Is there a preferred direction in the UniverseIs there a preferred direction in the Universe

P. Jain, IIT Kanpur

There appear to be several indications of the existence of a preferred direction in the Universe (or a breakdown of isotropy)

Optical polarizations from distant AGNsRadio polarizations from distant AGNsLow order multipoles of CMBR

On distance scales of less than 100 Mpc the Universe is not homogeneous and isotropic

The Virgo cluster sits at the center of this disc like structure

Most galaxies in our vicinity lie in a plane (the supercluster plane) which is approximately perpendicular to the galactic plane.

On larger distance scales the universe appears isotropic

CMBRCMBR

What does CMBR fluctuations imply about the isotropy of the universe?

*)1( lmlml aallC

),(),(2

l

l

lmlmlmYaT

CMBR is isotropic to a very good approximation

TT Cross Power SpectrumTT Cross Power Spectrum

The power is low at small l (quadrupole l=2)

The probability for such a low quadrupole to occur by a random fluctuation is 5%

Oliveira-Costa et al 2003

The Octopole is not small but very planar

Surprisingly the Octopole and Quadrupole appear to be aligned with one another with the chance probability =1/62

Quadrupole

Octopole

Cleaned Map

Oliveira-Costa et al 2003

All the hot and cold spots of the Quadrupole and Octopole lie in a plane, inclined at approx 30o to galactic plane

Extraction of Preferred AxisExtraction of Preferred AxisImagine T as a wave function

Maximize the angular momentum dispersion

Oliveira-Costa et al 2003

Extraction of Preferred AxisExtraction of Preferred Axis

k = 1 …3, m = -l … l

Preferred frame ek is obtained by Singular Value Decomposition

e represent 3 orthogonal axes in space

The preferred axes is the one with largest eigenvalue

Ralston, Jain 2003

Alternatively Define

The preferred axis for both Quadrupole and Octopole points roughly in the direction (l,b) (-110o,60o) in Virgo Constellation

Hence WMAP data suggests the existence of a preferred direction (pointing towards Virgo)

We (Ralston and Jain, 2003) show that there is considerable more evidence for this preferred direction

CMBR dipole

Anisotropy in radio polarizations from distant AGNs

Two point correlations in optical polarizations from AGNs

Also point in this directionAlso point in this direction

CMBR Dipole

The dipole is assumed to arise due to the local (peculiar) motion of the milky way, arising due to local in-homogeneities

The observed dipole also points in the direction of Virgo

Physical ExplanationsPhysical ExplanationsMany explanations have been proposed for the anomalous behavior of the low order harmonics

Non trivial topology (Luminet, Weeks, Riazuelo, Leboucq and Uzan, 2003)

Anisotropic Universe due to background magnetic field (Berera, Buniy and Kephart, 2003)

Sunyaev Zeldovich effect due to local supercluster (Abramo and Sodre, 2003)

A satisfactory explanation of the observations is still lacking

Offset angle RM)

RM : Faraday Rotation Measure

= IPA (Polarization at source)

Anisotropy in Radio PolarizationsAnisotropy in Radio Polarizations

shows a Dipole ANISOTROPY

Radio Polarizations from distant AGNs show a dipole anisotropy

Birch 1982Jain, Ralston, 1999Jain, Sarala, 2003

Likelihood Analysis The Anisotropy

is significant at 1% in full (332 sources) data set and 0.06% after making a cut in RM (265 sources)

RM - <RM>| > 6 rad/m

<RM> = 6 rad/m

= polarization offset angle

Distribution of RMDistribution of RM

The cut eliminates the data near the central peak

The radio dipole axis also points towards Virgo

Jain and Ralston, 1999

Anisotropy in Extragalactic Radio PolarizationsAnisotropy in Extragalactic Radio Polarizations

beta = polarization offset angle

Using the cut |RM - <RM>| > 6 rad/m2

Anisotropy in Extragalactic Radio PolarizationsAnisotropy in Extragalactic Radio Polarizations

