1: \documentclass{emulateapjx}
2:
3: \def\al{\alpha}
4: \def\be{\beta}
5: \def\ga{\gamma}
6: \def\de{\delta}
7: \def\ep{\epsilon}
8: \def\ve{\varepsilon}
9: \def\ze{\zeta}
10: \def\et{\eta}
11: \def\th{\theta}
12: \def\vt{\vartheta}
13: \def\io{\iota}
14: \def\ka{\kappa}
15: \def\la{\lambda}
16: \def\vpi{\varpi}
17: \def\rh{\rho}
18: \def\vr{\varrho}
19: \def\si{\sigma}
20: \def\vs{\varsigma}
21: \def\ta{\tau}
22: \def\up{\upsilon}
23: \def\ph{\phi}
24: \def\vp{\varphi}
25: \def\ch{\chi}
26: \def\ps{\psi}
27: \def\om{\omega}
28: \def\Ga{\Gamma}
29: \def\De{\Delta}
30: \def\Th{\Theta}
31: \def\La{\Lambda}
32: \def\Si{\Sigma}
33: \def\Up{\Upsilon}
34: \def\Ph{\Phi}
35: \def\Ps{\Psi}
36: \def\Om{\Omega}
37:
38: \def\cl{{\cal L}}
39: \def\fr#1#2{{{#1} \over {#2}}}
40: \def\Frac#1#2{{\textstyle{{#1}\over {#2}}}}
41: \def\lsim{\mathrel{\rlap{\lower4pt\hbox{\hskip1pt$\sim$}}
42: \raise1pt\hbox{$<$}}}
43: \def\gsim{\mathrel{\rlap{\lower4pt\hbox{\hskip1pt$\sim$}}
44: \raise1pt\hbox{$>$}}}
45: \def\half{{\textstyle{1\over 2}}}
46: \def\prt{\partial}
47:
48: \newcommand{\beq}{\begin{equation}}
49: \newcommand{\eeq}{\end{equation}}
50: \newcommand{\bea}{\begin{eqnarray}}
51: \newcommand{\eea}{\end{eqnarray}}
52: \newcommand{\rf}[1]{(\ref{#1})}
53:
54: \def\mbf#1{\mbox{\boldmath$#1$}}
55: \def\syjm#1#2{\phantom{}_{#1}Y_{#2}}
56:
57: \def\kf{\hat k_F}
58: \def\kaf{\hat k_{AF}}
59: \def\kfd#1{k_{F}^{(#1)}}
60: \def\kafd#1{k_{AF}^{(#1)}}
61:
62: \def\kjm#1#2#3{k^{(#1)}_{(#2)#3}}
63: \def\kI{\kjm{d}{I}{jm}}
64: \def\kE{\kjm{d}{E}{jm}}
65: \def\kB{\kjm{d}{B}{jm}}
66: \def\kV{\kjm{d}{V}{jm}}
67: \def\kVt{\kjm{3}{V}{jm}}
68: \def\kIdjm#1#2{\kjm{#1}{I}{#2}}
69: \def\kEdjm#1#2{\kjm{#1}{E}{#2}}
70: \def\kBdjm#1#2{\kjm{#1}{B}{#2}}
71: \def\kVdjm#1#2{\kjm{#1}{V}{#2}}
72:
73: \def\etal{et al.}
74:
75: \begin{document}
76:
77: \title{Astrophysical Tests of Lorentz and CPT Violation with Photons}
78:
79: \author{V.\ Alan Kosteleck\'y\altaffilmark{1} and Matthew Mewes\altaffilmark{2}}
80:
81: \altaffiltext{1}{Physics Department, Indiana University,
82: Bloomington, IN 47405, U.S.A.}
83: \altaffiltext{2}{Physics Department, Marquette University,
84: Milwaukee, WI 53201, U.S.A.}
85:
86: \begin{abstract}
87: A general framework for tests of Lorentz invariance
88: with electromagnetic waves is presented,
89: allowing for operators of arbitrary mass dimension.
90: Signatures of Lorentz violations include
91: vacuum birefringence, vacuum dispersion, and anisotropies.
92: Sensitive searches for violations
93: using sources such as active galaxies, gamma-ray bursts,
94: and the cosmic microwave background are discussed.
95: Direction-dependent dispersion constraints are obtained
96: on operators of dimension 6 and 8 using
97: gamma-ray bursts and the blazar Markarian 501.
98: Stringent constraints on operators of dimension 3
99: are found using 5-year data from
100: the Wilkinson Microwave Anisotropy Probe.
101: No evidence appears for isotropic Lorentz violation,
102: while some support at 1$\si$ is found
103: for anisotropic violation.
104: \end{abstract}
105:
106: \keywords{ relativity --- gravitation ---
107: cosmic microwave background --- gamma rays: bursts ---
108: galaxies: active}
109:
110: \maketitle
111:
112: Recent years have seen a resurgence in tests of relativity,
113: spurred in part by the prospect of relativity violations
114: arising in a unified description of nature
115: \citep{ks89,kp91}.
116: Experimental searches for violations of Lorentz invariance,
117: the symmetry underlying relativity,
118: have been performed in a wide range of systems
119: (for data tables, see \citet{tables}).
120: Historically,
121: experiments probing the behavior of light have been central
122: in confirming relativity.
