ExDiP 2012, Hokkaido, 11 August, 2012 Anisotropic Infaltion --- Impact of gauge fields on inflation --Jiro Soda Kyoto University Standard isotropic inflation Action S M p2 1 4 4 d x g R d x g V ( ) 2 2 K-G eq. H2 1 3M p2 1/ 2 2.4 1018 GeV ds 2 dt 2 a 2 (t ) dx 2 dy 2 dz 2 Isotropic homogeneous universe Friedman eq. M p 8 G 1 2 V ( ) 2 H a a 3H V '( ) 0 inflation a eH t 2 Origin of fluctuations Curvature perturbations H 2 Rc N H t H 2 H4 1 H2 Ps 4 2 8 2 M p2 Rc ( x1 ) inflation end 1 Rc ( x2 ) 一定 scale invariant spectrum N ( x1 ) Action for GW The relation S M p2 8 4 ij ij d x h h h h ij ij M pl h / 2 Tensor perturbations The tensor to the scalar ratio yields h ( x1 ) N ( x2 ) ( x2 ) initial H M pl 2 H2 Pt 2 2 Mp 2 polarizations P 8 d r t 16 2 Ps M p dN 2 3 From COBE to WMAP! gravitational red shift T T T T : CMB angular power spectrum a mY m , a ma ' m' C 'mm' m The predictions have been proved by cosmological observations. COBE WMAP provided more precise data! We now need to look at a percent level fine structure of primordial fluctuations! 4 Primordial gravitational waves via B-modes E and B Polarizations Primordial gravitational waves Statistical Anisotropy? Eriksen et al. 2004 Hansen et al. 2009 2 P (k ) P0 (k ) 1 g ( k ) k n n : prefered direction Groeneboomn & Eriksen (2008) The preferred direction may be produced by gauge fields. 6 What should we look at? There are two directions in studying fine structures of fluctuations! quantitative improvement qualitative improvement -> spectral tilt -> non-Gaussianity PGW Yet other possibility is the statistical anisotropy If there exists coherent gauge fields during inflation, the expansion of the universe must be anisotropic. Thus, we may have statistical anisotropy in the primordial fluctuations. 7 Gauge fields and cosmic no-hair 8 The anomaly suggests gauge fields? gauge fields FLRW universe S 1 d 4 x g g g F F 4 ds 2 dt 2 a 2 (t) dx 2 dy 2 dz 2 a 2 ( ) d 2 dx 2 dy 2 dz 2 The key feature of gauge fields is the conformal invariance. In terms of a conformal time, we can explicitly see the conformal invariance S 1 4 4 1 1 d x a ( ) F F 2 2 4 a ( ) a ( ) cancelled out Thus, gauge fields are decoupled from the cosmic expansion and hence no interesting effect can be expected. In particular, gauge fields are never generated during inflation. However, ….. 9 Gauge fields in supergravity Supergravity action S d 4 x g R gi j i j e 2 K ( ) g i j DiW ( )D jW ( ) 3 2 W ( ) 2 1 1 d 4 x g Re f ab ( i )F a Fb Im f ab ( i ) Fa Fb 8 4 K ( ) : Kahler potential W ( ) : superpotential fab ( ) : gauge kinetic function Cosmological roles of Kahler potential and super potential in inflation has been well discussed so far. While, the role of gauge kinetic function in inflation has been overlooked. The non-trivial gauge coupling breaks the conformal invariance because the scalar field has no conformal invariance. Thus, there is a chance for gauge fields to make inflation anisotropic. Why, no one investigated this possibility? 