm{m}\mu \nu }$
181: are separately conserved and there is no energy exchange between them.
182: Hence, from the vanishing covariant divergence of $T_{\mathrm{YM}\mu \nu }$
183: we find that the YM equations are
184: \begin{equation}
185: \vec{F}_{\ \ \ ;\nu }^{\mu \nu }-\vec{A}_{\nu }\times \vec{F}^{\mu \nu }=0\,,
186: \end{equation}%
187: where $g_{\mathrm{YM}}$ is the self-coupling constant of YM field, and $\vec{
188: F}^{\mu \nu }=F^{(A)\mu \nu }\vec{\tau}_{A}$ and $\vec{A}_{\nu }=A_{\nu
189: }^{(A)}\vec{\tau}_{A}$ with $\vec{\tau}_{A}$ being the $SU(2)$ basis.
190: Setting our units as $8\pi G/g_{\mathrm{YM}}^{2}=1$ \footnote{%
191: This choice of units has the following consequences. If $g_{\mathrm{YM}}=1$,
192: then $8\pi G=1$; if $g_{\mathrm{YM}}=10^{-10}$, then $8\pi G=10^{-20}$ and
193: in this case the unit of time is $10^{10}t_{\mathrm{pl}}\sim 10^{-34}\mathrm{
194: sec}$ and the present Hubble time is $H_{0}^{-1}=1.2\times 10^{10}\mathrm{yr}
195: =4\times 10^{17}\mathrm{sec}=10^{51}$.}
196: and $c=\hbar =1$, the basic
197: equations become free of the value of $g_{\mathrm{YM}}$.
198:
199: We adopt the orthonormal-frame formalism, which has been developed in ref.
200: \cite{wainwright-ellis}. In Bianchi-type spacetimes, there exists a
201: 3-dimensional homogeneous spacelike hypersurface $\Sigma _{t}$ which is
202: parameterized by a time coordinate $t$. The timelike basis is given by
203: $\bm{e}_{0}=\bm{\partial}_{t}$. The triad basis $\{\bm{e}_{a}\}$ on the
204: hypersurface $\Sigma _{t}$ is defined by the commutation function $\gamma
205: _{\ ab}^{c}$ as
206: \begin{equation}
207: \lbrack \bm{e}_{a},\ \bm{e}_{b}]=\gamma _{\ ab}^{c}\bm{e}_{c}\,.
208: \end{equation}%
209: It is convenient to decompose $\gamma _{\ ab}^{c}$ into the geometric
210: variables denoted conventionally by the Hubble expansion $H$, the
211: acceleration $\dot{u}_{a}$, the shear $\sigma _{ab}$, and the vorticity $%
212: \omega _{ab}$, and the variables $\Omega _{a}$ (the rotation of $\bm{e}_{a}$%
213: ), and the variables $a_{a}$ and $n_{ab}$, which distinguish the type of
214: Bianchi model, so
215: \begin{eqnarray}
216: \gamma _{\ 0b}^{a} &=&-\sigma _{b}^{\ a}-H\delta _{b}^{\ a}-\epsilon _{\
217: bc}^{a}(\omega ^{c}+\Omega ^{c}), \label{gamma1} \\
218: \gamma _{\ 0a}^{0} &=&\dot{u}_{a}, \label{gamma2} \\
219: \gamma _{\ ab}^{0} &=&-2\epsilon _{ab}^{\quad c}\omega _{c}, \label{gamma3}
220: \\
221: \gamma _{\ ab}^{c} &=&\epsilon _{abd}n^{dc}+a_{a}\delta _{b}^{\
222: c}-a_{b}\delta _{a}^{\ c}. \label{gamma4}
223: \end{eqnarray}%
224: \emph{\ }If Bianchi spacetimes are expressed in this way, the vorticity $\omega
225: _{ab}$ always vanishes because they are hypersurface orthogonal.
226:
227: In order to analyze the EYM system, we shall study the simplest case. First,
228: using the gauge freedom of the YM field, we set $A_{0}^{(A)}(t)=0$, which
229: simplifies the vector potential to $\vec{\bm{A}}=A_{a}^{(A)}(t)\vec{\tau}_{A}
230: \bm{\omega}^{a}$, where $\bm{\omega}^{\alpha }$ is the dual basis of $\bm{e}
231: _{\alpha }$ \cite{okuyama}. Next we shall restrict attention to cosmologies
232: of Bianchi type class A, in which the vector $a_{a}$ vanishes. We can also
233: diagonalize $n_{ab}$, i.e. $n_{ab}=\mbox{diag}\{n_{1},n_{2},n_{3}\}$, using
234: the remaining freedoms of a time-dependent rotation of the triad basis.
235: Then, we can show that if $\sigma _{ab},\ A_{a}^{(A)}$ and $\dot{A}
236: _{a}^{(A)} $ do not have off-diagonal components initially, the equations of
237: motion guarantee that those variables will remain diagonal during the
238: subsequent evolution for the class A Bianchi spacetimes (see the Appendix in
239: \cite{jin}). As a result, $\Omega _{a}$ vanishes and the number of basic
240: variables defining the initial value problem reduces from 21 (12 for
241: spacetime [$H,N_{a},\sigma _{ab},\Omega _{a}$] and 9 for YM field
242: [$A_{a}^{(A)}$]) to 10 (7 for spacetime [$H,N_{a},\sigma _{aa}$] and 3
243: for YM field [$Y_{a}^{(a)}$]).
244:
245: In order to discuss the dynamics most efficiently, it is convenient to
246: introduce the Hubble-normalized variables, which are defined as follows:
247: \begin{equation}
248: \Sigma _{ab}\equiv \frac{\sigma _{ab}}{H},~~~\ N_{a}\equiv \frac{n_{a}}{H}.
249: \label{def}
250: \end{equation}
251:
252: We also re-express the shear variables of the Bianchi spacetime as
253: \[
254: \Sigma _{+}\equiv \frac{1}{2}\left( \Sigma _{22}+\Sigma _{33}\right)
255: ,~~\Sigma _{-}\equiv \frac{1}{2\sqrt{3}}\left( \Sigma _{22}-\Sigma
256: _{33}\right) ,~~\Sigma ^{2}\equiv \Sigma _{+}^{2}+\Sigma _{-}^{2}\,.
257: \]%
258: The diagonal components of the YM field potential are described by new
259: variables $a(t),\ b(t),$ and $c(t)$ as
260: \[
261: a\equiv A_{1}^{(1)},\ b\equiv A_{2}^{(2)},\ c\equiv A_{3}^{(3)}\,.
262: \]%
263: and a new time variable, $\tau $, defined by $d\tau =Hdt$ is introduced so
264: that $\tau $ denotes the e-folding number of the scale length.
265:
266: Using these variables we can write down the evolution and constraint
267: equations explicitly. They consist of the generalized Friedmann equation,
268: the dynamical Einstein equations and the YM evolution equations. The
269: generalized Friedmann equation, which is the constraint equation, is
270: \begin{equation}
271: \Sigma ^{2}+\Omega _{\mathrm{YM}}+\Omega _{\mathrm{m}}+K=1\,,
272: \label{eq:GFRW}
273: \end{equation}%
274: where $\Omega _{\mathrm{YM}}$ is the density parameter of the YM field, i.e.