123: Contemporary versions
124: of the classic Michelson-Morley and Kennedy-Thorndike experiments
125: \citep{cavities1,cavities2,cavities4}
126: remain among the most sensitive tests today.
127:
128: Some tight constraints on relativity violations
129: have been achieved by seeking tiny changes in light
130: that has propagated over astrophysical distances.
131: Many of these search for a change in polarization
132: resulting from vacuum birefringence,
133: using sources such as galaxies
134: \citep{cfj,sme2,km_agn,km},
135: gamma-ray bursts (GRB)
136: \citep{grb_bire1,grb_bire2,km_grb,grb_bire3,grb_bire4},
137: and the cosmic microwave background (CMB)
138: \citep{cmb_bire12,cmb_bire23,km_cmb,cmb_bire13,wmap_cpt,xia,kdm}.
139: Others seek a frequency-dependent velocity
140: arising from vacuum dispersion,
141: using GRB, pulsars, and blazars
142: \citep{disp,km,boggs,mmtp,emnss,lps,magic}.
143: Here,
144: we present a general theoretical framework
145: that characterizes Lorentz-violating effects
146: on the vacuum propagation of electromagnetic waves
147: and includes operators of all mass dimensions.
148: We discuss several techniques that
149: can be used to search for the unconventional signals
150: of Lorentz violation,
151: Using vacuum-dispersion constraints
152: from GRB and the blazar Markarian 501,
153: we place new direction-dependent limits
154: on several combinations of coefficients for Lorentz violation.
155: We also perform a search for Lorentz violations
156: in the 5-year results from
157: the Wilkinson Microwave Anisotropy Probe (WMAP)
158: \citep{wmap_cpt,wmap5yr1,wmap5yr2},
159: finding some evidence for anisotropic violations
160: but no support for isotropic violations.
161:
162: At attainable energies,
163: violations of Lorentz invariance are described
164: by a framework called
165: the Standard-Model Extension (SME)
166: \citep{sme1,sme2,sme3}
167: that is based on effective field theory
168: \citep{kp95}.
169: Approaches outside field theory also exist
170: \citep{review}.
171: The SME characterizes
172: all realistic violations affecting known particles and fields,
173: while incorporating otherwise established physics.
174: Much of the work on Lorentz violation
175: has focused on the minimal SME,
176: which restricts attention to
177: gauge-invariant operators of renormalizable dimension.
178: In this work,
179: we consider the gauge-invariant pure-photon sector of the full SME
180: with Lorentz-violating operators of arbitrary dimension,
181: which has Lagrange density
182: \citep{km_cmb}
183: \bea
184: \cl &=& -\Frac 1 4 F_{\mu\nu}F^{\mu\nu}
185: +\Frac 1 2 \ep^{\ka\la\mu\nu}A_\la (\kaf)_\ka F_{\mu\nu}
186: \nonumber \\
187: && \quad
188: - \Frac 1 4 F_{\ka\la} (\kf)^{\ka\la\mu\nu} F_{\mu\nu}\ ,
189: \label{lagrangian}
190: \eea
191: where $A_\mu$ is the 4-potential with field strength $F_{\mu\nu}$.
192: In a flat background with energy-momentum conservation,
193: the Lorentz violation arises through the differential operators
194: \bea
195: (\kaf)_\ka &=&
196: \hspace{-3pt}\sum_{d=\mbox{\scriptsize odd}}
197: {(\kafd{d})_\ka}^{\al_1\ldots\al_{(d-3)}}
198: \prt_{\al_1}\ldots\prt_{\al_{(d-3)}} ,
199: \label{kafs}\\
200: (\kf)^{\ka\la\mu\nu} &=&
201: \hspace{-5pt}\sum_{d=\mbox{\scriptsize even}}
202: (\kfd{d})^{\ka\la\mu\nu\al_1\ldots\al_{(d-4)}}
203: \prt_{\al_1}\ldots\prt_{\al_{(d-4)}} .
204: \label{kfs}
205: \eea
206: The constant coefficients
207: ${(\kafd{d})_\ka}^{\al_1\ldots\al_{(d-3)}}$
208: and
209: $(\kfd{d})^{\ka\la\mu\nu\al_1\ldots\al_{(d-4)}}$
210: characterize the degree of Lorentz violation.
211: The former control CPT-odd operators
212: and are nonzero for odd dimension $d\geq 3$,
213: while the latter control CPT-even operators
214: and are restricted to even $d\geq 4$.
215:
216: The Lagrange density \rf{lagrangian}
217: yields modified Maxwell equations.
218: At leading order in coefficients for Lorentz violation,
219: two plane-wave solutions exist.
220: The corresponding two modified dispersion relations
221: can be written in the form
222: \beq
223: p(\om) \approx [1 + \vs^0 \mp
224: \sqrt{(\vs^1)^2+(\vs^2)^2+(\vs^3)^2}\, ] \om ,
225: \label{disp}
226: \eeq
227: where $p$ and $\om$ are the wavenumber and frequency,
228: respectively.
229: It follows that electromagnetic waves
230: generically contain two propagating modes
231: with different velocities and polarizations.