10 Black hole no-hair theorem Israel 1967,Carter 1970, Hawking 1972 Let me start with analogy between black holes and cosmology. any initial configurations Black hole has no hair other than M,J,Q J Q M gravitational collapse event horizon By analogy, we also expect the no-hair theorem for cosmological event horizons. 11 Cosmic no-hair conjecture cosmic expansion Gibbons & Hawking 1977, Hawking & Moss 1982 deSitter Inhomogeneous and anisotropic universe T T no-hair? 12 Cosmic No-hair Theorem Wald (1983) Statement: The universe of Bianchi Type I ~ VIII will be isotropized and evolves toward de Sitter space-time, provided there is a positive cosmological constant . Assumption: , p : energy density & pressure other than cosmological constant Dominant energy condition: ex) Strong energy condition: ex) Type IX needs a caveat. 0 3p 0 p p0 Sketch of the proof Ricci tensor (0,0) ds2 dt 2 hij ( x )dx dx K 1 2 1 1 K ij ij 2 T g T t t t 3 2 Mp 0 Einstein equation (0,0) 1 2 K 3 1 1 i j 1 (3) R T t t j i 2 2 Mp 2 Bianchi Type I ~ VIII : 0 Strong Energy Condition Dominant Energy Condition R0 3 K K 3 in time scale : ij 0 t 3 tanh 3 t 3 No Shear = Isotropized (3) T00 0 1 1 hij K ij K ij ij 2 t 3 (3) K 1 K2 0 t 3 R0 ij0 in time scale : t 3 Spatially Flat No matter 14 The cosmic no-hair conjecture kills gauge fields! Since the potential energy of a scalar field can mimic a cosmological constant, we can expect the cosmic no-hair can be applicable to inflating universe. Inflationary universe T T Initial gauge fields No gauge fields! Any preexistent gauge fields will disappear during inflation. Actually, inflation erases any initial memory other than quantum vacuum fluctuations. Predictability is high! 15 Can we evade the cosmic no-hair conjecture? One may expect that violation of energy conditions makes inflation anisotropic. ・Vector inflation with vector potential Ford (1989) 1 L F F V ( A A ) F A A 4 ・・・Fine tuning of the potential is necessary. ・Lorentz violation 1 L F F ( A A m 2 ) 4 Ackerman et al. (2007) ・・・Vector field is spacelike but A0 0 is necessary. ・A nonminimal coupling of vector to scalar curvature 1 1 1 L Fi F i m2 R Ai Ai 4 2 6 Golovnev et al. (2008), Kanno et al. (2008) ・・・More than 3 vectors or inflaton is required All of these models breaking the energy condition have ghost instabilities. Himmetoglu et al. (2009) Anisotropic inflation 17 Gauge fields in inflationary background de Sitter background abelian gauge fields ds2 dt 2 a 2 (t) dx 2 dy 2 dz 2 S 1 4 d x g f 2 (a) F F 4 a () dt 2 2 dx 2 dy 2 dz 2 F A A a( ) e Ht 1 H f depends on time. f ( (t)) gauge symmetry A A It is possible to take the gauge A0 0 , i Ai 0 d 3k ikx Ai 3/2 Ak ( ) i (k)e 2 ki i (k) 0, i (k)i ' (k) ' Ak ( ) Ak ( ) A 2 1 4 2 1 4 k 2 Ak ( ) Ak ( ) S d x g f 2 i Ai 2i Ai 2 d x g f 2 canonical commutation relation Ak , k' ' i ' (k k') k S Ak / f 2 Ak 18 Do gauge fields survive? 