Using the cut |RM - <RM>| > 6 rad/m2

Galactic Coordinates

Equatorial Coordinates

Anisotropy in Extragalactic Radio PolarizationsAnisotropy in Extragalactic Radio PolarizationsA generalized (RM dependent) statistic indicates that the entire data set shows dipole anisotropy

HutsemHutseméékerskers EffectEffectOptical Polarizations of QSOs appear to be locally aligned with one another. (Hutsemékers, 1998)

A very strong alignment is seen in the direction of Virgo cluster

1<z<2.3

HutsemHutseméékerskers EffectEffect

Equatorial Coordinates

1<z<2.3

Statistical AnalysisStatistical Analysis A measure of alignment is obtained by comparing

polarization angles in a local neighborhood

The polarizations at different angular positions are compared by making a parallel transport along the great circle joining the two points

•Maximizing di() with respect to gives a measure of alignment Di and the mean angle

StatisticStatistic

k, k=1…nv are the polarizations of the nv nearest neighbours of the source i

ki = contribution due to parallel transport

Statistic Jain, Narain and Sarala, 2003

We find a strong signal of redshift dependent alignment in a data sample of 213 quasars

Alignment ResultsAlignment Results

Low polarization sample (p < 2%) High redshift sample (z > 1)

The strongest signal is seen in

Significance LevelSignificance Level

Significance LevelSignificance Level

Significance LevelSignificance Level

Large redshifts (z > 1) show alignment over the entire sky

Alignment Statistic (z > 1)Alignment Statistic (z > 1)

Alignment ResultsAlignment Results

Strongest correlation is seen at low polarizations ( p < 2%) at distance scales of order Gpc

Large redshifts z > 1 show alignment over the entire sky

Jain, Narain and Sarala, 2003

Preferred AxisPreferred AxisTwo point correlation

Define the correlation tensor

Define

where

is the matrix of sky locations

S is a unit matrix for an isotropic uncorrelated sample

Preferred AxisPreferred Axis

Optical axis is the eigenvector of S with maximum eigenvalue

Alignment StatisticAlignment Statistic

Preferred axis points towards (or opposite) to Virgo

Degree of Polarization < 2%

dipole quad octo radio optical

dipole 0.020 0.061 0.042 0.024

quad 0.015 0.023 0.004

octo 0.059 0.026

radio 0.008

Prob. for pairwise coincidencesProb. for pairwise coincidences

Ralston and Jain, 2003

A satisfactory explanation of the observations is so far not available

It is possible that the universe may not be isotropic even at cosmological scales. One should then explore generalization of the FRW metric

the large scale anisotropies could arise due to :

• propagation in a large scale anisotropic medium• The active galactic nuclei may be intrinsically correlated on very large distance scales. Similarly the CMBR quadrupole and octopole may be aligned at the source

Physical ExplanationPhysical Explanation

Alternatively the anisotropies could arise due to the local inhomogeneous distribution of matter

This possibility cannot be ruled out for the CMBR and radio anisotropies but is unlikely to account for the large scale optical correlations, which is a redshift dependent effect

Physical ExplanationPhysical Explanation

The observations may also represent a fundamental violation of Lorentz invariance

Lorentz invariance has been observed to be a very good symmetry of nature.

Theoretically we expect that it is violated due to quantum gravity effects.

We expect violations of order

(M Susy/M Planck)2 (Jain, Ralston 2005)

Physical ExplanationPhysical Explanation

We have been exploring the possibility that the effects may be explained by a light scalar (or pseudoscalar)

Very light mass pseudoscalars (or scalars) are predicted by many theories beyond the Standard Model

Axion (Peccei-Quinn) Supergravity String theory

A very light scalar or pseudoscalar may also be required to explain dark energy

A common model for dark energy is a scalar field slowly rolling towards its true vacuum

Light ScalarsLight Scalars

Coupling to Photons

• Such a scalar field will have an effective coupling to photons

• It does not matter whether is a scalar or a pseudoscalar

• If is a scalar then this interaction breaks parity but parity is not a symmetry of nature.