232: The symbols
233: $\vs^0$,
234: $\vs^1$,
235: $\vs^2$, and
236: $\vs^3$
237: represent certain combinations of coefficients
238: for Lorentz violation,
239: and they depend on the frequency $\om$
240: and direction of propagation $\mbf{\hat p}$.
241: With convenient normalizations,
242: $\vs^1$, $\vs^2$, and $\vs^3$
243: are the Stokes parameters
244: $s^1=Q$, $s^2=U$, and $s^3=V$
245: of the faster mode,
246: while $\vs^0$
247: is a scalar combination analogous to the intensity $s^0=I$.
248: These four combinations completely control
249: the leading-order effects of Lorentz violation
250: on light propagating through empty space.
251: The combination $\vs^3$ depends only
252: on the coefficients
253: ${(\kafd{d})_\ka}^{\al_1\ldots\al_{(d-3)}}$,
254: while $\vs^0$, $\vs^1$, and $\vs^2$
255: depend only on the coefficients
256: $(\kfd{d})^{\ka\la\mu\nu\al_1\ldots\al_{(d-4)}}$.
257:
258: It is convenient to identify a minimal set
259: of coefficient combinations
260: that affect light propagating \it in vacuo. \rm
261: This can be accomplished through
262: spherical-harmonic decomposition.
263: Since $\vs^0$, $\vs^3$ are rotation scalars
264: while $\vs^1$, $\vs^2$ are rotation tensors,
265: their decomposition must involve some form of
266: tensor spherical harmonics.
267: The spin-weighted harmonics
268: $\syjm{s}{jm}(\mbf{\hat p})$
269: provide a well-understood set
270: \citep{sYjm1,sYjm2}.
271: The index $s$ is the spin weight,
272: which up to a sign is equivalent to helicity.
273: Decomposing yields
274: \bea
275: \vs^0 &=&
276: \sum_{djm} \om^{d-4} \, \syjm{0}{jm}(\mbf{\hat n})\, \kI \ ,
277: \nonumber \\
278: \vs^1\pm i \vs^2 &=&
279: \sum_{djm} \om^{d-4} \, \syjm{\pm2}{jm}(\mbf{\hat n})\,
280: \big(\kE\mp i\kB\big) \ ,
281: \nonumber \\
282: \vs^3 &=&
283: \sum_{djm} \om^{d-4} \, \syjm{0}{jm}(\mbf{\hat n})\, \kV \ ,
284: \label{vac_exp}
285: \eea
286: where $j\leq d-2$ and $\mbf{\hat n}=-\mbf{\hat p}$ is a unit vector
287: pointing to the source in astrophysics tests.
288:
289: With this decomposition,
290: all types of Lorentz violations
291: for propagation {\it in vacuo}
292: can now be simply characterized
293: using four sets of spherical coefficients,
294: $\kI$, $\kE$, $\kB$ for CPT-even effects
295: and $\kV$ for CPT-odd effects.
296: For each coefficient,
297: the underlying Lorentz-violating operator
298: has mass dimension $d$
299: and eigenvalues of total angular momentum given by $jm$,
300: as usual.
301: For light from astrophysical sources,
302: dispersion arises when the speed of
303: propagation depends on frequency,
304: which occurs for any nonzero coefficient with $d\neq 4$.
305: Birefringence results when the usual degeneracy
306: among polarizations is broken,
307: for which at least one of $\kE$, $\kB$, $\kV$
308: is nonzero.
309: For example,
310: all operators producing lightspeed corrections
311: that are linear in the energy
312: have $d=5$ and are necessarily birefringent.
313: The only coefficients for nonbirefringent dispersion
314: are therefore
315: $\kI$ with even $d\geq 6$.
316: Since birefringence tests using polarimetry are typically
317: many orders of magnitude more sensitive
318: than dispersion tests using timing,
319: in the following discussion of dispersion
320: we focus only on coefficients for nonbirefringent dispersion.
321:
322: Tests for vacuum dispersion seek differences
323: in the velocity of light at different wavelengths.
324: In the present context
325: with zero birefringent coefficients,
326: the change in velocity is $\de v \simeq -\vs^0$.
327: We see from Eq.\ \rf{vac_exp}
328: that the velocity generically depends
329: on the direction $\mbf{\hat n}$ as well as the frequency $\om$.
330: Typical analyses study explosive or pulsed sources of radiation
331: producing light over a wide wavelength range in short time periods,
332: comparing the arrival times of different wavelengths.
333: This idea has been the focus of many searches
334: based on modified dispersion relations
335: \citep{disp,km,boggs,mmtp,emnss,lps,magic}.
336: Many of these studies assume isotropic violations,
337: which corresponds to the limit $j=m=0$.
338: However,
339: at each dimension $d$,
340: this isotropic restriction misses $(d^2 - 2d - 2)$ possible effects
341: from anisotropic violations.
342:
343: To calculate arrival-time differences
344: in an expanding universe,
345: some care is required
346: \citep{jacob}.
347: In the present case,
348: the photons propagate between two comoving objects,
349: so the relevant coordinate interval is
350: $dl_c = (1+z) dl_p = -v_z dz / H_z$.