2 uk 1 f uk 2 k 2 uk 0 2 f † † Ak uk ()ak uk* ()ak Ak Canonical commutation relation leads to commutation relations and the normalization uk* u i uk uk* k 2 f ak , ak' ' k k' vacuum '† ak 0 0 Thus, it is easy to obtain power spectrums dk 0 Ai (x)A (0) 0 PA (k)eikx k i P(k) blue k 3 u () PA (k) 2 k2 a () 2 The blue spectrum means no gauge fields remain during inflation. The red spectrum means gauge fields survive during inflation. red k 19 Mode functions on super-horizon scales 2 vk 2 1 2 f k vk 0 2 f 2 vk f uk sub-horizon k Take a parametrization a f af ii) c 1 4 uk 1 eik f 2k 2 uk 1 f uk 2 0 2 f 2 uk f 0 uk c1 c% 2 d f2 2c 1 d da d Ha 2 d H H 2 matching at the horizon crossing 1 i) c 4 1 ik e 2k k 0 super-horizon Since we know vk ak H k 1 c2 ak4 c1 fk 2k we get uk c1 c% 2 da c1 c2 a 4 c1 2 2 Ha f Length Super-horizon a k 1 c1 fk 2k H 1 Sub-horizon time 20 Gauge fields survive! Finally, at the end of inflation, we obtain the power spectrum of gauge fields on super-horizon scales i) c 1 4 1 uk fk 2k k 3 uk ( f ) af a k 4 c 1 2 H Ha f PA (k) 2 2 a ( f ) 2 k ii) c 1 4 uk 2c 2 c 1 red spectrum 1 fk 2k k 3 uk ( ) H Ha f PA (k) 2 2 a ( ) 2 k 2 2c1 c 1 2 red spectrum For a large parameter region, we have a red spectrum, which means that there exists coherent long wavelength gauge fields! 21 We should overcome prejudice! According to the cosmic no-hair conjecture, the inflation should be isotropic and no gauge fields survive during inflation. However, we have shown that gauge fields can survive during inflation. It implies that the cosmic no-hair conjecture does not necessarily hold in inflation. Hence, there may exist anisotropic inflation. 22 Gauge fields and backreaction power-law inflation M p2 2 1 S d x g R V ( ) 2 2 4 V V0 e Mp In this case, it is well known that there exists an isotropic power law inflation ds 2 dt 2 t 4/ dx 2 dy 2 dz 2 2 Mp 2 logt In this background, one can consider generation of gauge fields M p2 2 1 1 S d x g R V ( ) f 2 ( ) F F 2 4 2 4 our universe gauge kinetic function f f 0e Mp There appear coherent gauge fields in each Hubble volume. Thus, we need to consider backreaction of gauge fields. H 1 23 Watanabe, Kanno, Soda, PRL, 2009 Exact Anisotropic inflation Kanno, Watanabe, Soda, JCAP, 2010 For homogeneous background, the time component can be eliminated by gauge transformation. Let the direction of the vector to be x – axis. A 0, A (t), 0, 0 x 2 2 4 0, we found the following new solution Mp ds2 dt 2 e2 (t ) e4 (t ) dx 2 e2 (t ) dy 2 dz 2 ds2 dt 2 t 2 t 4 dx 2 t 2 dy 2 dz 2 Then, the metric should be Bianchi Type-I For the parameter region 2 logt Ax (t) Ct 2 8 12 2 8 6 2 4 2 2 4 3 2 4 Apparently, the expansion is anisotropic and its degree of anisotropy is given by & 1 I H & 3 H 2 2 4 I 2 2 0 I 1 H 6 2 H& H 2 2 8 12 2 8 slow roll parameter 24 The phase space structure Kanno, Watanabe, Soda, JCAP, 2010 Quantum fluctuations generate seeds of coherent vector fields. 2 2 4 0 vector Anisotropic inflation scalar Isotropic inflation anisotropy After a transient isotropic inflationary phase, the universe enter into an anisotropic inflationary phase. The result universally holds for other set of potential and gauge kinetic functions. 