275: the Hubble-normalized energy density of the YM field, which is defined by
276: \begin{eqnarray}
277: \Omega _{\mathrm{YM}} &\equiv &\frac{\mu _{\mathrm{YM}}}{3H^{2}} \nonumber
278: \\
279: &=&\frac{1}{6}\left[ \left\{ a^{\prime }+(-2\Sigma _{+}+1)a\right\}
280: ^{2}+\left\{ b^{\prime }+(\Sigma _{+}+\sqrt{3}\Sigma _{-}+1)b\right\}
281: ^{2}+\left\{ c^{\prime }+(\Sigma _{+}-\sqrt{3}\Sigma _{-}+1)c\right\}
282: ^{2}\right. \nonumber \\
283: &&\left. +\left( N_{1}a+{\frac{bc}{H}}\right) ^{2}+\left( N_{2}b+{\frac{ca}{H%
284: }}\right) ^{2}+\left( N_{3}c+{\frac{ab}{H}}\right) ^{2}\right] ;
285: \end{eqnarray}%
286: $\Omega _{\mathrm{m}}$ is the density parameter of perfect fluid, defined by
287: $\Omega _{\mathrm{m}}\equiv \mu _{\mathrm{m}}/(3H^{2})$, and $K$ is the
288: Hubble normalized curvature, which is defined by
289: \[
290: K\equiv -\frac{{}^{(3)}R}{6H^{2}}=\frac{1}{12}\left\{
291: N_{1}^{2}+N_{2}^{2}+N_{3}^{2}-2(N_{1}N_{2}+N_{2}N_{3}+N_{3}N_{1})\right\}
292: \,.
293: \]%
294: Note that the positive 3-curvature corresponds to $K<0$. Hence, except for
295: the Bianchi type IX models we have $K\geq 0$. The energy density of YM field
296: and perfect fluid are always positive definite. Thus we find that $\Sigma
297: ^{2},\ \Omega _{\mathrm{YM}},$ and $K$ are restricted to the domains $0\leq
298: \Sigma ^{2}\leq 1,\ 0\leq \Omega _{\mathrm{YM}}\leq 1,$ and $\ 0\leq K\leq 1$%
299: , except for the Type IX universes.
300:
301: The dynamical Einstein equations are
302: \begin{eqnarray}
303: H^{\prime } &=&-(1+q)H, \label{eq:H} \\
304: \Sigma _{+}^{\prime } &=&(q-2)\Sigma _{+}-S_{+}+\Pi _{+}, \label{eq:S+} \\
305: \Sigma _{-}^{\prime } &=&(q-2)\Sigma _{-}-S_{-}+\Pi _{-}, \label{eq:S-} \\
306: N_{1}^{\prime } &=&(q-4\Sigma _{+})N_{1}, \label{eq:N1} \\
307: N_{2}^{\prime } &=&(q+2\Sigma _{+}+2\sqrt{3}\Sigma _{-})N_{2}, \label{eq:N2}
308: \\
309: N_{3}^{\prime } &=&(q+2\Sigma _{+}-2\sqrt{3}\Sigma _{-})N_{3}, \label{eq:N3}
310: \end{eqnarray}%
311: where $q$ is deceleration parameter:
312: \begin{equation}
313: q\equiv 2\Sigma ^{2}+\frac{1}{2}\Omega +\frac{p}{2H^{2}}=2\Sigma ^{2}+\Omega
314: _{\mathrm{YM}}+\frac{1}{2}(3\gamma -2)\Omega _{\mathrm{m}}\,.
315: \end{equation}%
316: The quantities $S_{\pm }$ are given only by the $N_{a}:$
317: \begin{eqnarray}
318: S_{+} &=&\frac{1}{6}\left\{
319: (N_{2}-N_{3})^{2}-N_{1}(2N_{1}-N_{2}-N_{3})\right\} , \\
320: S_{-} &=&\frac{1}{2\sqrt{3}}(N_{3}-N_{2})(N_{1}-N_{2}-N_{3})\,.
321: \end{eqnarray}%
322: Last, we define the Hubble-normalized anisotropic pressures of YM field, $%
323: \Pi _{+}$ and $\Pi _{-},$ so that
324: \begin{eqnarray}
325: \Pi _{+} &=&-\frac{1}{6}\left[ -2\left\{ a^{\prime }+(-2\Sigma
326: _{+}+1)a\right\} ^{2}+\left\{ b^{\prime }+(\Sigma _{+}+\sqrt{3}\Sigma
327: _{-}+1)b\right\} ^{2}+\left\{ c^{\prime }+(\Sigma _{+}-\sqrt{3}\Sigma
328: _{-}+1)c\right\} ^{2}\right. \nonumber \\
329: &&\left. -2\left( N_{1}a+{\frac{bc}{H}}\right) ^{2}+\left( N_{2}b+{\frac{ca}{%
330: H}}\right) ^{2}+\left( N_{3}c+{\frac{ab}{H}}\right) ^{2}\right] ,
331: \label{eq:Pi+} \\
332: \Pi _{-} &=&\frac{\sqrt{3}}{6}\left[ -\left\{ b^{\prime }+(\Sigma _{+}+\sqrt{%
333: 3}\Sigma _{-}+1)b\right\} ^{2}+\left\{ c^{\prime }+(\Sigma _{+}-\sqrt{3}%
334: \Sigma _{-}+1)c\right\} ^{2}\right. \nonumber \\
335: &&\left. -\left( N_{2}b+{\frac{ca}{H}}\right) ^{2}+\left( N_{3}c+{\frac{ab}{H%
336: }}\right) ^{2}\right] \,. \label{eq:Pi-}
337: \end{eqnarray}
338:
339: Here we can see the complexity introduced by the YM dynamics.
340: We can see how the anisotropic pressures are driven by the directional scale
341: factors and their time derivatives, and not solely by the fractions of the
342: density as was the case with the simple model of anisotropic stresses
343: defined by eq. (\ref{an}).
344:
345: The isotropic matter conservation equation is
346: \begin{equation}
347: \Omega _{\mathrm{m}}^{\prime }=\{2q-(3\gamma -2)\}\Omega _{\mathrm{m}}
348: \label{con}
349: \end{equation}%
350: and the YM evolution equations are
351: \begin{eqnarray}
352: a^{\prime \prime } &=&(q-2)a^{\prime } \nonumber \\
353: &&+(q-1-4K+4\Sigma _{+}^{2}-4\Sigma _{+}-N_{1}(N_{2}+N_{3})+2\Pi _{+})a
354: \nonumber \\
355: &&-\frac{1}{H}(N_{1}+N_{2}+N_{3})bc-\frac{1}{H^{2}}a(b^{2}+c^{2}),
356: \label{eq:a} \\
357: b^{\prime \prime } &=&(q-2)b^{\prime } \nonumber \\
358: &&+(q-1-4K+(\Sigma _{+}+\sqrt{3}\Sigma _{-})(\Sigma _{+}+\sqrt{3}\Sigma
359: _{-}+2)-N_{2}(N_{3}+N_{1})-\Pi _{+}-\sqrt{3}\Pi _{-})b \nonumber \\
360: &&-\frac{1}{H}(N_{1}+N_{2}+N_{3})ca-\frac{1}{H^{2}}b(c^{2}+a^{2}),
361: \label{eq:b} \\
362: c^{\prime \prime } &=&(q-2)c^{\prime } \nonumber \\
363: &&+(q-1-4K+(\Sigma _{+}-\sqrt{3}\Sigma _{-})(\Sigma _{+}-\sqrt{3}\Sigma
364: _{-}+2)-N_{3}(N_{1}+N_{2})-\Pi _{+}+\sqrt{3}\Pi _{-})c \nonumber \\
365: &&-\frac{1}{H}(N_{1}+N_{2}+N_{3})ab-\frac{1}{H^{2}}c(a^{2}+b^{2})\,.