351: Here,
352: $v_z$ is the particle velocity at redshift $z$,
353: and
354: $H_z = H_0(\Om_r\ze^4 + \Om_m\ze^3
355: +\Om_k\ze^2+\Om_\La)^{1/2}$
356: with $\ze = 1+z$
357: is the Hubble expansion rate at $z$ in terms of the
358: present-day Hubble constant $H_0 \simeq71$ km/s/Mpc,
359: radiation density $\Om_r\simeq 0$,
360: matter density $\Om_m\simeq 0.27$,
361: vacuum density $\Om_\La\simeq 0.73$,
362: and curvature density
363: $\Om_k = 1-\Om_r-\Om_m-\Om_\La$.
364: The total coordinate distance is the same for all wavelengths,
365: but the travel times may differ.
366: Integrating $dl_c$ from the same initial
367: time to the two arrival times for the two velocities
368: gives a relation for the arrival-time difference $\De t$,
369: which depends on the two energies
370: and the source location on the sky.
371: For the present case with Lorentz violation at dimension $d$,
372: we find
373: \beq
374: \De t \approx -\De \om^{d-4}
375: \int_0^z \frac{(1+z)^{d-4}}{H_z}dz
376: \sum_{jm}
377: \syjm{0}{jm}(\mbf{\hat n}) \kI ,
378: \label{det}
379: \eeq
380: where $\De \om^{d-4}$
381: is the difference in $\om^{d-4}$
382: between the two frequencies.
383:
384: As an illustration,
385: consider the bright gamma-ray burst GRB 021206
386: at right ascension $240^\circ$ and
387: declination $-9.7^\circ$.
388: Over energies from 3 to 17 MeV,
389: arrival-time differences are no more than $\De t < 4.8$ ms
390: for this source at $z \simeq 0.3$
391: \citep{boggs}.
392: Numerical integration of Eq.\ \rf{det}
393: leads to a bound on one direction-specific combination
394: of the 25 independent coefficients
395: for nonbirefringent dispersion with $d=6$:
396: \beq
397: \sum_{jm}\syjm{0}{jm}(99.7^\circ,240^\circ)\,
398: \kIdjm{6}{jm} < 1\times 10^{-16} \mbox{ GeV}^{-2}\ .
399: \label{d6}
400: \eeq
401: For the 49 independent nonbirefringent dispersive operators
402: with $d=8$,
403: we obtain
404: \beq
405: \sum_{jm}\syjm{0}{jm}(99.7^\circ,240^\circ)\,
406: \kIdjm{8}{jm} < 3\times 10^{-13} \mbox{ GeV}^{-4}\ .
407: \label{d8}
408: \eeq
409: Operators with higher $d$ can be treated similarly.
410: Note that many sources are required to constrain fully
411: the coefficient space for a given $d$.
412: In contrast,
413: only one source is needed to constrain fully
414: the corresponding coefficient
415: in the restrictive isotropic limit $j=m=0$.
416: In this limit,
417: the bounds \rf{d6} and \rf{d8} reduce to
418: $\kIdjm{6}{00} < 4\times 10^{-16} \mbox{ GeV}^{-2}$
419: and
420: $\kIdjm{8}{00} < 9\times 10^{-13} \mbox{ GeV}^{-4}$,
421: respectively.
422:
423: As another example,
424: consider Markarian 501,
425: which lies at $z\simeq 0.03$.
426: This source produces flares
427: with photon energies in the TeV range,
428: making it particularly sensitive
429: to an energy-dependent velocity
430: and also to threshold analyses
431: \citep{threshold}.
432: A recent analysis of observations
433: by the MAGIC collaboration
434: found some evidence for a nonbirefringent velocity defect
435: of the form
436: $\de v = - \om/M$ or $\de v = - \om^2/M^2$
437: \citep{magic}.
438: The first case is incompatible with the present treatment;
439: a reanalysis incorporating the necessary birefringence
440: could yield comparatively weak but compatible new bounds.
441: The second case suggests dispersion
442: with $M\simeq 6^{+5}_{-1}\times 10^{10}$ GeV,
443: assuming an arrival-time lag
444: due entirely to nonbirefringent Lorentz violation.
445: For $d=6$,
446: this yields the single constraint
447: \beq
448: \sum_{jm}\syjm{0}{jm}(50.2^\circ,253^\circ)\,
449: \kIdjm{6}{jm} \simeq 3^{+1}_{-2}\times 10^{-22} \mbox{ GeV}^{-2}\ ,
450: \eeq
451: consistent with the GRB bound \rf{d6}.
452: In the isotropic limit,
453: this becomes
454: $\kIdjm{6}{00} \simeq 10^{+4}_{-7}\times 10^{-22} \mbox{ GeV}^{-2}$.
455:
456: Next,
457: we consider tests for vacuum birefringence.
458: In birefringent scenarios,
459: the two plane-wave eigenmodes travel
460: at slightly different velocities,
461: which alters their superposition
462: and hence the net polarization of the light
463: as it propagates in free space.
464: The polarization change is equivalent to
465: a rotation of the Stokes vector
466: $\mbf s = (s^1,s^2,s^3)^T$
467: about the birefringent axis
468: $\mbf\vs = (\vs^1,\vs^2,\vs^3)^T$.
469: The total rotation angle is equal to
470: the relative phase change between the two eigenmodes.
471: Infinitesimally,
472: the rate of rotation is
473: $d\mbf s/dt = -i\Si\cdot\mbf s$,
474: where $\Si^{ab} = -2i\om\ep^{abc}\vs^c$
475: is the rotation generator.