25 More general cases Hamiltonian Constraint Mp 1 t 1 1 &2 E 2 2 4 4 & & V ( ) f ( )e 32 2 2 2 Scale factor 2 E && 3&2 V ( ) f 2 ( )e4 4 6 Anisotropy E 2 2 && 3&& f ( )e4 4 3 Scalar field ' 3 V ( ) E 2 f 3 ( ) f ( )e4 4 ds 2 dt 2 e2 (t ) e4 (t ) dx 2 e2 (t ) dy 2 dz 2 A (0, v(t), 0 , 0 ) (t ) v f 2 ( )e 4 E const. of integration Behavior of the vector is determined by the coupling Hamiltonian Constraint 2 1 E 1 2 f 2 ( )e4 4 &2 2 V ( ) 2 3 2 Scalar field Conventional slow-roll equations (The 1st Inflationary Phase) d & V ( ) d & V ( ) 3 V ( ) E 2 f 3 ( ) f ( )e4 4 V d V Now we can determine the functional form of f f ( ) e 2 2 e m2 2 V 2 V V d Critical Case 2 2 f e f ( ) e2 To go beyond the critical case, we generalize the function by introducing a parameter c 2 2 f e c 1 Vector grows 1 Vector remains const. 1 Vector is negligible The vector field should grow in the 1st inflationary phase. Can we expect that the vector field would keep growing forever? Attractor mechanism Hamiltonian Constraint 1 3 2 &2 2 1 &2 m 2 E 2 c 2 4 4 e 2 2 2 Scalar field e 2 2 2 e 3 m cE e 2 c2 4 4 2 The opposite force to the mass term Define the ratio of the energy density A E e R m2 2 c 1 2 c 2 4 e4( c 1) The growth should be saturated around Typically, inflation takes place at cE 2ec O (10) Irrespective of initial conditions, we have R= 102 2 4 m2 1 R 2 c 102 Inflaton dynamics in the attractor phase The modified slow-roll equations: The second inflationary phase Hamiltonian Constraint 1 6 &2 m2 2 d 2cE 2 c 2 4 2 2 e d m Scalar field Energy Density 2 2 c 2 2 4 & & 3 m c E e E 2 c 2 4 A e 2 Remember c 1 Solvable e c 2 4 m (c 1) 4( c 1) De 1 c2 E 2 2 1 m2 (c 1) 2c 2 const. of integration We find A becomes constant during the second inflationary phase. m2 E.O.M. for Φ : 3 c 3 m2 (2nd inflationary phase) (1st inflationary phase) 1 c Phase flow: Inflaton Scalar field 3 m2 (1st inflationary phase) 2 m 3 (2nd inflationary phase) Numerically solution at c c2 m 105 1 2 c0 17 The degree of Anisotropy Ratio Anisotropy A E e R m2 2 Attractor Point 2 c 2 4 2 E c 3&& e 3 2 4 4 The degree of anisotropy is determined by & E 2ec 4 H & 9&2 2 Attractor point 2 R (t ) 3 3&2 e c 2 4 m2 (c 1) c2 E 2 2 2 c 1 R (t) H 3 3 c2 2 1 2 2 m :Hamiltonian Constraint 2 The slow-roll parameter is given by 2 Attractor point 2 c 2 Compare We find that the degree of anisotropy is written by the slow-roll parameter. 1 c 1 H 3 c : A universal relation Evolutions of the degree of anisotropy Numerically solution at c0 17 = R (t ) A H becomes constant H 0.3% disappears x grows fast increase Initially negligible 32 Anisotropic Inflation is an attractor ds2 dt 2 e2 Ht e4 t dx 2 e2 t dy 2 dz 2 1 I H H 3 0 I 1 It is true that exponential expansion erases any initial memory. In this sense, we have still the predictability. However, the gauge kinetic function generates a slight anisotropy in spacetime. Statistical Symmetry Breaking in the CMB 33 Phenomenology of anisotropic inflation 34 What can we expect for CMB observables? In the isotropic inflation, scalar, vector, tensor perturbations are decoupled. The power spectrum is isotropic (k1 ) (k 2 ) (k1 k 2 )Ps (k1 | k1 |) h(k1 )h(k 2 ) (k1 k 2 )Pt (k1 | k1 |) However, in anisotropic inflation, we have the following couplings vector-scalar g g g {f ( ) {F F ff 1 f2 em n em I H f V 1 f V H f f length 2 em 1 H I H I Preferred direction n A a k H 1 N (k) t vector-tensor g g g f 2 ( )F F 14 2 43 em n em I H 35 Predictions of anisotropic inflation Thus, we found the following nature of primodordial fluctuations in anisotropic inflation. Watanabe, Kanno, Soda, PTP, 2010 Dulaney, Gresham, PRD, 2010 Gumrukcuoglu,, Himmetoglu., Peloso PRD, 2010 statistical anisotropy in curvature perturbations 2 Ps (k) Ps (k) 1 g s n kö gs 24 I N 2 (k ) n statistical anisotropy in primordial GWs Pt (k) Pt (k) 1 gt n kö 2 preferred direction gt 6 I H N 2 (k ) kˆ cross correlation between curvature perturbations and primordial GWs rc G 24 I H N 2 (k ) TB correlation in CMB These results give consistency relations between observables. 64gt r gs rc 4 gt 36 How to test the anisotropic inflation? The current observational constraint is given by WMAP constraint Pullen & Kamionkowski 2007 Now, suppose we detected gs 24 I N 2 (k ) 0.3 gs 24 I N 2 (k ) 0.3 H 0.02 Then we could expect • statistical anisotropy in GWs gt 1.5 103 • cross correlation between curvature perturbations and GWs G 6 103 If these predictions are proved, it must be an evidence of anisotropic inflation! 37 How does the anisotropy appear in the CMB spectrum? Angular power spectrum of X and Y reads C XY d k PXY k sY *m kˆ For isotropic spectrum, P k P k Y m kˆ s C XY , we have For anisotropic spectrum, there are off-diagonal components. For example, C TB d k PTB k Y *m kˆ Y m kˆ 2 , 1 The off-diagonal part of the angular power spectrum tells us if the gauge kinetic function plays a role in inflation. 38 We should look for the following signals in PLANCK data! When we assume the tensor to the scalar ratio r 0.3 and scalar anisotropy gs 0.3 The off-diagonal spectrum becomes Watanabe, Kanno, Soda, MNRAS Letters, 2011 The anisotropic inflation can be tested through the CMB observation! 39 Non-gaussianity in isotropic inflation 2 1 1 1 S d 4 x g R V ( ) f 2 ( ) F F 2 4 2 f ( ) e c 2 2 e2c a( )2c deSitter k 3 uk ( ) H Ha f PA (k) 2 2 a ( ) 2 k 2 a( ) 1 H 2c2 For c=1, we have the scale invariant spectrum. 2 P P 1192 P NCMB (Ntot NCMB ) : P 3 f NLlocal 1280PNCMB (Ntot NCMB ) : 102 (Ntot NCMB ) 600 k13k33 k1 k2 k3 1 cos2 ( kˆ1 kˆ3 ) cosY00 sinY20 anisotropy Barnaby, Namba, Peloso 2012 ; 0.22 Non-gaussianity in anisotropic inflation c 1 1 6 &2 m2 2 2 d 2 2c 2 E 2 ec 4 d m 3&& m2 c E 2ec 4 2 f ( ) e Scale invariant ceff 1 c 2 2 e2 a( )2 Is an attractor! e c 2 c2 E 2 2 e4 m (c 1) Summary Anisotropic inflation can be realized in the context of supergravity. As a by-product, we found a counter example to the cosmic no-hair conjecture. We have shown that anisotropic inflation with a gauge kinetic function induces the statistical symmetry breaking in the CMB. More precisely, we have given the predictions: the statistical anisotropy in scalar and tensor fluctuations the cross correlation between scalar and tensor the sizable non-gaussianity Off-diagonal angular power spectrum can be used to prove or disprove our scenario. We have already given a first cosmological constraint on gauge kinetic functions. gs 24 I N (k) 0.3 2 2 2 4 0.3 I 2 2 24 N 2 (k) 42