This leads to reduced intensity of wave if the incident pseudoscalar flux is assumed negligible

As the EM wave passes through large scale background magnetic field, photons (polarized parallel to transverse magnetic field) mix with pseudoscalars

We are basically interested in electromagnetic waves propagating over astrophysical or cosmological distances in the presence of a background magnetic field.

This may also be partially responsible for dimming of distant supernovae (Csaki, Kaloper and Terning, 2002)

The reduction in intensity due to pseudoscalar photon mixing in the local supercluster magnetic field may explain the anomalous CMBR quadrupole and octopole

(Jain and Saha, work in progress)

The wave gets polarized perpendicular to the transverse magnetic field sinceonly the component parallel to the background magnetic field mixes with pseudoscalars

This may explain the optical alignment

However we require magnetic field coherent on cosmologically large distance scales

Polarization

Limit on the coupling

For the invisible axion the current limit on the Peccei-Quinn symmetry breaking scale is 109 GeV,

Mass < 0.01 eV (PDG)

This particle gives very little contribution to mixing for galactic or intergalactic propagation.

It may contribute in regions of strong magnetic fields and plasma density.

We are interested in a pseudoscalar whose mass may be much smaller

g < 6 x 10-11 /GeV (PDG)if we assume that the mass is negligible

We will assume that its mass is smaller or comparable to the plasma density of the medium

Typical scalesBackground magnetic field for the case of Virgo supercluster is roughly 0.1 G, distance 1-10 MpcPlasma density 10-6 cm-3

For intergalactic propagation it may be reasonable to assume many domains of size 1 Mpc and B ≈ 0.005 G Plasma density 10- 8 cm-3

We are interested in the frequency regime from radio to optical, = 10- 5 – 1 eV

Pseudoscalar Photon mixingPseudoscalar Photon mixing

We have considered this mixing in great detail so that it can be tested in future observations

Uniform backgroundTurbulent background (Jain, Panda, Sarala, 2002)Slowly varying background (background magnetic field direction fixed)

(Das, Jain, Ralston, Saha, 2004)Slowly varying background with the direction of magnetic field varying with distance.

(Das, Jain, Ralston, Saha, 2004)

Degree of Polarization as a function of l (or )

Uniform Background

Uniform BackgroundAt source Q=0, U=0.4, V = 0.1

Stokes Parameters as a function of (we set I = 1)

Degree of Polarization as a function of the distance of propagationThe wave is unpolarized at source

Resonant Mixing

Stokes parameter V as a function of Q for several different parameters (varying background magnetic field direction)

V

Q

A background pseudoscalar (scalar) field also leads to a rotation of the polarization of the wave

Background pseudoscalar field

Rotation in polarization =g (

change in the pseudoscalar field along the path

Possible Explanation of Radio Possible Explanation of Radio AnisotropyAnisotropy

An anisotropically distributed background pseudoscalar field of sufficiently large strength can explain the observations

To account for the RM dependence

Pseudoscalar field at source

Concluding RemarksConcluding RemarksThere appears to be considerable evidence that there is a preferred direction in the Universe pointing towards Virgo

However the CMBR observations may also be explained in terms of some local distortion of microwave photons due to supercluster.

The physical mechanism responsible for this is not known so far.

We are considering the possibility that it may be explained due to conversion of photons into pseudoscalars due to propagation through local supercluster magnetic field.

Concluding RemarksConcluding Remarks

It is not possible to attribute optical alignment to a local effect since it is intrisically redshift dependent.We can explain this in terms of pseudoscalar photon mixing provided there exist magnetic fields coherent on cosmological distance scales

Future observations will hopefully clarify the situation

Radio anisotropy may also arise due to some local unknown effect. However it is difficult to find a physical mechanism which can accomplish this.

An anisotropically distributed background pseudoscalar field may explain this effect.