476: Integration from source redshift $z$ to $0$
477: taking into account the cosmological expansion
478: yields the net change in the Stokes vector,
479: \beq
480: \De \mbf s = \int_z^0 \frac{i\Si_z\cdot\mbf s}{(1+z)H_z}\ dz\ ,
481: \label{dstokes}
482: \eeq
483: where $\Si_z$ is the rotation matrix
484: at the blue-shifted frequency $(1+z)\om$
485: and source direction $\mbf{\hat n}$.
486: The net polarization change $\De \mbf s$
487: can depend on frequency and direction of propagation.
488: To seek birefringence,
489: we can either model the polarization
490: at the source and search for discrepancies
491: in the observed polarization,
492: or we can test for unexpected frequency dependences.
493:
494: In what follows,
495: we investigate vacuum birefringence via the CMB,
496: leaving the use of GRB polarimetry in this context
497: to be discussed elsewhere
498: \citep{km-long}.
499: The CMB has a long baseline but comparatively low frequency,
500: which implies lesser sensitivities to $d>3$ violations
501: relative to higher-frequency sources.
502: Here,
503: we focus on the four $d=3$ Lorentz-violating operators.
504: These induce energy-independent polarization changes,
505: so the best constraints are expected from the most distant sources
506: irrespective of frequency.
507: The CMB therefore has the potential to yield maximal sensitivity
508: to these CPT-odd operators.
509: For any CPT-odd case,
510: birefringence causes a rotation of the Stokes vector
511: about the $s^3$ axis,
512: corresponding to a rotation
513: of the linear-polarization angle $\ps$
514: with no change in the degree of linear or circular polarization.
515: For $d=3$,
516: the value of $\ps$ at present is
517: $\ps = \ps_z + \de\ps_z$,
518: where $\ps_z$ is the blueshifted angle
519: and $\de\ps_z$ is its rotation,
520: \beq
521: \de\psi_z=
522: \int_0^z \frac{dz} {(1+z)H_z}
523: \sum_{jm} \syjm{0}{jm}(\mbf{\hat n})\ \kVt .
524: \label{delta_psi}
525: \eeq
526: Taking $z=1100$ for the CMB
527: and including a small radiation component $\Om_r\simeq 0.015$,
528: the rotation reduces to the direction-dependent approximation
529: \beq
530: \de\psi_{\rm CMB} \simeq
531: 3.5^\circ\times10^{43}\ {\rm GeV}^{-1}
532: \sum_{jm}
533: \syjm{0}{jm}(\mbf{\hat n})\
534: \kVdjm{3}{jm} \ .
535: \label{cmb_delta_psi}
536: \eeq
537: We remark in passing that CPT-even operators
538: produce a complicated mixing of circular and linear polarization,
539: rather than a simple rotation of $\psi$
540: \citep{km-long}.
541:
542: We next search for the above effect in the
543: recent WMAP 5-year results
544: \citep{wmap5yr1,wmap5yr2}.
545: We generate initial sky maps of the Stokes parameters
546: using the best fit correlation coefficients $C_j$
547: as calculated by the WMAP collaboration
548: within the $\La$-CDM model
549: assuming gaussianality.
550: The Stokes parameters at each point on the sky
551: are then rotated appropriately and used
552: to calculate the $C_j$ coefficients
553: at the present epoch.
554: The likelihood of these coefficients
555: is determined using available WMAP software.
556: The underlying cosmology is kept fixed,
557: so we are comparing the likelihood of Lorentz violation
558: relative to a reasonable Lorentz-invariant cosmology.
559: Our analysis uses $TE$ and $TB$ data at high-$l$
560: corresponding to $j=24$-450,
561: disregarding the $TT$ data.
562: The latter is a good approximation because
563: the $TT$ data would dominate
564: an analysis with varying cosmology
565: and therefore hold the cosmology comparatively fixed.
566:
567: The correlation coefficients $C_j$ are rotationally invariant,
568: so our analysis has sensitivity only
569: to rotationally invariant combinations
570: of Lorentz-violating coefficients.
571: In the present context,
572: these are the isotropic coefficient
573: $(\kafd{3})^T=- \kVdjm{3}{00}/\sqrt{4\pi}$
574: and the scalar magnitude
575: $|\mbf{\kafd{3}}|=
576: \big(6|\kVdjm{3}{11}|^2+3|\kVdjm{3}{10}|^2\big)^{1/2}/\sqrt{4\pi}$.
577: In particular,
578: our results are independent of the direction of $\mbf{\kafd{3}}$.
579: Although the analysis contains no {\it a priori} anisotropies,
580: the procedure involves generating random realizations
581: that contain anisotropies.
582: As a result, the likelihood $L\big(\kVdjm{3}{jm},r\big)$
583: for a given realization $r$ is anisotropic.
584: In obtaining the total likelihood
585: for a given set of coefficients for Lorentz violation,
586: we sum over the likelihoods of all possible realizations
587: weighted by the probability density $P(r)$,
588: yielding
589: $L\big(\kVdjm{3}{jm}\big) =\sum_r P(r) L\big(\kVdjm{3}{jm},r\big)$.
590: This total likelihood
591: is simply the average over all possible universes
592: and is a rotationally invariant indicator of Lorentz violation.
593: Here,
594: we estimate
595: $L\big(\kVdjm{3}{jm}\big)$ for a range of
596: values of $\kVdjm{3}{jm}$
597: by averaging over 3,000 realizations per value.
598: The results for the four coefficients with $d=3$
599: are shown in Fig.\ \ref{like}.
600:
601: \begin{figure}
602: \epsscale{1}
603: \plotone{f1.eps}
604: \caption{\label{like}
605: Relative likelihood versus the four $d=3$ coefficients
606: for Lorentz violation.
607: Points indicate the values at which the ensemble averages
608: were made,
609: and the bars represent their standard errors.
610: The line is an extrapolation through the points.
611: Dark-gray regions indicate the estimated 68\% confidence interval,
612: while the light-gray region shows the 95\% level.
613: All coefficients are in units of $10^{-42}$ GeV.}
614: \end{figure}
615:
616: For the isotropic coefficient,
617: we obtain the $1\si$ result
618: \beq
619: \kVdjm{3}{00}=(2.3\pm5.4)\times 10^{-43} {\rm ~GeV}.
620: \label{iso}
621: \eeq
622: This improves by about an order of magnitude
623: on the previous limit from radiogalaxy data
624: \citep{cfj}.
625: The result is also consistent with that of the WMAP collaboration,
626: which found a rotation
627: of $\de\psi \simeq 1.2^\circ \pm 2.2^\circ$
628: \citep{wmap_cpt}
629: corresponding to
630: $\kVdjm{3}{00} < (1.2\pm2.2)\times 10^{-43}$ GeV,
631: and that of another recent analysis yielding
632: $\kVdjm{3}{00} < 2.5\times 10^{-43}$ GeV
633: \citep{kdm}.
634: Other reported limits obtained from WMAP 3-year data
635: \citep{wmap3yr}
636: include
637: $\kVdjm{3}{00} = (6.0\pm 4.0) \times 10^{-43}$ GeV
638: \citep{cmb_bire12}
639: and
640: $\kVdjm{3}{00} = (2.5\pm 3.0) \times 10^{-43}$ GeV
641: \citep{cmb_bire13}.
642: Some indication of a nonzero rotation has been found
643: in previous studies.
644: One involving BOOMERANG (B03) data
645: \citep{boomerang1,boomerang2}
646: alone yielded the possibility
647: $\kVdjm{3}{00} = (12\pm 7) \times 10^{-43}$ GeV
648: \citep{km_cmb},
649: while the result from another analysis
650: combining B03 and WMAP 5-year data
651: corresponds to
652: $\kVdjm{3}{00} < (2.6\pm1.9)\times 10^{-43}$ GeV
653: \citep{xia}.
654: While consistent with the latter,
655: our result \rf{iso}
656: shows little evidence for isotropic Lorentz violation.
657:
658: For each anisotropic coefficient,
659: Fig.\ \ref{like} displays the likelihood.
660: As expected,
661: the results are independent
662: of the direction of $\mbf{\kafd{3}}$.
663: The plot symmetries reflect the expected behavior under
664: $\mbf{\kafd{3}}\rightarrow -\mbf{\kafd{3}}$.
665: These plots yield the result
666: \beq
667: |\mbf{\kafd{3}}|=(10^{+4}_{-8})\times 10^{-43} {\rm ~GeV},
668: \label{aniso}
669: \eeq
670: revealing some evidence at 1$\si$
671: for anisotropic Lorentz violation in the WMAP 5-year data.
672: This agrees with the indication
673: of anisotropic Lorentz violation found
674: from an analysis of B03 data,
675: which corresponds to
676: $|\mbf{\kafd{3}}|=(15\pm 6)\times 10^{-43}$ GeV
677: \citep{km_cmb}.
678: The data are consistent with no Lorentz violation at $2\si$,
679: with a 95\% confidence level of
680: $|\mbf{\kafd{3}}|<2\times 10^{-42}$ GeV.
681: This fully constrains the vector components
682: of $\mbf{\kafd{3}}$,
683: and the results \rf{iso} and \rf{aniso}
684: provide a measurement of all four of the $d=3$
685: coefficients for Lorentz violation.
686:
687: \acknowledgments
688: This work is supported in part by the U.S.\ D.o.E.\
689: under grant DE-FG02-91ER40661.
690:
691: \begin{thebibliography}{}
692:
693: \bibitem[Albert \etal (2008)]{magic}
694: Albert,
695: J.\
696: \etal\
697: 2008,
698: Phys.\ Lett.\ B {\bf 668}, 253
699:
700: \bibitem[Amelino-Camelia \etal (1998)]{disp}
701: Amelino-Camelia,
702: G.,
703: Ellis,
704: J.,
705: Mavromatos,
706: N.E.,
707: Nanopoulos,
708: D.V.,
709: and
710: Sarkar,
711: S.\
712: 1998,
713: Nature {\bf 393}, 763
714:
715: \bibitem[Amelino-Camelia \& Piran(2001)]{threshold}
716: Amelino-Camelia,
717: G.,
718: and
719: Piran, T.\
720: Phys.\ Rev.\ D {\bf 64}, 036005
721:
722: \bibitem[Amelino-Camelia(2008)]{review}
723: Amelino-Camelia,
724: G.\
725: 2008,
726: arXiv:0806.0339
727:
728: \bibitem[Antonini \etal (2005)]{cavities2}
729: Antonini,
730: P.\
731: \etal\
732: 2005,
733: Phys.\ Rev.\ A {\bf 71}, 050101(R)
734:
735: \bibitem[Boggs \etal (2004)]{boggs}
736: Boggs,
737: S.E.,
738: Wunderer,
739: C.B.,
740: Hurley,
741: K.,
742: and
743: Coburn,
744: W.\
745: 2004,
746: Astrophys.\ J.\ {\bf 611}, L77
747:
748: \bibitem[Cabella \etal (2007)]{cmb_bire13}
749: Cabella P.,
750: Natoli P.,
751: and
752: Silk, J.\
753: 2007,
754: Phys.\ Rev.\ D {\bf 76}, 123014
755:
756: \bibitem[Carroll \etal (1990)]{cfj}
757: Carroll,
758: S.M.,
759: Field,
760: G.B.,
761: and
762: Jackiw,
763: R.\
764: 1990,
765: Phys.\ Rev.\ D {\bf 41}, 1231
766:
767: \bibitem[Colladay \& Kosteleck\'y(1997)]{sme1}
768: Colladay,
769: D.\
770: and
771: Kosteleck\'y,
772: V.A.\
773: 1997,
774: Phys.\ Rev.\ D {\bf 55}, 6760
775:
776: \bibitem[Colladay \& Kosteleck\'y(1998)]{sme2}
777: Colladay,
778: D.\
779: and
780: Kosteleck\'y,
781: V.A.\
782: 1998,
783: Phys.\ Rev.\ D {\bf 58}, 116002
784:
785: \bibitem[Ellis \etal (2006)]{emnss}
786: Ellis,
787: J.,
788: Mavromatos,
789: N.E.,
790: Nanopoulos,
791: D.V.,
792: Sakharov,
793: A.S.,
794: and
795: Sarkisyan,
796: E.K.G.\
797: 2006,
798: Astropart.\ Phys.\ {\bf 25} 402
799:
800: \bibitem[Fan \etal (2007)]{grb_bire4}
801: Fan, Y.-Z.,
802: Wei, D.-M.,
803: and Xu, D.\
804: 2007,
805: Mon.\ Not.\ R.\ Astron.\ Soc.\ {\bf 376}, 1857
806:
807: \bibitem[Feng \etal (2006)]{cmb_bire12}
808: Feng,
809: B.,
810: Li,
811: M.,
812: Xia,
813: J.-Q.,
814: Chen,
815: X.,
816: and
817: Zhang,
818: X.\
819: 2006,
820: Phys.\ Rev.\ Lett.\ {\bf 96}, 221302
821:
822: \bibitem[Gamboa \etal (2006)]{cmb_bire23}
823: Gamboa, J.,
824: L\'opez-Sarri\'on, J.,
825: and
826: Polychronakos, A.P.\
827: 2006,
828: Phys.\ Lett.\ B {\bf 634}, 471
829:
830: \bibitem[Goldberg(1967)]{sYjm2}
831: Goldberg,
832: J.N.\
833: 1967,
834: J.\ Math.\ Phys.\ {\bf 8}, 2155
835:
836: \bibitem[Hinshaw \etal (2009)]{wmap5yr1}
837: Hinshaw,
838: G.\
839: \etal\
840: 2009,
841: Astrophys.\ J.\ Suppl., in press
842: [arXiv:0803.0732]
843:
844: \bibitem[Jacob \& Piran(2008)]{jacob}
845: Jacob, U.\
846: and
847: Piran, T.\
848: 2008,
849: J.\ Cosmol.\ Astropart.\ Phys.\ {\bf 0801}, 031
850:
851: \bibitem[Jacobson \etal (2004)]{grb_bire2}
852: Jacobson, T.,
853: Liberati, S.,
854: Mattingly, D.,
855: and
856: Stecker, F.W.\
857: 2004,
858: Phys.\ Rev.\ Lett.\ {\bf 93}, 021101
859:
860: \bibitem[Kahniashvili \etal (2006)]{grb_bire3}
861: Kahniashvili,
862: T.\
863: Gogoberidze,
864: G.\
865: and
866: Ratra,
867: B.\
868: 2006,
869: Phys.\ Lett.\ B {\bf 643}, 81
870:
871: \bibitem[Kahniashvili \etal (2008)]{kdm}
872: Kahniashvili,
873: T.,
874: Durrer,
875: R.,
876: and
877: Maravin,
878: Y.\
879: 2008,
880: arXiv:0807.2593
881:
882: \bibitem[Komatsu \etal (2009)]{wmap_cpt}