366: \label{eq:c}
367: \end{eqnarray}
368:
369: The basic variables are therefore [$H,\ \Sigma _{+},\ \Sigma _{-},\ N_{1},\
370: N_{2},\ N_{3},\ a,\ b,\ c$], and they are determined uniquely and
371: completely by Eqs. (\ref{eq:H})-(\ref{eq:N3}), (\ref{eq:a})-(\ref{eq:c}),
372: along with the constraint equation (\ref{eq:GFRW}).
373:
374:
375: \section{Bianchi I co-evolution of Yang-Mills fields and perfect fluids}
376:
377: We compute the solutions of the dynamical equations (\ref{eq:H})-(\ref{eq:N3})
378: and (\ref{eq:a})-(\ref{eq:c}) numerically and illustrate the typical
379: evolutionary behaviours. In this section, we consider for simplicity only
380: the Bianchi type I spacetime (so $K=0$) with black-body radiation ($\gamma
381: =4/3$) or dust ($\gamma =1$) and a YM field. We set initial value of $H$ as $%
382: H_{\mathrm{ini}}=10^{-10}$ (our unit), which is not a restriction because
383: our time unit depends on $g_{\mathrm{YM}}$ which determines the physical
384: identity of the YM field and can be scaled.
385:
386: First, we display the behavior of the three leading quantities,
387: $\Sigma^{2},\ \Omega _{\mathrm{YM}},$ and $\Omega _{\mathrm{m}}$.
388: %%%%%%%%%%%%%%%%%%%%
389: \begin{figure}[ht]
390: \includegraphics[height=6cm]{GFRW.eps}
391: \caption{The typical behaviors of $\Sigma^2,\ \Omega_{\mathrm{YM}},$
392: and $\Omega_{\mathrm{m}}$ with increasing time.
393: This figure shows how $\Sigma^2\rightarrow 0$, and the spacetime approaches
394: the isotropic Friedmann model.
395: It is remarkable that $\Omega_{\mathrm{YM}}$ does not damp.}
396: \label{fig:GFRW}
397: \end{figure}
398: %%%%%%%%%%%%%%%%%%%%
399: Figure \ref{fig:GFRW} shows $\Sigma ^{2}\rightarrow 0$, i.e., spacetime
400: approaches to the flat isotropic FRW universe. We also find that $\Omega
401: _{\mathrm{YM}}$ does not damp, but approaches a constant.
402: It is interesting to compare this behavior with that which obtains
403: close to isotropy in universes containing radiation and a magnetic field.
404: There we found that $\Sigma $ and $\Omega _{\mathrm{mag}}$ both fell slowly
405: as $(\ln t)^{-1}$ as $t$ increased, with asymptotic approach to an attractor
406: where $\Omega_{\mathrm{mag}}/\Omega _{\mathrm{rad}}=\lambda \Sigma $,
407: where $\lambda \approx O(1)$ is a calculable constant depending
408: on the orientation of the field and the number of effectively massless spin states
409: in the radiation background.
410:
411: Second, we can display the evolution of the components of the YM field.
412: %%%%%%%%%%%%%%%%%%%%
413: \begin{figure}[ht]
414: \begin{tabular}{ccc}
415: \includegraphics[height=6cm]{YM.eps} &
416: \includegraphics[height=6cm]{YM_abs.eps} \\
417: (a)&(b)\\
418: \end{tabular}
419: \caption{Typical behaviors of the Yang-Mills field. \\
420: (a) $a,\ b,$ and $c$ damp whilst undergoing chaotic oscillations of increasing frequency. \\
421: (b) The averaged damping rate of $a,\ b,$ and $c$ has the approximate form $ e^{-\tau} $. }
422: \label{fig:YM}
423: \end{figure}
424: %%%%%%%%%%%%%%%%%%%
425: Figure \ref{fig:YM} shows that $a,\ b,$ and $c$ follow a sequence of damped
426: chaotic oscillations, and decay as $e^{-\tau }$. Figure \ref{fig:YM} also
427: shows that the frequencies of the oscillations of $\ a,\ b,$ and $c$ are
428: growing, although their amplitudes are decaying. This is further supported
429: by Figure \ref{fig:YM'}, which shows that the mean amplitudes of $a^{\prime },\
430: b^{\prime },$ and $c^{\prime }$ remain nearly constant.
431: %%%%%%%%%%%%%%%%%%%%
432: \begin{figure}[ht]
433: \includegraphics[height=6cm]{YM2.eps}
434: \caption{Typical behaviors of the time derivative of the principal components of
435: the Yang-Mills field. The mean amplitudes of this shows amplitudes of
436: $a',\ b',$ and $c'$ remain nearly constant.}
437: \label{fig:YM'}
438: \end{figure}
439: %%%%%%%%%%%%%%%%%%%%
440:
441: Third, we show the behavior of $\Pi _{\pm }$, the Hubble-normalized
442: anisotropic pressures.
443: %%%%%%%%%%%%%%%%%%%%
444: \begin{figure}[ht]
445: \includegraphics[height=6cm]{Pi.eps}
446: \caption{The typical behaviors of the anisotropic pressure components $\Pi_{\pm}$.
447: This shows that $\Pi_{\pm}$ oscillate in a complicated way with nearly constant mean amplitudes.}
448: \label{fig:Pi}
449: \end{figure}
450: %%%%%%%%%%%%%%%%%%%%
451: Figure \ref{fig:Pi} shows that the $\Pi _{\pm }$ oscillate in a complicated
452: fashion with approximately constant mean amplitudes. This is consistent with the
453: observation that $\Omega _{\mathrm{YM}}\propto \Omega _{\mathrm{m}}$ but it is surprising that
454: the pressure anisotropy remains constant despite the fall in the shear.
455: These chaotic oscillations are a familiar feature of the YM evolution which
456: resembles the chaotic Hamiltonian dynamics of a point bouncing inside a
457: potential whose four steep walls in the $x-y$ plane are formed by the
458: branches of the rectangular hyperbolae $x^{2}y^{2}=$ constant
459: \cite{blev,jin}.
460:
461: In the next section we try to understand these features
462: emerging from the numerical studies of Bianchi I spacetime with the YM field
463: and radiation or dust. \emph{\ }
464:
465: \subsection{The approach to isotropy}
466:
467: In this section we consider Bianchi I spacetime again, i.e., $N_{a}=0$ and $K=0$.
468: We assume that the FRW limit holds, that is $\Sigma \rightarrow 0,$ so
469: the dynamics can be close to isotropy. This results in
470: \begin{eqnarray}
471: q = \Omega _{\mathrm{YM}}+\frac{1}{2}(3\gamma -2)\Omega _{\mathrm{m}}\,, \label{eq:q} \\
472: \Omega _{\mathrm{YM}}+\Omega _{\mathrm{m}} = 1\,. \label{eq:FRW}
473: \end{eqnarray}%
474: These conditions imply
475: \begin{equation}
476: q=1+\frac{1}{2}(3\gamma -4)\Omega _{\mathrm{m}}\,. \label{eq:q2}
477: \end{equation}%
478: Motivated by the results of the numerical analysis, we also assume that
479: $\Omega _{\mathrm{YM}} \rightarrow \mbox{const.}$ and
480: $\Omega _{\mathrm{m}} \rightarrow \mbox{const.}$
481: These assumptions reduce (\ref{con}) to the condition that
482: \begin{equation}
483: \Omega _{\mathrm{m}}^{\prime }=\{2q-(3\gamma -2)\}\Omega _{\mathrm{m}}=0\,,
484: \end{equation}%
485: and if $\Omega _{\mathrm{m}} \neq 0$, we have
486: \begin{equation}
487: q=\frac{1}{2}(3\gamma -2)\,. \label{eq:q_sol}
488: \end{equation}%
489: Hence the evolution of $H$ is given by $H^{\prime }=-(1+q)H=-\frac{3}{2}%
490: \gamma H$. Solving this equation, we have
491: \begin{equation}
492: H \propto \exp \left(-\frac{3}{2}\gamma \tau \right)\,. \label{eq:expansion}
493: \end{equation}%
494: This means that\ to leading order the YM field does not determine the
495: expansion rate in FRW limit.
496:
497: From eqs. (\ref{eq:q2}) and (\ref{eq:q_sol}), we get
498: \begin{equation}
499: 3\gamma -4=(3\gamma -4)\Omega _{\mathrm{m}}\,.
500: \end{equation}%
501: Therefore, in the radiation case $(\gamma =4/3)$, $\Omega _{\mathrm{m}}$ can converge
502: to any value. Otherwise $(\gamma \neq 4/3)$, $\Omega _{\mathrm{m}}$ becomes
503: dominant, and we must have $\Omega _{\mathrm{m}} \rightarrow 1$.
504: In the radiation case ($\gamma =4/3$), the results become
505: \[
506: q=1,\ \ \Omega _{\mathrm{YM}}+\Omega _{\mathrm{m}}=1,\ \ H \propto e^{-2\tau }\,.
507: \]%
508: Hereafter, we will consider only the radiation case.
509:
510: From our numerical results, we know that the YM field oscillates with
511: growing frequency while its amplitude is damped. Therefore, we
512: assume that $a\rightarrow 0,a^{\prime }\not\rightarrow 0$. Because the $\Pi
513: _{\pm }$ are both finite\footnote{%
514: In Bianchi I, II, and VI$_{0}$ spacetimes, $-1\leq \Pi _{+}\leq 2$ and
515: $-\sqrt{3}\leq \Pi _{-}\leq \sqrt{3}$ (see \cite{jin}).},
516: eq. (\ref{eq:a}) becomes
517: \begin{equation}
518: a^{\prime \prime }=-a^{\prime }-\frac{1}{H^{2}}a(b^{2}+c^{2})=-a^{\prime
519: }-e^{4\tau }a(b^{2}+c^{2})\,. \label{eq:a_2}
520: \end{equation}%
521: If we divide $a$ into an amplitude and a phase, with $a=e^{-p\tau
522: }e^{i\theta _{1}}$, where $p$ is the damping rate and $\theta _{1}(\tau )$
523: is the time-dependent oscillation frequency, then
524: \begin{equation}
525: a^{\prime }=-pa+i\theta _{1}^{\prime }a=(-p+i\theta _{1}^{\prime })a
526: \label{A}
527: \end{equation}%
528: In order to keep $a^{\prime }\not\rightarrow 0$, $\theta _{1}^{\prime }$ has
529: to grow like
530: \[
531: \theta _{1}^{\prime }\sim O(e^{p\tau })\,.
532: \]%
533: Differentiating eq. (\ref{A}), this gives
534: \begin{eqnarray}
535: a^{\prime \prime } &=&-pa^{\prime }+i\theta _{1}^{\prime \prime }a+i\theta
536: _{1}^{\prime }a^{\prime } \nonumber \\
537: &=&\{p^{2}-\theta _{1}^{\prime }{}^{2}+i(\theta _{1}^{\prime \prime
538: }-2p\theta _{1}^{\prime })\}a\,, \label{eq:a''}
539: \end{eqnarray}%
540: where $\theta _{1}^{\prime }{}^{2}$ grows $O(e^{2p\tau })$ and $\theta
541: _{1}^{\prime \prime }$ depends on the phase of $\theta _{1}^{\prime }$.
542: Substituting in eq. (\ref{eq:a_2}) yields
543: \begin{eqnarray}
544: p^{2}-\theta _{1}^{\prime }{}^{2}+i(\theta _{1}^{\prime \prime }-p\theta
545: _{1}^{\prime }) &=&-(-p+i\theta _{1}^{\prime })-e^{4\tau }(b^{2}+c^{2})
546: \nonumber \\
547: &=&p-i\theta _{1}^{\prime }-e^{4\tau }e^{-2p\tau }(e^{2i\theta
548: _{2}}+e^{2i\theta _{3}})\,, \label{eq:a_3}
549: \end{eqnarray}%
550: where, for the simplicity, we assume $b$ and $c$ damp at the same rate as $a$%
551: , so $b=e^{-p\tau }e^{i\theta _{2}}$ and $c=e^{-p\tau }e^{i\theta _{3}}$.
552: Picking out the dominant terms in eq. (\ref{eq:a_3}), we have
553: \begin{eqnarray}
554: -\theta _{1}^{\prime }{}^{2}+i\theta _{1}^{\prime \prime } &=&-e^{(4-2p)\tau
555: }(e^{2i\theta _{2}}+e^{2i\theta _{3}}) \nonumber \\
556: &=&-e^{(4-2p)\tau }2\cos (\theta _{2}-\theta _{3})e^{i(\theta _{2}+\theta
557: _{3})}\,.
558: \end{eqnarray}%
559: It is generic to assume that there is no relation between $\theta _{2}$ and $%
560: \theta _{3}$, so $\theta _{2}\neq \theta _{3}$. As a result, the
561: characteristic frequencies of the above equation will be neither purely
562: real nor purely imaginary. Therefore it is likely that $O(\theta _{1}^{\prime
563: }{}^{2})\sim O(\theta _{1}^{\prime \prime })$, which means that the
564: leading-order solution of the above equation grows\emph{\ }as $e^{2p\tau }$,
565: i.e., $p=1$.
566:
567: \medskip
568:
569: From this, we can estimate $\Pi _{\pm }$ as follows,
570: \begin{eqnarray}
571: \Pi _{+} &=&-\frac{1}{6}\left( -2a^{\prime }{}^{2}+b^{\prime
572: }{}^{2}+c^{\prime }{}^{2}-2e^{2i(\theta _{2}+\theta _{3})\tau }+e^{2i(\theta
573: _{3}+\theta _{1})\tau }+e^{2i(\theta _{1}+\theta _{2})\tau }\right) \sim O(%
574: \mbox{const.}) \nonumber \\
575: \Pi _{-} &=&\frac{\sqrt{3}}{6}\left( -b^{\prime }{}^{2}+c^{\prime
576: }{}^{2}-e^{2i(\theta _{3}+\theta _{1})\tau }+e^{2i(\theta _{1}+\theta
577: _{2})\tau }\right) \sim O(\mbox{const.})\,.
578: \end{eqnarray}%
579: It follows from our discussion that $\Pi _{\pm }\not\rightarrow 0$ despite
580: the fact that$\ \Sigma \rightarrow 0\,.$ This result is quite unexpected.
581:
582: It is worth checking explicitly that these non-zero anisotropic pressures, $%
583: \Pi _{\pm },$ do not break isotropy. That is, we need to confirm whether
584: these non-zero $\Pi _{\pm }$ make $\Sigma \not\rightarrow 0$ or not. Using
585: eqs. (\ref{eq:Pi+}) and (\ref{eq:Pi-}), and the property that $\Sigma _{\pm
586: }\ll \Pi _{\pm }$, and assuming
587: $\mathcal{A}^2 + \tilde{\mathcal{A}}^2 \sim
588: \mathcal{B}^2 + \tilde{\mathcal{B}}^2 \sim
589: \mathcal{C}^2 + \tilde{\mathcal{C}}^2$,
590: where $\mathcal{A}\equiv a^{\prime }+(-2\Sigma _{+}+1)a$,
591: $\mathcal{B}\equiv b^{\prime }+(\Sigma _{+} + \sqrt{3}\Sigma _{-}+1)b$,
592: $\mathcal{C}\equiv c^{\prime }+(\Sigma _{+} - \sqrt{3}\Sigma _{-}+1)c$,
593: $\tilde{\mathcal{A}}\equiv N_{1}a+\frac{bc}{H}$,
594: $\tilde{\mathcal{B}}\equiv N_{2}b+\frac{ca}{H}$, and
595: $\tilde{\mathcal{C}}\equiv N_{3}c+\frac{ab}{H}$, we have
596: \begin{eqnarray}
597: \Sigma ^{\prime } &=&\frac{(\Sigma ^{2})^{\prime }}{2\Sigma }=\frac{1}{%
598: \Sigma }(\Sigma _{+}\Sigma _{+}^{\prime }+\Sigma _{-}\Sigma _{-}^{\prime
599: })\sim \frac{1}{\Sigma }(\Sigma _{+}\Pi _{+}+\Sigma _{-}\Pi _{-}) \nonumber
600: \\
601: &=&\frac{1}{3}\left[ (\mathcal{A}^{2}+\tilde{\mathcal{A}}^{2})\cos \Psi +(%
602: \mathcal{B}^{2}+\tilde{\mathcal{B}}^{2})\cos \left( \Psi +\frac{2}{3}\pi
603: \right) +(\mathcal{C}^{2}+\tilde{\mathcal{C}}^{2})\cos \left( \Psi -\frac{2}{%
604: 3}\pi \right) \right] \nonumber \\
605: &\sim &\frac{1}{3}\left[ (\mathcal{A}^{2}+\tilde{\mathcal{A}}^{2})\left(
606: \cos \Psi +\cos \left( \Psi +\frac{2}{3}\pi \right) +\cos \left( \Psi -\frac{%
607: 2}{3}\pi \right) \right) \right] =0\,,
608: \end{eqnarray}%
609: where $\Psi \equiv \tan ^{-1}(\Sigma _{-}/\Sigma _{+})\,.$ This shows that
610: the non-zero $\Pi _{\pm }$ modes do \textit{not} break isotropy.
611: The assumption, $\mathcal{A}^2 + \tilde{\mathcal{A}}^2 \sim
612: \mathcal{B}^2 + \tilde{\mathcal{B}}^2 \sim
613: \mathcal{C}^2 + \tilde{\mathcal{C}}^2$, is justified by numerical solutions.
614: Figure \ref{fig:ABC} shows the time averages of $\mathcal{A}^2 + \tilde{\mathcal{A}}^2 \sim
615: \mathcal{B}^2 + \tilde{\mathcal{B}}^2 \sim \mathcal{C}^2 + \tilde{\mathcal{C}}^2$
616: over a period of $0.1 \tau$, which confirms our assumption.
617: %%%%%%%%%%%%%%%%%%%%
618: \begin{figure}[ht]
619: \includegraphics[height=6cm]{ABC.eps}
620: \caption{The typical behaviors of the
621: $\langle\mathcal{A}^2 + \tilde{\mathcal{A}}^2\rangle$,
622: $\langle\mathcal{B}^2 + \tilde{\mathcal{B}}^2\rangle$, and
623: $\langle\mathcal{C}^2 + \tilde{\mathcal{C}}^2\rangle$,
624: where $\langle \cdots \rangle$ means time average per $0.1\tau$.
625: This shows that $\mathcal{A}^2 + \tilde{\mathcal{A}}^2$,
626: $\mathcal{B}^2 + \tilde{\mathcal{B}}^2$, and $\mathcal{C}^2 + \tilde{\mathcal{C}}^2$
627: are nearly equal after the shear is significantly reduced, i.e. $\tau>2$,
628: although $\mathcal{A}^2,\ \mathcal{B}^2,\ \mathcal{C}^2,\
629: \tilde{\mathcal{A}}^2,\ \tilde{\mathcal{B}}^2,$ and $\tilde{\mathcal{C}}^2$
630: oscillate strongly.}
631: \label{fig:ABC}
632: \end{figure}
633: %%%%%%%%%%%%%%%%%%%%
634:
635:
636: \medskip In summary: so far our analyses have shown that if we assume that
637: the radiation-YM universe approaches FRW, in the sense that $\Sigma
638: \rightarrow 0,$ and also that $\Omega _{\mathrm{YM}} \rightarrow \mbox{const.}$
639: and $\Omega _{\mathrm{m}} \rightarrow \mbox{const.},$ then the YM field components
640: satisfy $a\rightarrow 0$, and $a^{\prime} \rightarrow O(\mbox{const.}),$
641: with $\mathcal{A}^2 + \tilde{\mathcal{A}}^2 \sim
642: \mathcal{B}^2 + \tilde{\mathcal{B}}^2 \sim
643: \mathcal{C}^2 + \tilde{\mathcal{C}}^2$,
644: and it is found that $H\propto e^{-2\tau },$ $a\propto
645: O(e^{-\tau }),$ $\Sigma ^{\prime }\rightarrow 0$, and the normalised
646: pressures $\Pi _{\pm }$ oscillate with nearly constant amplitudes. In the
647: next section we provide some further details of this evolution which can be
648: obtained from the numerical studies.
649:
650:
651: \section{A survey of typical evolutionary behaviors}
652:
653: Our numerical studies of the Bianchi I expansion dynamics were performed for
654: various combinations of radiation or dust and YM field. We give here a
655: summary of the principal conclusions from these investigations. These
656: results are stable against small changes in the initial data ($H_{\mathrm{ini}}$
657: and $\Sigma _{\mathrm{ini}}$) and appear to be robust. This robustness
658: is a consequence of the generic nature of the chaotic oscillations of the YM
659: field and the fact that these chaotic oscillations have a small effect
660: on the expansion dynamics of the universe.
661:
662: \subsection{ Radiation and YM field}
663:
664: \subsubsection{YM field dominates: $\ \Omega _{\mathrm{YM}} \gg \Omega _{\mathrm{rad}}$}
665:
666: In the case where the YM field density dominates the density of the
667: isotropic black-body radiation field, we find that
668: $\Omega _{\mathrm{YM}} \sim \mathrm{const.}$ and $\Omega _{\mathrm{rad}}\sim $ \textrm{const. }%
669: during the evolution. If we define an averaged expansion scale factor for
670: the expanding universe, $\ell (t)$, by
671:
672: \[
673: H\equiv \frac{d\ell /dt}{\ell }
674: \]
675: then the normalised shear is found to fall as $\Sigma \propto \ell ^{-1.01}$
676: in the numerical integrations, while the Hubble expansion rate falls at the
677: rate expected in a FRW model, with $H\propto \ell ^{-2.00},$ so the mean
678: scale factor evolves as $\ell \propto t^{0.500}\sim t^{1/2}$, as in
679: a FRW universe.
680:
681: \subsubsection{Radiation dominates: $\Omega _{\mathrm{YM}} \ll \Omega _{\mathrm{rad}}$}
682:
683: If the black-body radiation density dominates the YM field then we still
684: find $\Omega _{\mathrm{YM}}\sim \mathrm{const.}$ and $\Omega _{\mathrm{rad}%
685: }\sim \mathrm{const.}$ and the normalised shear falls almost linearly with
686: the scale factor, as $\Sigma \propto \ell ^{-1.07},$with $H\propto \ell
687: ^{-2.00}$ and $\ell \propto t^{0.500}\sim t^{1/2}$ as in the FRW model. Thus
688: we see that the evolution in both of the radiation plus YM field situations
689: is similar, with
690:
691: \[
692: \Sigma \propto \ell ^{-1} \ \text{and} \ \ell \propto t^{1/2}
693: \]%
694: holding to an excellent approximation. Although the anisotropic pressure can
695: be significant, it is not driving significant shear expansion anisotropy.
696:
697: \subsection{Dust and YM field}
698:
699: \subsubsection{YM field dominates: $\Omega _{\mathrm{YM}} \gg \Omega_{\mathrm{dust}} $}
700:
701: When the YM field dominates over the dust we are in the very early stages of
702: the overall evolution because on average the dust density redshifts away
703: more slowly than the YM field. When the YM field still dominates, so $\Omega
704: _{\mathrm{YM}}\sim 1$, the numerical evolution gives
705:
706: \begin{eqnarray*}
707: \Omega _{\mathrm{dust}} &\propto &\ell ^{0.983}\ \\
708: \Sigma &\propto &\ell ^{-1.01}
709: \end{eqnarray*}%
710: while $H\propto \ell ^{-2.00}$ so $\ell \propto t^{0.500}\sim t^{1/2}.$ This
711: reflects the assumption of the YM field domination and is not inconsistent
712: with eq. (\ref{eq:expansion}) because the assumption used there, $\Omega _{%
713: \mathrm{YM}}$ and $\Omega _{\mathrm{m}}\rightarrow \mathrm{const}$, is
714: violated. We see that the damping rate of shear and the expansion law are
715: the same as that in the radiation-dominated case. It is understandable that $%
716: \Omega _{\mathrm{dust}}\propto \ell ^{0.983}$, because $\mu _{\mathrm{dust}%
717: }\propto \ell ^{-3}$ and $H\propto t^{-1}\propto \ell ^{-2}$. This results
718: in the dependence $\Omega _{\mathrm{dust}}\equiv \mu _{\mathrm{dust}%
719: }/3H^{2}\propto \ell $, as expected. Thus, $\Omega _{\mathrm{dust}}$ grows
720: and will eventually become dominant and the assumption that $\Omega
721: _{\mathrm{YM}} \gg \Omega _{\mathrm{dust}}$ will eventually fail.
722:
723: \subsubsection{Dust dominates: $\Omega _{\mathrm{YM}} \ll \Omega _{\mathrm{dust}}$}
724:
725: This is the natural situation for the universe to evolve into after the
726: radiation-dominated era. Our numerical studies find that in the dust-dominated
727: phase the YM density falls as
728:
729: \[
730: \Omega _{\mathrm{YM}}\propto \ell ^{-0.954}
731: \]
732: while the normalised shear falls as
733:
734: \[
735: \Sigma \propto \ell ^{-1.48}\sim \ell ^{-3/2}
736: \]%
737: and the Hubble rate falls as $H\propto \ell ^{-1.52}$. Hence, we obtain a close
738: approximation to the evolution expected in a dust-dominated FRW universe,
739: with $\ell \propto t^{0.658}\sim t^{2/3}$. As expected, the shear falls off
740: more rapidly than in the radiation-dominated case because of the growing
741: influence of the isotropising dust density. It is also interesting that the
742: shear falls off more rapidly than in the simpler dust plus magnetic
743: universes \cite{Barrow}. We note that the fall of the shear in the YM case
744: is close to the $\Sigma \propto \ell ^{-3/2}$ fall-off that would occur if
745: the YM field were absent in a dust-dominated Bianchi I universe. We see how
746: this arises by noting that
747:
748: \[
749: \Sigma _{+}^{\prime }=-\frac{3}{2}\Sigma _{+}+\Pi _{+}\sim -\frac{3}{2}%
750: \Sigma _{+}
751: \]%
752: because $|\Pi _{\pm }|<\Omega _{\mathrm{YM}}$, and therefore,
753:
754: \[
755: \Sigma _{+}\sim \exp \left(-\frac{3}{2}\tau \right)\sim \ell ^{-3/2}.
756: \]
757:
758:
759: \section{Comparison of Yang-Mills and magnetic fields}
760:
761: One of the original motivations for our study of the evolution of YM fields
762: in the presence of perfect fluids was the unusual behavior found in the
763: case of a pure magnetic field and a perfect fluid, notably in the situation
764: where the perfect fluid is black-body radiation. Since the YM field is a
765: generalisation of a magnetic field it is instructive to compare and contrast
766: the results for these two cases. \emph{\ }
767:
768: In the case of an almost isotropic universe ($\ell \propto t^{1/2}$)
769: containing a pure magnetic field (or other anisotropic stresses with
770: pressure anisotropy of the form (\ref{an})) and black-body radiation with
771: $\Omega _{\mathrm{mag}}\ll \Omega _{\mathrm{rad}}$, the expansion-normalised
772: shear falls logarithmically in time, and
773:
774: \[
775: \Sigma \propto \mu _{\mathrm{mag}}/\mu _{\mathrm{rad}}\propto 1/\ln t\propto
776: 1/\ln \ell
777: \]%
778: This unusual 'critical' evolution arises because there is a zero eigenvalue
779: when we linearise the Bianchi type I magnetic radiation universe around
780: the isotropic Friedmann radiation universe \cite{collins, Barrow}. Note also
781: that the ratio of the magnetic to the black-body density is not constant as
782: is often assumed, but falls slowly due to the coupling to the shear \cite%
783: {zeld}.
784:
785: There is a late-time attractor with $\mu _{\mathrm{mag}}/\mu _{\mathrm{rad}%
786: }\approx \Sigma $ and $\ $%
787: \[
788: \Omega _{\mathrm{mag}}\equiv \mu _{\mathrm{mag}}/3H^{2}\propto \mu _{\mathrm{%
789: mag}}/\mu _{\mathrm{rad}}\propto \Sigma .
790: \]%
791: It is interesting that $\Omega _{\mathrm{mag}}\propto \Sigma $. This means
792: that $\Omega _{\mathrm{mag}}$ can be constrained directly by the CMB
793: temperature anisotropy, $\Delta T/T$, since the presently observed $\Delta
794: T/T\sim \Sigma _{\mathrm{rec}}$ where $z_{\mathrm{rec}}\sim 1100$ is the recombination
795: redshift. After accounting for the short period of dust dominated evolution
796: from the equal-density redshift, $z_{\mathrm{eq}}\sim 10^{4}$, to $z_{\mathrm{rec}}$ this
797: leads to strong limits on any spatially homogeneous cosmological magnetic
798: field today of $3.4\times 10^{-9}(\Omega _{0}h_{50}^{2})^{1/2}G$, \cite{maglim},
799: or on any anisotropic stress with pressures of the form (\ref{an}),
800: see ref. \cite{Barrow} for details and examples.
801:
802: In a dust-dominated era ($\Omega _{\mathrm{mag}}\ll \Omega _{\mathrm{dust}}$%
803: ) the magnetic stresses still slow the fall off of the shear to $\Sigma
804: \propto \ell ^{-1}\propto t^{-2/3},$ whereas we would have had $\Sigma
805: \propto \ell ^{-3/2}$ if the magnetic field was absent. The magnetic density
806: evolves as
807:
808: \[
809: \Omega _{\mathrm{mag}}\propto \mu _{\mathrm{mag}}/\mu _{\mathrm{dust}%
810: }\propto \ell ^{-1}.
811: \]
812:
813: The results for the different magnetic and YM evolutions are summarised in
814: Table \ref{table1}.
815:
816: \begin{table}[h]
817: \begin{tabular}{c||c|c}
818: Material content & Shear Evolution, $\Sigma $ & \ Anisotropic stress density \
819: \\ \hline \hline
820: Dust + magnetic field & $\Sigma \propto \ell ^{-1}\propto t^{-2/3}$ &
821: $\Omega _{\mathrm{mag}}\propto \Sigma \propto t^{-2/3}$ \\ \hline
822: Dust + YM field & $\Sigma \propto \ell ^{-3/2}\propto t^{-1}$ &
823: $\Omega_{\mathrm{YM}}\propto t^{-2/3}$ \\ \hline
824: \ Radiation + magnetic field \ &
825: \ $\Sigma \propto (\ln \ell )^{-1}\propto (\ln t)^{-1}$ \ &
826: $\Omega _{\mathrm{mag}}\propto \Sigma \propto (\ln t)^{-1}$ \\ \hline
827: Radiation + YM field & $\Sigma \propto \ell ^{-1}\propto t^{-1/2}$ &
828: $\Omega_{\mathrm{YM}}=$ const.%
829: \end{tabular}
830: \caption{Evolution of the normalised shear, $\Sigma \equiv \sigma /H$.} \label{table1}
831: \end{table}
832:
833: \section{Observational bounds}
834:
835: \subsection{The microwave background }
836:
837: The key difference that these results display between the magnetic and YM
838: fields is the rapid fall-off in the shear that accompanies the YM evolution.
839: As a result the YM field density does not determine a shear attractor at the
840: end of the radiation era and does not have a significant effect on the CMB
841: anisotropy in the way that the magnetic field does. As a corollary,
842: observations of the CMB temperature anisotropy do not provide a direct and
843: powerful upper bound on the present YM field density in the way that they do
844: for a magnetic field density. A scenario with which $\mu _{\mathrm{YM}}\sim
845: \mu _{\mathrm{rad}}$ is not constrained by shear anisotropy of CMB,
846: although it may be constrained by Big Bang nucleosynthesis (BBN), as we
847: discuss below.
848:
849: The slow decay of the shear in the magnetic universe means that the CMB
850: anisotropy places a stronger limit on the magnetic field density than can be
851: obtained by considering its effects on the expansion rate of the universe at
852: the time of BBN, $z_{\mathrm{BBN}}\sim 10^{9}-10^{10}$.
853: However, in the YM case the rapid fall in the shear
854: as we go forward in time means that a negligible
855: shear at the the time of recombination will be much larger at earlier
856: epochs, with an enhancement factor of
857:
858: \[
859: \frac{\Sigma _{\mathrm{BBN}}}{\Sigma _{\mathrm{rec}}}\sim
860: \frac{1+z_{\mathrm{BBN}}}{1+z_{\mathrm{eq}}}\sim 10^{5}.
861: \]%
862: Since $\Sigma _{\mathrm{BBN}}<0.35$ from the helium-4 abundance \cite%
863: {wainwright-ellis, JB, JB2, JB3, sw}, we see that the BBN anisotropy bound
864: is
865: \begin{equation}
866: \Sigma _{\mathrm{rec}}\lesssim 3.5\times 10^{-6}. \label{bbn}
867: \end{equation}%
868: This is about an order of magnitude stronger than the limit imposed by the
869: CMB since \cite{maglim}%
870: \begin{equation}
871: \Delta T/T=\Sigma _{\mathrm{rec}}\times f(\theta ,\phi ,\Omega _{0}),
872: \end{equation}%
873: where $f\sim O(1)$ is a dimensionless pattern orientation factor, which is
874: bounded as $0.6<f<2.2$ in flat or open universe by the COBE results over $%
875: \theta >10^{0}$ angular scales and so the Bianchi I CMB constraint on $%
876: \Sigma _{\mathrm{rec}}$ is only
877: \begin{equation}
878: \Sigma _{\mathrm{rec}}<10^{-5}\,. \label{cmb}
879: \end{equation}%
880: As discussed in Barrow \cite{pep}, this simple anisotropy evolution leads to
881: an unphysical super-Planck anisotropy energy density in the shear modes at
882: very early times if extrapolated backwards to $z \gg z_{\mathrm{eq}}$ from modest values
883: of shear at $z_{\mathrm{eq}}$. A physically reasonable requirement (the 'Planck
884: Equipartition Principle (PEP)' would be that at $t_{pl}\sim G^{1/2}\sim
885: 10^{-43}s$ all energy densities contributing to the cosmological dynamics
886: should be bounded above by the Planck density $\mu _{pl}\sim t_{pl}^{-4}\sim
887: 10^{94}gm.cm^{-3}$. This would require $\Sigma _{pl}\leq 1$ at \ $z_{pl}\sim
888: 10^{32}$ and hence
889:
890: \[
891: \frac{\Sigma _{\mathrm{rec}}}{\Sigma _{pl}}\sim \frac{1+z_{\mathrm{rec}}}{1+z_{pl}}
892: \sim 10^{-29}
893: \]%
894: and the residual anisotropy in the CMB on large angular scales in the YM
895: Bianchi I universe would be completely negligible. By contrast, in the
896: magnetic universe the slow logarithmic fall in $\Sigma $ allows an
897: interesting anisotropy level $\Sigma _{\mathrm{rec}}\sim O(10^{-5})$ to persist at
898: recombination even if $\Sigma _{pl}\leq 1.$ The predicted value of $\Delta
899: T/T\sim \Sigma _{\mathrm{rec}}$ depends upon the number of relativistic spin states
900: contributing to the equilibrium radiation sea and can also differ
901: if other forms of anisotropy are assumed \cite{pep}. In general, we
902: would expect that YM fields in the more complex Bianchi type VII universes
903: would display slow logarithmic decay of their shear regardless of the
904: presence or absence of the YM fields. This is due to the anisotropic
905: 3-curvature which mimics the presence of an anisotropic 'fluid' of
906: long-wavelength gravitational waves satisfying (\ref{an}) to a good
907: approximation and leads to $\Sigma \propto (\ln t)^{-1}$, see refs. \cite%
908: {DLN, lukash, BJS, BS, Jaffe}. Note however, that these modes are excluded
909: in anisotropic open universes if their spatial topology is compact \cite%
910: {bkodama1,bkodama2}: hyperbolic Bianchi spaces with compact topology must be
911: isotropic.
912:
913: \subsection{Constraints from BBN}
914:
915: In the scenario containing YM fields studied above, the YM density $\Omega _{%
916: \mathrm{YM}}$ does not evolve in proportional to the shear. The YM pressure
917: anisotropy does not dominate the simple 'adiabatic' decay of the shear ($%
918: \sigma \propto \ell ^{-3}$) that occurs in the absence of anisotropic
919: sources or anisotropic 3-curvature. Therefore it is impossible to constrain $%
920: \Omega _{\mathrm{YM}}$ by the shear anisotropy alone, in the way that $%
921: \Omega _{\mathrm{mag}}$ was constrained by shear. Primordial nucleosynthesis
922: gives the strongest constraint on $\Omega _{\mathrm{YM}}$. In our YM
923: scenario, the expansion rate $H$ is larger than in the isotropic FRW case,
924: because
925:
926: \[
927: H^{2}=(8\pi G/3)(\mu _{\mathrm{rad}}+\mu _{\mathrm{YM}})
928: \]%
929: where $\mu _{\mathrm{rad}}$ $\propto T^{4}$ is fixed by observation. The
930: increase in the expansion rate raises the temperature at which
931: neutron-proton abundance ratio freezes out of equilibrium, and this leads to
932: an enhancement in the final helium-4 abundance. The YM field evolves on
933: average like a radiation field and we can use constraints on the density of
934: dark radiation (DR) \cite{Ichiki} to limit its allowed effect in this
935: process. This gave a bound of $\Omega _{\mathrm{DR_{BBN}}}<0.105\Omega _{%
936: \mathrm{rad_{BBN}}}$ at $2\sigma $ confidence level. Therefore, we expect
937: the YM field density to be similarly constrained with the bound
938:
939: \[
940: \Omega _{\mathrm{YM}}<0.105\Omega _{\mathrm{rad}}.
941: \]
942:
943: A similar bound would be obtained from BBN considerations for the magnetic
944: energy density. However, the bound previously obtained from the CMB
945: anisotropy is far stronger in this case: $\Omega _{\mathrm{mag}} \lesssim
946: 10^{-5}\Omega _{\mathrm{rad}}.$
947:
948: \section{Conclusions}
949:
950: We have analysed the evolution of YM fields in anisotropic radiation and
951: dust cosmologies which evolve close to isotropy. The YM field undergoes
952: chaotic oscillations which produce small chaotic vibrations about the
953: Friedmann expansion when the anisotropy level is small. We investigated the
954: evolution of the shear anisotropy and the YM density during the radiation
955: and dust eras. This revealed significant differences to the unusual
956: situation that is known to exist in magnetic cosmologies containing perfect
957: fluids. In particular, unlike in magnetic universes, there is no attractor
958: in the radiation era which couple the YM energy density to the shear
959: anisotropy. Consequently, there is no direct bound on the YM field density
960: from the CMB temperature anisotropy. We have carried out a comparative
961: analysis of the magnetic and YM universes which shows how magnetic universes
962: of Bianchi type I are principally constrained by the CMB anisotropy whilst
963: the YM universes of this type are constrained by the effects of the YM
964: energy density on the synthesis of helium-4. The YM evolution reveals
965: unusual features. Despite the fall in shear to Hubble expansion ratio, the
966: Hubble-normalised pressure anisotropies induced by the YM field do not decay
967: during the radiation era, but oscillate chaotically with constant
968: amplitudes.
969:
970: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
971: \section*{ACKNOWLEDGMENTS}
972: This work was partially supported by a Grant for The 21st Century COE Program
973: (Holistic Research and Education Center for Physics Self-organization Systems)
974: at Waseda University.
975: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
976:
977:
978: \begin{thebibliography}{99}
979:
980: \bibitem{sub} A. Brandenburg and K. Subramanian, Physics Reports \textbf{417},
981: 1 (2005).
982:
983: \bibitem{zeld} Y.B. Zeldovich, Sov. Phys. JETP \textbf{36}, 811 (1972).
984:
985: \bibitem{collins} C.B. Collins, Comm. Math. Phys. \textbf{27}, 37 (1972).
986:
987: \bibitem{Barrow} J.D.~Barrow, Phys. Rev. D \textbf{55}, 7451 (1997).
988:
989: \bibitem{BM} J.D. Barrow and R. Maartens, Phys. Rev. D
990: \textbf{59}, 043502 (1998).
991:
992: \bibitem{BMKK} J.D. Barrow and R. Maartens, Phys. Lett B. \textbf{532}, 153
993: (2002).
994:
995: \bibitem{maglim} J.D. Barrow, P.G. Ferreira, and J. Silk, Phys. Rev. Lett.
996: \textbf{78}, 3610 (1997).
997:
998: \bibitem{YMtypeI} B.K. Darian and H.P. Kunzle, Classical and Quantum Gravity
999: \textbf{13}, 2651 (1996).
1000:
1001: \bibitem{kun} B.K. Darian and H.P. Kunzle, J. Math. Phys. \textbf{38}, 4696
1002: (1997).
1003:
1004: \bibitem{blev} J.D. Barrow and J. Levin, Phys. Rev. Lett. \textbf{80}, 656
1005: (1998).
1006:
1007: \bibitem{jin} Y.~Jin and K.~Maeda, Phys. Rev. D \textbf{71}, 064007 (2005).
1008:
1009: \bibitem{YMKS} E. Donets, D.V. Galt'sov and M.Y. Zotov, Phys. Rev. D \textbf{%
1010: 56}, 3459 (1997).
1011:
1012: \bibitem{mix} C.W. Misner, Phys. Rev. Lett. \textbf{22}, 1071 (1969).
1013:
1014: \bibitem{bkl} V. Belinski, I. Khalatnikov and E.M. Lifshitz, Adv. Phys.
1015: \textbf{19}, 525 (1970).
1016:
1017: \bibitem{jdbchaos1} J.D. Barrow, Phys. Reports \textbf{85}, 1 (1982).
1018:
1019: \bibitem{jdbchaos2} J.D. Barrow, Phys. Rev. Lett., \textbf{46}, 963 (1981).
1020:
1021: \bibitem{jdbchaos3} D. Chernoff and J.D. Barrow, Phys. Rev. Lett. \textbf{50},
1022: 134 (1983).
1023:
1024: \bibitem{wainwright-ellis} J.~Wainwright and G.F.R.~Ellis, \textit{Dynamical
1025: Systems in Cosmology}, (Cambridge University Press, Cambridge, 1997).
1026:
1027: \bibitem{okuyama} P.~Forg\'{a}cs and N.S.~Manton, Commun. Math. Phys. \textbf{%
1028: 72}, 15 (1980).
1029:
1030: \bibitem{JB} J.D. Barrow, Mon. Not. Roy. astr. Soc., \textbf{175}, 359
1031: (1976).
1032:
1033: \bibitem{JB2} J.D. Barrow, Mon. Not. Roy. astr. Soc., \textbf{211}, 221
1034: (1984).
1035:
1036: \bibitem{JB3} J.D. Barrow, Canadian J. Phys., \textbf{64}, 152 (1986).
1037:
1038: \bibitem{sw} S.W. Hawking and R.J. Tayler, Nature \textbf{309}, 1278 (1966).
1039:
1040: \bibitem{pep} J.D. Barrow, Phys. Rev. D \textbf{51}, 3113 (1995).
1041:
1042: \bibitem{DLN} A.D. Doroshkevich, V.N. Lukash and I.D. Novikov, Sov. Phys.
1043: JETP \textbf{37}, 739 (1973).
1044:
1045: \bibitem{lukash} A.D. Doroshkevich, V.N. Lukash and I.D. Novikov, Sov.
1046: Astron. \textbf{18}, 554 (1974).
1047:
1048: \bibitem{BJS} J.D. Barrow, R. Juszkiewicz and D.H. Sonoda, Mon. Not. Roy.
1049: astr. Soc., \textbf{213}, 917 (1985).
1050:
1051: \bibitem{BS} J.D. Barrow and D.H. Sonoda, Physics Reports \textbf{139}, 1
1052: (1986).
1053:
1054: \bibitem{Jaffe} T.~R.~Jaffe, et al. astro-ph/0503213.
1055:
1056: \bibitem{bkodama1} J.D. Barrow and H. Kodama, Class. Quantum Gravity,
1057: \textbf{18}, 1753 (2001).
1058:
1059: \bibitem{bkodama2} J.D. Barrow and H. Kodama, Int. J Mod. Phys. D \textbf{10},
1060: 785 (2001).
1061:
1062: \bibitem{Ichiki} K.~Ichiki, et al. Phys. Rev. D \textbf{66}, 043521 (2002).
1063: \end{thebibliography}
1064:
1065: \end{document}
1066: