Lepus
X- ray survey: dipole velocity profile
Figure 4.3. Left: A projetion of the intermediate-sale struture inour
neigh-bourhoodto thesupergalati
x − y
plane. Theontoursindiatedensityaordingto
(1, 3, 0.5) × 10 − 3
lusters Mp− 2
respetively. The Shapley onentration repre-sents the most massive strutureinthe shown distane range. Onean learly seethezoneofavoidane,fromwhihoptialdataannotbetaken. Right: dipoleprole
as derived fromreent X-ray galaxy surveys. TheShapley superluster dominantly
ontributesto thedipolebetween
∼ 100
Mpand∼ 200
Mp. Thepituresare takenfrom[TSVZ92 ℄and[KME04 ℄.
onditionandhave
v 2 c ≃ φ
, whihleadsusto(4.2)
∆T (θ, ϕ) T
RS
∼ φ 3/2 ∼ δM
d 3/2
.
Wearegoingtomodelthenon-linearstruturebyaspheriallysymmetriLTBmodelembedded
in a at (
Ω = 1
) Friedmann-Robertson-WalkerUniverse. Substituting theexpression for the massexesswithinthismodel[Pan92℄weobtainthePaneksaling(4.3)
∆T (θ, ϕ) T
RS
∼ δρ
ρ 3/2
d t
3
.
Werepeat,
t
istheosmitimeatwhihtheCMBphotonsrossedthestruture,d
isitsphysialsizeand
δρ/ρ
itsdensityontrast. InsertingtheharateristisoftheShapleysuperluster,we seethatindeed aCMBanisotropyof10 −5
dueto aloalRees-Siamaeetisreasonable.Foralargeangularsaleof thesoureloaland nearbystrutures this indues
ontri-butionsto thelow-
ℓ
multipoles, espeiallythedipole,quadrupole and otopole. This, in turn, ould inlude a non-Doppler ontribution to thedipole. This would imply ahange of afewperent in the inferred dipole veloity, whih might also explain someof theCMB anomalies
[FGM
+
06℄. The Shapleyonentrationis anon-linear struture,and the amplitudeof the
in-duedanisotropiesannot be reliablyalulatedin linearperturbation theory. Aordingto a
omparisonoflinearandexatalulationsforGreatAttrator-likeobjetswiththeLTBmodel
in [FSA94℄, linear theoryis reliableat distanes omparableto the Hubblesale, but failsfor
strutureswithin1000
h −1
Mporso.The advantage of the spherial symmetry of the LTB model is that it allows exat
al-ulations fornon-linear objets; thedrawbakis that theobserved non-linearobjetssuh as
theGreatAttratorandtheShapleyonentrationdonotappearto bespheriallysymmetri.
Figure4.4. AMollweidemapshowingtheforegroundswhihtheWMAP
ollabora-tiontakesintoaountformapleaning.Theonlyextendedforegroundisrepresented
by the galatiregion. Theregions shownin pinkand beige indiatethe so alled
Kp0andKp2diuseemissionmasksusedbytheWMAPollaborationtoobtain
os-mologialmaps. Fordetails ofthemap-makingproess see[J
+
07a ℄ and[H
+
07℄. All
oftheremainingforegroundsthathavebeentakenintoaountarepointsoures. In
this workwe areonsideringthe(Rees-Siama)eetofextendedloalforegrounds,
seeg.4.8. Thepitureistakenfrom[WMAa ℄.
theShapleyonentrationdoesseem toberoughlyspherial[PQC
+
06℄. Also,if thepreferred
diretion indiated by the low-
ℓ
anomalies is due to loal strutures, this implies that thereindeedisadegreeofsymmetryintheloalmassdistribution.
In addition, there is aseond motivation for studying aspheriallysymmetri
inhomoge-neous model, namely dark energy. If interpreted in the framework of isotropi and
homoge-neousosmology,observationsofSNIaimplythattheexpansionoftheUniverseisaelerating,
.f. se. 1.2.2. However, in an inhomogeneous spaetime the observations are not neessarily
inonsistentwith deeleration, see se. 1.3. Inpartiular, in theLTB model theparameter
q 0
denedwiththeluminositydistane isnolongeradiret measureofaeleration[HMM97℄. It
hasbeensuggestedbyseveralgroupsthataspheriallysymmetriinhomogeneityouldbeused
to explain the SNIa data, see se. 1.3, though it is not lear whether suh amodel ould be
onsistentwithwhatisknownaboutstruturesintheloalUniverse[Bol05℄ortheobservation
ofbaryonosillationsinthematterpowerspetrum. HerewewillonernonlytheCMB.
ThepitureoftheloalUniversethatweadoptisaspheriallysymmetridensity
distribu-tion,withtheloalgroupfallingtowardstheoreoftheoverdensityat theentre,.f. g.4.2.
Thelinebetweenourloationandtheentredenesapreferreddiretion
z ˆ
,whihinthepresentaseorrespondstothediretionofthedipole aftersubtratingourmotionwithrespettothe
loalgroupandtheloalgroup'sinfalltowardsthenearbyVirgolusterassumingthe
primor-dialomponentofthedipoletobenegligible. Thediretions ontheskythatareimportantfor
ouranalysis aregivenin tab.4.2. Thissetupexhibits rotationalsymmetrywithrespetto the
axis
z ˆ
negletingtransverseomponentsofourmotion. Consequently,onlyzonal harmonis (m = 0
in thez ˆ
-frame)aregenerated. Wehavealreadyantiipated thisresult,it isonsistentwithourpreditionthatameoutfromtheanalytialtreatmentoftheRees-Siamaeetusing
anLTBmodel inse. 1.3.3. Notethat anyothereet withaxialsymmetrywouldalsoindue
anisotropyonlyin thezonalharmonis.
ThedensityeldhastwoeetsontheCMBseenbyano-entreobserver.First,photons
omingfrom dierentdiretions traveldierent routes throughtheloal overdensity,and this
reatesanisotropyevenwithaperfetlyhomogeneousdistributionofphotons. Inastationary
Figure 4.5. Left: thegalatioordinatesystem. Thegalatioordinatesystem
is denedasbeing parallelwiththeplaneoftheMilkyWayandentredonthesun.
Sotheequatoringalatioordinates(redirle,
0 ◦
galatilatitude)liesintheplaneof ourgalaxy. Thegalatilatitude
b
is theangle aboveor belowthis plane(yellowangle)andthegalatilongitude
l
(greenangle)ismeasuredfrom0 ◦
to360 ◦
,ounterlokwisewithrespettothenorthgalatipole.
0 ◦
ofgalatilongitudeisarbitrarily dened asthe diretionpointing tothe galatientre(Sagittarius). Sometimes,inastronomytheequatorialoordinatesystemisused. Right: therelationofthegalati
oordinatesystemtotheequatorialoordinatesystem. Thelatterisdenedthrough
the planeofthe Earth'sequator. Importantreferene diretionsontheskythat we
usehereare,ingalatioordinates: thenortheliptipole
(l, b) ≃ (96.4 ◦ , 29.8 ◦ )
,theequinox
(l, b) ≃ (276.3 ◦ , 60.2 ◦ )
andthenorthgalatipole(l, b) = (0 ◦ , 90 ◦ )
. Pituresare takenfrom[Ast℄and[Org ℄.
CMB. Seond, the environmentwill aet the evolution of the intrinsi anisotropies as the
homogeneousbakgroundspaedoes,byhangingtheangulardiameterdistane. Theomplete
alulation taking into aount both of these eets would be to study the evolution of the
CMBanisotropiesastheytravelarossthedensityeldusing perturbationtheoryontheLTB
bakground. Asinearliertreatments,weneglettheseondeetandsimplyaddtheanisotropy
generatedbytheLTBmodelontopoftheintrinsiontribution. Itispossiblethatthistreatment
missessomeeetsofproessingtheanisotropiesalreadypresent. Inpartiular,simplylinearly
adding a new soure of anisotropy will in general add multipole power, not redue it, while
aproperanalysis of the proessing of the intrinsi anisotropies ould leadto amultipliative
modiationoftheamplitudes ofthelowmultipoles, asmentionedin[GHHC05℄.
It hasbeen suggestedthat spheriallysymmetri inhomogeneities of the order of horizon
sizeorlargerwouldontributetothelowCMBmultipoles [DZS78,RT81,PP90, LP96℄; itwas
laimedin[Mof05℄ thatthis ouldexplaintheobservedpreferredaxis. Leavingasidethe issue
thatassumingspherialsymmetryfortheentireUniverseseemsquestionable,theobservational
signatureonthelowmultipoles isidentialto thatfrom theLTB modelused todesribeloal
strutures,possiblyapartfromtheamplitude.
4.3. Angular PowerAnalysis
Firstweaddressthequestionhowtheosmimirowaveskyis aetedbythe loal
Rees-Siama eet. We are going to study how maps of the CMB are aeted by the anisotropy
induedbyadditionalaxisymmetri ontributions
a axial ℓ0
onthelargestangular salesbyusing0 1 2 3 4 5 6
0 0.05 0.1 0.15 0.2 0.25 0.3
Likelihood
C 2 in [0.1 mK] 2
WMAP(1yr) WMAP(3yr) a 20 axial = 0 µK [WMAP(1yr) best-fit ΛCDM]
20 µK 40 µK 70 µK
0 2 4 6 8 10 12 14
0 0.05 0.1 0.15 0.2 0.25 0.3
Likelihood
C 3 in [0.1 mK] 2
WMAP(1yr) WMAP(3yr) a 30 axial = 0 µ K [WMAP(1yr) best-fit Λ CDM]
20 µ K 40 µ K 70 µK
Figure 4.6. Likelihood of quadrupole and otopole power for inreased axial
ontributions. Vertial lines denote experimental data: WMAP(1yr) ut-sky and
WMAP(3yr)maximumlikelihoodestimate. Consideringthe quadrupoleaddingany
multipolepower was exluded at
> 99%
C.L. with respet to WMAP(1yr) but itis possible to add up to
60µ
K within the same exlusion level with respet to theWMAP(3yr)value. Adding
80µ
K(100µ
K)tothequadrupoleleadstoanexlusionof99.7%
C.L.(99.9%
C.L.). Theotopoleismoreresistantagainstaxialontaminations as itispossibletoadd awhole100µ
KbeforereahingthesameexlusionlevelwithrespettotheupdatedWMAPdata.
Wesawin se. 3.3.2that theangularpowerspetrumin termsof theoeients
a ℓm
anbeexpressedas
(4.4)
C ℓ = 1
2ℓ + 1 X ℓ
m=−ℓ
| a ℓm | 2 .
Aspreditedbythestandardpereptionofinationaryosmology,theprimordialperturbations
arebelievedtofollowagaussianstatisti. Deviationsfromthiswouldbehardtoreonilewith
thestandardinationary paradigm. Therefore,theomplexoeients
a ℓm = a Re ℓm + i a Im ℓm
areexpetedtobegaussianlydistributed withzeromeanand varianegivenbytheangularpower
C ℓ
,aordingto(4.5)
f (a ℓ0 ) = 1
√ 2πC ℓ
exp
− (a Re ℓ0 ) 2 2C ℓ
and f (a Re,Im ℓm ) = 1
√ πC ℓ
exp − (a Re,Im ℓm ) 2 C ℓ
! .
Therefore,in thestandardmodel, theoeients
a ℓm
are fullyharaterised by theirangular power,forwhihweusethevaluesfromthebesttΛ
CDMtemperaturespetrumtotheWMAP data. Inouraxisymmetrimodel,weparameterisetheeetofaloalstruturebyaddingaxialontributions
a axial ℓ0
to thequadrupoleandotopole. Itis obviousthattheadditivemehanismannotmakethepowerdeitanomaly disappear. Forthestatistial analysiswegenerate
10 5
MonteCarlorealisationsof thequadrupoleand theotopole. Inthe followingwedesribethe
resultsofourMonte Carloanalysisfortheangularpower(4.4)withrespettoone-yearaswell
asthree-yearWMAPdata.
4.3.1. WMAP(1yr) Angular Power. Consideringone-yeardata,thevaluesof
C 2
andC 3
determinedfrom theWMAP ut-sky[H+
03℄, thesoalled TOHmap [TdOCH03℄, the
La-grange ILC map [EBGL04℄ and the ILC map [B
+
03a℄ are listed in tab. 4.3. The extrated
quadrupoleshavebeenDoppler-orretedasdesribedin[SSHC04℄,exeptfortheut-skyvalue.
Thevaluesof
C 2
andC 3
fromthefull-skymapsaresigniantlylargerthantheut-skyvalues.In g. F.1 we show how the
C 2
andC 3
histograms ompare with the one-year data asa axial ℓ0
is inreased. Fora axial ℓ0 = 40µ
K, the number of Monte Carlo hits that are onsistentwith theWMAP ut-skydata is smallerby afator of
∼ 2
for bothC 2
andC 3
asompared0 0.5 1 1.5 2 2.5 3 3.5
0 0.2 0.4 0.6 0.8 1
Likelihood
S nNEP
ILC(3yr)
ILC(1yr)
axis in direction of WMAP dipole a l0 axial = 40 µK 70 µK 0 µK
0 0.5 1 1.5 2 2.5 3 3.5
0 0.2 0.4 0.6 0.8 1
S nEQX
ILC(3yr) ILC(1yr) axis in direction of WMAP dipole a l0 axial = 0 µK 40 µK 70 µK
0 0.5 1 1.5 2 2.5 3 3.5
0 0.2 0.4 0.6 0.8 1
S nNGP
ILC(3yr) ILC(1yr) axis in direction of WMAP dipole
a l0 axial = 40 µK 70 µK 0 µK
0 0.5 1 1.5 2 2.5 3 3.5 4
0 0.2 0.4 0.6 0.8 1
Likelihood
S nNEP
ILC(3yr)
ILC(1yr)
axis in direction of north ecliptic pole a l0 axial = 40 µK 70 µK 0 µK
0 0.5 1 1.5 2 2.5 3 3.5 4
0 0.2 0.4 0.6 0.8 1
S nEQX
ILC(3yr) ILC(1yr) axis in direction of north ecliptic pole
a l0 axial = 40 µK 70 µK 0 µK
0 0.5 1 1.5 2 2.5 3 3.5 4
0 0.2 0.4 0.6 0.8 1
S nNGP
ILC(3yr) ILC(1yr) axis in direction of north ecliptic pole
a l0 axial = 40 µK 70 µK 0 µK
Figure 4.7. WMAPone- and three-year ILCmapsomparedtothe likelihoodof
analignment(4.7)ofquadrupoleandotopolenormalswithastrophysialdiretions
[northeliptipole(NEP),equinox(EQX)andnorthgalatipole(NGP)inolumns℄,
for twoorthogonal realisations ofthe preferred diretion
ˆ z
(WMAP dipole, NEPinrows). The bold histograms represent statistially isotropi and gaussian skies as
preditedbythe
Λ
CDMmodel. Inreasingtheaxialontributionmakestheanomalies worseforˆ z
beingalignedwiththeWMAPdipole,butwiththeexlusionsbeing lesssigniantfor theILC(3yr)thanfor theILC(1yr). AtthesametimeaSolarsystem
eet is preferred by the data. Thenumberof MonteCarlo realisations pertest is
always
10 5
.with the duial CMB sky. For
a axial ℓ0 = 70 µ
K, the number of onsistent Monte Carlo hitsfor
C 2 (C 3 )
is redued by a fator of∼ 5(15)
ompared with the standard CMB sky. Notethatadding anypowerto thetheoretiallyexpeted quadrupole isexluded atthe
> 99%
C.L.level from the ut-sky analysis, but for the otopole the same exlusion level is not reahed
until
a axial 30 = 80µ
K.Further,adding50µ
K(100µ
K)to thequadrupoleleadsto anexlusionof99.6%
C.L.(99.9%
C.L.). Ing.4.6 weshowaomparisonofone-andthree-yeardata.4.3.2. WMAP(3yr) Angular Power. In g. 4.6 we show howthe histograms for the
quadrupoleandotopolepoweromparewiththemeasuredvaluesfromWMAP(1yr,3yr).
Con-sideringtheWMAP(1yr)ut-sky,addinganypowertothequadrupolewasalreadyexludedat
> 99%
C.L. whereasthe WMAP(3yr) data allowsfor addingup toa axial 20 = 60µ
Kin order toreahthesameexlusionlevel. Theotopoleisquiterobustagainstaxialontaminations asit
liesbetteronthet: inordertoreahthesameexlusionlevelof
> 99%
C.L.itisneessarytoadd
a axial 30 = 80µ
KwithrespettotheWMAP(1yr)ut-skyandawholea axial 30 = 100µ
Kwithre-spettotheWMAP(3yr)value. Addingamoderateaxialontributionof
a axial ℓ0 = 40µ
Kleadstoanapproximatebisetionof thenumberofonsistentMonteCarlohits regardingWMAP(1yr)
data(exluded at
99.5%
C.L forC 2
and91.5%
C.L forC 3
), where for theupdated ut-skyaontributionof
a axial ℓ0 = 40µ
Kanbeexludedat> 98%
C.L.forC 2
andonlyat∼ 71%
C.Lfortheotopole.
4.4. Extrinsi AlignmentAnalysis
Nowweaskwhat kindwhat kindofdiretionalpatterns theontribution
a axial ℓ0
indues ontheCMBsky. Inthemultipolevetorrepresentation[CHS04℄anyrealmultipole
T ℓ
onasphereanbeexpressedwith
ℓ
unit vetorsv ˆ (ℓ,i)
andonesalarA (ℓ)
as(4.6)
T ℓ (θ, ϕ) =
X ℓ
m=−ℓ
a ℓm Y ℓm (θ, ϕ) ≃ A (ℓ) Y ℓ
i=1
ˆ
v (ℓ,i) · ˆ e(θ, ϕ) ,
where
e(θ, ϕ) = (sin ˆ θ cos ϕ, sin θ sin ϕ, cos θ)
isaradialunitvetor.Notethattherighthandsideofequation(4.6)ontainsontributionswith`angularmomentum'
ℓ − 2
,ℓ − 4
,...Theuniquenessofthemultipolevetorsisensuredbyremovingthese termsbytakingtheappropriatetraeless
symmetriombination;for detailssee [CHS04℄. Beausethesigns ofallthemultipole vetors
anbeabsorbedintothequantity
A (ℓ)
, theirsignsareunphysialand sooneisfreeto hoosethe hemisphere of eah vetor. Also note that the multipole vetors are independent of the
angularpower. Withthedeomposition(4.6)weahievedauniquefatorisationofamultipole
intoasalarpart
A (ℓ)
,whihmeasuresitstotalpower,andℓ
unitvetorsv ˆ (ℓ,i)
thatontainallthephaseinformation.
Now it is neessary to dene a suitable statisti to ope with the information from the
multipolevetors. Introduingthe
ℓ(ℓ − 1)/2
orientedareasn (ℓ;i,j) ≡ ˆ v (ℓ,i) × v ˆ (ℓ,j) / | v ˆ (ℓ,i) × ˆ v (ℓ,j) |
,wearereadytodeneastatistiinordertoprobealignmentofthenormals
n (ℓ;i,j)
withagivenphysialdiretion
x ˆ
[SSHC04℄,(4.7)
S nx ≡ 1
4 X
ℓ=2,3
X
i<j
n (ℓ;i,j) · x ˆ .
This statistiis a sumoverall dot produtsfor a given
x ˆ
, so it does notimply anyorderingbetweenthetermsandisauniqueandompatquantity. Foromputingthemultipole vetors
we use themethod introdued by [CHS04℄. For mathematial details of the multipole vetor
formalismwereferto e.g.[Fis07℄.
AstheontributionofthestruturedesribedbytheLTBmodel,weaddtothequadrupole
andtheotopoleaomponent,denotedby
a axial ℓ0
,whihisapurem = 0
modewithrespettoagivenphysialdiretion
z ˆ
. Forthediretionx ˆ
weewantto inserttherelevantastrophysial diretionswhih giveriseto alignment,likethediretionoftheelipti plane,theequinoxet.Butthereisaath. Onewerotatethe
z ˆ
axisofourinitialoordinatesystemintothediretionofthepreferredaxisofourmodel,thediretionsonthesky,likenortheliptipoleet.,haveto
berealulatedinthatframe. ThisanbedoneintermsofWignerrotationmatries[CHSS06℄.
Writtenasvetors,theoeients
a ′ ℓ
transformunderrotationsasa ′ ℓ = D † a ℓ
,wherethevetornotationmeansthat
a ℓ
isavetoroftheℓ
-thmultipole oeientwith(2ℓ + 1)
entries andD
denotingtherotation. TherotationsanbeparameterisedintermsofthesoalledEulerangles
α, β, γ
andaregivenin matrixform by[CHSS06℄D m (ℓ) ′ m (α, β, γ) = e im ′ γ d (ℓ) m ′ e imα with d (ℓ) m ′ m = X
k
( − 1) ℓ−m ′ −k [(ℓ + m ′ )!(ℓ − m ′ )!(ℓ + m ′ )!(ℓ − m ′ )!] 1/2 k!(l − m ′ − k)!(l − m − k)!(m + m ′ + k)!
×
cos β 2
2k+m ′ +m sin β
2
2ℓ−2k−m ′ −m
.
(4.8)
WehavearriedouttherotationswiththehelpofaMATHEMATICAroutine. Next,letusreviewour
resultsoftheMonteCarloanalysisforthealignmentstatisti(4.7)withrespettoastrophysial
diretions.
4.4.1. WMAP(1yr) Alignment. Welook for alignmentwith three dierentdiretions
ˆ
x
: thenorthelipti pole,theequinoxandthenorthgalatipole. Thersttwoarepreferreddiretionsin theSolar systemandthe lastdenes theplaneof thedominantforeground. The
observed
S
-valuesfromthedierentCMBmapsaregivenintab.4.3. Theresultsoftheorre-dipole and equinox lieveryloseto eah other, so analignmenttest with the dipole will give
resultsverysimilartotheonewiththeequinox.
In g.F.2 thepreferredaxis
z ˆ
ishosentobethemeasured WMAP(1yr)dipole [B+
03b℄.Weperformalignmenttests(4.7)withrespettothethreetestdiretions
x ˆ
. Forallthreeteststheanomaly gets learly worse, that is the axial mehanismdrives thehistograms awayfrom
thedata. Next,instead of using themotion of theloal group with respet to theCMB rest
frame [KLS
+
93℄ as the test diretion, we takethe veloityof theloal groupwhen orreted
forVirgoentrimotion[PK98℄,sinethisdiersmorefromtheWMAPdipole. Theresultsare
shown in g. F.3. The situation forthe alignment with theequinoxis again worse, but there
is not muh eet on the elipti alignment. For the alignment with the galati plane, the
axialontributionmakesanapparentgalatiorrelationmoreprobable,i.e.thereis aertain
probability of overestimating the galati foreground. For both test diretions by now, the
alignment with the equinox gets worse. Forexample, in the diretion of the Virgo-orreted
loal groupmotionanexlusion of
∼ 99.9%
C.L. fora axial l0 = 50µ
K anbe givenwith respetto all three leaned maps. Note that adding any multipole powerin this test analready be
exludedat the
≥ 99.4%
C.L.Asaomplementarytestweshowthealignmentlikelihoodwithregardtoanorthogonaltest
diretion,namelythenortheliptipole,ing.F.4. Anelipti extraontributionintheCMB
wouldindeedindue analignmentofnormalvetorssimilartotheobservedone. Inpartiular,
for
a axial ℓ0 = 50µ
K, the probability of nding an alignment with the north elipti pole itself beomesroughly5%
, andtheprobabilityfortheequinoxalignmentrisesto1%
.Table 4.3. Testsappliedto variousleanedmaps,asdenedinequation(4.7) ,for
one- andthree-year data,as wellas thevaluesfor angularpower(4.4) .
Foreground-leaned maps: TOH(1yr)is due to [TdOCH03 ℄, LILC (1yr) to [EBGL04℄, the ILC
mapsto[H
+
03 ,H
+
07 ℄andtheMaximumLikelihoodEstimate(MLE)forlow
multi-polesto[H
+
07 ℄. Allone-yearquadrupolesexepttheut-skyvaluehavebeen
Doppler-orreted.
utsky(1yr) TOH(1yr) LILC(1yr) ILC(1yr) ILC(3yr) MLE(3yr)
C 2
129µ
K2
203µ
K2
352µ
K2
196µ
K2
261µ
K2
221µ
K2 C 3
320µ
K2
454µ
K2
571µ
K2
552µ
K2
550µ
K2
545µ
K2
S nNEP
- 0.194 0.193 0.210 0.252-S nEQX
- 0.886 0.866 0.870 0.846-S nNGP
- 0.803 0.803 0.810 0.794-4.4.2. WMAP(3yr) Alignment. Similarly,wetestforalignmentwiththethreegeneri
diretions
x ˆ
: northeliptipole,equinoxandnorthgalatipole. Theresultsoftheorrelationanalysis are shown in g. 4.7: in the rst row the preferred diretion
z ˆ
oinides with thediretion of loal motion, the dipole. Here the anomaly beomes worse when inreasing the
amplitudeoftheaxialontribution. Butfor
x ˆ = NEP
theexlusionbeomessomewhatmildergoingfromone-yeartothree-yeardata;e.g.
a axial ℓ0 = 40µ
Kleadstoanexlusionof99.2%
C.L.forILC(1yr)butonly
98.2%
C.L.fortheupdatedILCmap. Findinganalignmentwiththeequinoxthoughisstronglyexludedat
> 99.2%
C.L.,evenwithanvanishingaxialontributionforboth one-and three-yeardata. Forinstane, forx ˆ = EQX
adding aontribution ofa axial ℓ0 = 20µ
K(
a axial ℓ0 = 70µ
K) leads to an exlusion level of99.4%
C.L. (99.9%
C.L.) with respet tothree-year data. Similarlyto above,aSolarsystemeet is preferredbythe data. Forexample, an
alignmentwiththeeliptiitself(
x ˆ = NEP
)mayonlybeexludedatthelevelof92.3%
C.L.afteraddingan axialontribution of