883: Komatsu,
884: E.\
885: \etal\
886: 2009,
887: Astrophys.\ J.\ Suppl., in press
888: [arXiv:0803.0547]
889:
890: \bibitem[Kosteleck\'y(2004)]{sme3}
891: Kosteleck\'y,
892: V.A.\
893: 2004,
894: Phys.\ Rev.\ D {\bf 69}, 105009
895:
896: \bibitem[Kosteleck\'y \& Mewes(2001)]{km_agn}
897: Kosteleck\'y,
898: V.A.\
899: and
900: Mewes,
901: M.\
902: 2001,
903: Phys.\ Rev.\ Lett.\ {\bf 87}, 251304
904:
905: \bibitem[Kosteleck\'y \& Mewes(2002)]{km}
906: Kosteleck\'y,
907: V.A.\
908: and
909: Mewes,
910: M.\
911: 2002,
912: Phys.\ Rev.\ D {\bf 66}, 056005
913:
914: \bibitem[Kosteleck\'y \& Mewes(2006)]{km_grb}
915: Kosteleck\'y,
916: V.A.\
917: and
918: Mewes,
919: M.\
920: 2006,
921: Phys.\ Rev.\ Lett.\ {\bf 97}, 140401
922:
923: \bibitem[Kosteleck\'y \& Mewes(2007)]{km_cmb}
924: Kosteleck\'y,
925: V.A.\
926: and
927: Mewes,
928: M.\
929: 2007,
930: Phys.\ Rev.\ Lett.\ {\bf 99}, 011601
931:
932: \bibitem[Kosteleck\'y \& Mewes(2009)]{km-long}
933: Kosteleck\'y,
934: V.A.\
935: and
936: Mewes,
937: M.\
938: 2009,
939: arXiv:0905.0031
940:
941: \bibitem[Kosteleck\'y \& Potting(1991)]{kp91}
942: Kosteleck\'y,
943: V.A.\
944: and
945: Potting,
946: R.\
947: 1991,
948: Nucl.\ Phys.\ B {\bf 359}, 545
949:
950: \bibitem[Kosteleck\'y \& Potting(1995)]{kp95}
951: Kosteleck\'y,
952: V.A.\
953: and
954: Potting,
955: R.\
956: 1995,
957: Phys.\ Rev.\ D {\bf 51}, 3923
958:
959: \bibitem[Kosteleck\'y \& Russell(2008)]{tables}
960: Kosteleck\'y,
961: V.A.\
962: and
963: Russell,
964: N.\
965: 2008,
966: {\it Data Tables for Lorentz and CPT Violation,}
967: arXiv:0801.0287
968:
969: \bibitem[Kosteleck\'y \& Samuel(1989)]{ks89}
970: Kosteleck\'y,
971: V.A.\
972: and
973: Samuel,
974: S.\
975: 1989,
976: Phys.\ Rev.\ D {\bf 39}, 683
977:
978: \bibitem[Lamon \etal (2008)]{lps}
979: Lamon,
980: R.,
981: Produit,
982: N.,
983: and
984: Steiner,
985: F.\
986: 2008,
987: Gen.\ Rel.\ Grav.\ {\bf 40}, 1731
988:
989: \bibitem[Lipa \etal (2003)]{cavities1}
990: Lipa,
991: J.\
992: \etal\
993: 2003,
994: Phys.\ Rev.\ Lett.\ {\bf 90}, 060403
995:
996: \bibitem[Mart\'\i nez \& Piran(2006)]{mmtp}
997: Mart\'\i nez,
998: M.R.\
999: and
1000: Piran,
1001: T.\
1002: 2006,
1003: JCAP 0604:006
1004:
1005: \bibitem[Mitrofanov(2003)]{grb_bire1}
1006: Mitrofanov,
1007: I.G.\
1008: 2003,
1009: Nature {\bf 426}, 139
1010:
1011: \bibitem[Montroy \etal (2006)]{boomerang1}
1012: Montroy,
1013: T.E.\
1014: \etal\
1015: 2006,
1016: Astrophys.\ J.\ {\bf 647}, 813
1017:
1018: \bibitem[M\"uller \etal (2007)]{cavities4}
1019: M\"uller,
1020: H.\
1021: \etal\
1022: 2007,
1023: Phys.\ Rev.\ Lett.\
1024: {\bf 99}, 050401
1025:
1026: \bibitem[Newman \& Penrose(1966)]{sYjm1}
1027: Newman,
1028: E.T.\
1029: and
1030: Penrose,
1031: R.\
1032: 1966,
1033: J.\ Math.\ Phys.\ {\bf 7}, 863
1034:
1035: \bibitem[Nolta \etal (2009)]{wmap5yr2}
1036: Nolta,
1037: M.\
1038: \etal\
1039: 2009,
1040: Astrophys.\ J.\ Suppl., in press
1041: [arXiv:0803.0593]
1042:
1043: \bibitem[Page \etal (2007)]{wmap3yr}
1044: Page,
1045: L.\
1046: \etal\
1047: 2007,
1048: Astrophys.\ J.\ Supp.\ {\bf 170}, 335
1049:
1050: \bibitem[Piacentini \etal (2006)]{boomerang2}
1051: Piacentini,
1052: F.\
1053: \etal\
1054: 2006,
1055: Astrophys.\ J.\ {\bf 647}, 833
1056:
1057: \bibitem[Xia \etal (2008)]{xia}
1058: Xia,
1059: J.-Q.,
1060: Li,
1061: H.,
1062: Zhao,
1063: G.-B.,
1064: and
1065: Zhang,
1066: X.\
1067: 2008,
1068: Astrophys.\ J.\ {\bf 679}, L61
1069:
1070: \end{thebibliography}
1071: \end{document}
1072:
1073: