Gravity-Driven Cosmology N. C. Tsamis (Crete) R. P. Woodard (Florida) CRETE-09-11, CRETE-09-13 1 Particle Production • pair production from the vacuum: 0 → 2E ⇒ energy non-conservation • uncertainty relation: ∆E ∆t ≥ h̄ h̄ violation is not detectable ⇒ ∀ ∆t ≤ 2E • extension to cosmological spacetimes: E(t, k) = q m2 c4 + h̄2 c2 |k|2 a−2(t) Z t+∆t t dt′ 2E(t′ , k) ≤ h̄ ⊲ virtual particle lifetime increases as the mass decreases ⊲ virtual particle lifetime increases if a(t) grows 2 • the de Sitter spacetime case: a(t) = eH0 t ⊲ massless lifetime bound: −H0 ∆t h 2 c |kphys| × 1 − e i ≤ H0 ⊲ thus, massless virtual particles with: k | ≤ H0 a(t) ⇒ real particle production c |kphys| = c | may never recombine • what is the rate for such massless particle production? key distinguishing principle is conformal invariance 3 • conformal transformations with parameter Ω(x): gµν → Ω2 gµν , Aµ → Aµ , 3 −2 ψb → Ω ψb , φ → Ω−1 φ • conformal co-ordinates: dη = a−1(t) dt ⇒ conformal metric tensor: gµν (η, x) = a2(η) ηµν ⇒ a conformally invariant lagrangian is the same as in flat space when expressed in terms of conformally rescaled fields with Ω = a−1 ⇒ production rates from such massless theories are highly suppressed for growing a(t): dη dn Γ dn = = dt dt dη a Γ ≡ number of virtual particles emerging per unit conformal time η 4 η _ H −1 Short wavelength (λphys < H −1 ) graviton pairs (violet) recombine. Long wavelength (λphys > H −1) graviton pairs (red) cannot recombine. • graviton uniqueness: massless, non-conformally invariant also massless & minimally coupled scalar, exotics 5 • how many such gravitons are produced? ⊲ number of infrared gravitons of a particular mode: NIR(t, k) ∼ !2 H t 0 H0 e 2k ≫ 1 ⊲ number of hubble volumes at time t : NH (t) = e3H0 t ≫ 1 ⊲ number of infrared gravitons of any kind per hubble volume is about one • what is their effect on a local observer? ⊲ infrared graviton production ⇒ source imprint ⇒ secular effect ⇒ gravitational response • which observables capture the effect? expansion rate, acceleration observable 6 A Newtonian Model • classical background: de Sitter spacetime ⊲ causal horizon at H0−1 ⊲ manifold T 3 × ℜ • free graviton kinematics: infrared ⊲ degrees of freedom are polarization and wave number k ⊲ infrared modes have H0 ≤ k ≤ H0 eH0t • free graviton dynamics: massless minimally coupled scalar ⊲ go to co-ordinates where background becomes conformally flat ⊲ time evolution of mode k is that of a harmonic oscillator of frequency k and time dependent mass H0 a2(η) ⊲ assume conformal flat space vacuum and do QM on this system 7 • physical energy of infrared mode k at time t: H02 H0 t 1 −H0 t ke + e Ek = 2 4k • newtonian energy density and potential of mode k at t: k e−H0 t Ek − 1 H04 −2H0 t 2 = e ρk = V3(t) 4k ⇒ π GH05 ϕk = − k3 • total infrared newtonian energy density and potential at t: Z H exp[H t] 0 0 H04 1 2 ρIR = dk k ρk = 3 2 8π 2 π H0 H0 Z H exp[H t] 0 0 GH02 1 2 dk k ϕk = − ϕIR = H0 t 3 2 π π H0 H0 8 • total infrared newtonian interaction energy density at t: GH06 H0 t ρnewton = ρIR × ϕIR = − 3 8π ⊲ ρnewton is negative and ever-increasing ⊲ ρnewton starts much smaller than ρIR: ρnewton ∼ GH02 (H0 t) , GH02 ≤ 10−12 ρIR but eventually dominates ⊲ agreement with quantum gravitational perturbative result 9 Quantum Gravity • effective lagrangian and parameters √ 1 LGR = (−2Λ + R ) −g + (counterterms) 16πG ⊲ classical gravity without matter only “knows” about Λ ⊲ the strength of quantum effects is set by G ⊲ the corresponding mass scales are: 1 Λ 4 , M ≡ G 8πG ⊲ the dimensionless coupling constant is: ǫ ≡ GΛ ⊲ perturbation theory is valid iff : 1 2 MPl ≡ ǫ ≡ GΛ < 1 ⇔ M MPl !4 < 1 10 • background geometry: ds2 = −dt2 + a2(t) dx · dx = −dt2 + exp [ 2 b(t) ] dx · dx = a2(η) (−dη 2 + dx · dx) • quantum corrections: h state | γµν (t, x) dxµdxν | state i = gµν (t, x) dxµ dxν γµν (t, x) is the metric operator • quantum-induced stress tensor: ⊲ imprint of geometrically significant differences between classical and quantum backgrounds ⊲ defined from the deficit by which the quantum background gµν fails to obey the classical equations of motion: 1 8πG Tµν ≡ Rµν − gµν R + gµν Λ 2 11 • quantum-induced energy density and pressure: T00(t) = −ρ(t) g00 , i 1 h 2 ḃ (t) − Λ ρ(t) = 8πG T0i(t) = 0 , , Tij (t) = p(t) gij i 1 h p(t) = −2b̈(t) − ρ(t) 8πG • quantum-induced expansion rate: a′(η) ȧ(t) H(t) ≡ = ḃ(t) = 2 = a(t) a (η) s 8πG Λ + ρ(t) 3 3 12 • perturbative results for the de Sitter background adS (t) = ebdS (t) = eH0 t = adS (η) = − 1 H0 η , H02 ≡ 1 Λ>0 3 −1 i ≤ 1 H −1 • spacetime manifold is T 3 × ℜ: − 1 H < x 2 0 2 0 (one causal volume) • initial value problem: bdS (0) = 0 , ḃdS (0) = H0 | statei = | Bunch Davies vacuum at t = 0 i ≡ |0i • fluctuating field ψµν (x): γµν ≡ a2 (ηµν + κ ψµν ) , κ2 ≡ 16πG 13 • results for large observation times (infrared limit) : ρdS (t) = −ǫH04 [ # (H0 t) + O(1) ] + O(ǫ2) pdS (t) = ǫH04 [ # (H0 t) + O(1) ] + O(ǫ2) HdS (t) = H0 n 1 − ǫ2 h #′ (H0 t) + O(1) i o 3 + O(ǫ ) ⊲ the rate of expansion decreases by an amount which becomes non-perturbatively large at late times ⊲ the perturbation theory breakdown occurs when the effective coupling constant becomes of order one : ǫ2 H0 t1 ∼ 1 ⇒ MPl 8 N1 ≡ H0 t1 ∼ ≫ 60 M 14 • why does the effect start at two loops? ⊲ screening represents the gravitational attraction between virtual infrared gravitons ripped from the vacuum ⊲ the graviton production process is a one-loop effect ⊲ the gravitational response to this production cannot occur until the next loop order • infrared logarithms (ln a) ⊲ factors of (H0 t) = ln a that appear in various perturbative results 15 • infrared logarithms rule : in an interaction involving N undifferentiated gravitons, along with any number of other fields, each new factor of the coupling constant squared can produce at most N additional infrared logarithms ⊲ e.g., quantum gravity in de Sitter background: √ basic interaction : G h ∂h ∂h each additional power of G brings at most one extra infrared logarithm ⇒ for instance, the general form of the H(t) corrections is: n H(t) = H0 1 − ∞ X ℓ=2 ǫ ℓ ℓ−1 X k=0 k cℓk (H0 t) o ℓ is the loop order cℓk are pure numbers of O(1) 16 • non-perturbative method ⊲ summation of leading infrared logarithms series: n H(t)|leading log = H0 1 − ǫ ∞ X ℓ−1 cℓ, ℓ−1 (ǫH0 t) ℓ=2 o ⊲ Starobinskiı̆ developed a stochastic technique which sums the leading infrared logarithms of scalar λφ4 • non-perturbative method extensions ⊲ scalar models with bounded below potentials (small, constant increase of the vacuum energy) ⊲ yukawa theory (unbounded decrease of the vacuum energy) ⊲ scalar quantum electrodynamics (small, constant decrease of the vacuum energy) ⊲ quantum gravity? (unknown yet, very complicated) • another approach: phenomenological construction 17 Gravitational versus Scalar Inflation • initial conditions ⊲ gravitational case: Λ>0 , matter | < Λ |Tµν (i.e. matter stress-energy does not “overwhelm” Λ) ⇒ both are robust conditions ⊲ scalar case: inflaton field must be homogeneous over more than one hubble region and without kinetic energy ⇒ quite unlikely • potential issues ⊲ scalar case: inflaton potential needs to be flat ⇒ constraints on couplings gravity case: Λ is a constant ⊲ scalar case: inflaton potential needs Vmin ∼ 0 ⇒ cosmological constant problem gravity case: screening of Λ 18 • potential issues (contd.) ⊲ scalar case: inflaton potential is arbitrary ⇒ not predictive gravity case: gravitation exists • reheating issues ⊲ scalar case: inflaton transfers energy via coupling to matter ⇒ different couplings to different matter fields ⇒ generation of Veff from quantum corrections ⊲ scalar Veff example: – start with V (φ) = λ φ4 , |λ| < 10−14 – couple φ to matter with, say, strength g ⇒ generation of Veff = ± g 2 φ4 ln(φ/µ) ( + for bosons, − for fermions; instability for the latter for large φ ) h i 2 λ ± g ln(φ/µ) φ4 ⇒ Vtot = ⇒ fine-tuning to make bracketed term very small 19 • reheating issues (contd.) ⊲ gravity case: gravitation couples universally but is weak ⇒ normally, reheating is generated non-gravitationally → exception: if - besides the zero-mode - all the modes participate in the reheating, gravity can become strong and capable of naturally potent reheating → danger: gravity may reheat the universe before it stopped inflating 20 A Simple Model for Early Cosmology • the phenomenological goal: what is the most cosmologically significant part of the effective field equations? 1 Gµν ≡ Rµν − gµν R = −Λ gµν + 8πG Tµν [g] 2 • the perfect fluid form: Tµν [g] = (ρ + p) uµ uν + p gµν is determined completely by providing: (i) the gravitationally induced energy density ρ[g](x) (ii) the gravitationally induced energy pressure p[g](x) (iii) the 4-velocity field uµ[g](x): g µν uµuν = −1 (timelike and normalized) 21 • why go to the effective field equations directly? ⊲ very hard to identify “in-in” effective action terms ⊲ avoid changes in the effective newton constant G • the perfect fluid conservation: D µ Tµν = 0 there are 4 conservation equations and 5 independent quantities in Tµν (ρ, p, normalized uµ) ⇒ must only specify one quantity; we choose the pressure 22 • requirements on the pressure: ⊲ should not alter the initial value problem ⊲ should be non-local, – effect is inherently non-local – any local modification simply renormalizes Λ ⊲ should be simple, – a simple non-local operator is the inverse of: √ 1 µν ≡ √ −g ∂ν ) ∂µ ( g −g (defined with retarded boundary conditions) – a simple scalar it can act on is the curvature scalar R: X ≡ 1 R ⊲ should reproduce our perturbative de Sitter result 23 • relevant spacetimes and parameters: ds2 = gµν (t) dxµdxν = −dt2 + a2(t) dx · dx d ȧ(t) = ln a(t) a(t) dt H(t) ≡ q(t) ≡ − T00 = ρ , T0i = 0 R = 12H 2 + 6Ḣ X = 1 , Ḣ(t) a(t) ä(t) = −1 − ȧ2(t) H 2(t) , 1 Tij = gij p uµ = −δµ0 ; t′ 1 = − dt′ 3 ′ dt′′ a3(t′′ ) a (t ) 0 0 Z t Z Z t′ h i 1 ′′ 3 ′′ 2 ′′ 2 ′′ ′ R = − dt a (t ) 12H (t ) + 6Ḣ (t ) dt 3 ′ a (t ) 0 0 Z t 24 • the field and conservation equations: 3H 2 = Λ + 8πG ρ −2Ḣ − 3H 2 = −Λ + 8πG p ρ̇ = −3H (ρ + p) ⇒ t 1 ρ(t) = −p(t) + 3 dt′ a3(t′ ) ṗ(t′ ) a (t) 0 Z • the de Sitter correspondence limit: HdS (t) = H0 > 0 RdS = 12H 2 , , 1 qdS (t) = −1 |dS = − Z t 0 , dt′ e−3H0 adS (t) = eH0 t t′ Z t′ 0 ′′ dt′′ e3H0 t i 4h −3H t 0 XdS = −4H0 t + 1−e ≃ −4 ln[ adS (t) ] + O(1) 3 25 • the de Sitter physical ansatz for the source: p[gdS ](t) = Λ2 f [−ǫ XdS ](t) , ǫ ≡ GΛ reproduces the monotonically unbounded perturbative result: (up to a positive coefficient) H 2 ≃ H02 { 1 − 32πǫ2 ln(adS ) + O(ǫ3) } provided f is some monotonically unbounded function: f [−ǫ XdS ] = −ǫ XdS + O(ǫ2) ⊲ XdS is monotonically increasing and negative ⊲ f, p are monotonically increasing and positive 26 • the general physical ansatz for the source for a general geometry, we take the gravitationally induced pressure to equal: p[g](x) = Λ2 f [−ǫ X](x) , X ≡ 1 R f [−ǫ X] = −ǫ X + O(ǫ2) ⊲ given p we determine ρ and uµ by conservation (up to their initial value data) • the linear and exponential models two simple choices for f are: (i) the linear model : (ii) the exponential model : f [−ǫ X] ≡ −ǫ X f [−ǫ X] ≡ e−ǫ X − 1 27 • numerical evolution ⊲ the relevant equation is: 2Ḣ + 3H 2 = 3H02 { 1 − 8πǫ f [−ǫ X] } , X ≡ 1 R ⊲ its discretization involves constants: step size in Hubble units ⇒ δ ≡ H0 ∆t ǫ ≡ GΛ = 3GH02 coupling constant ⇒ ⊲ and the basic variables with their initial value data: a(t) → a(i ∆t) = ebi (ρ + p)(t) → [ρ + p](i ∆t) , , b0 = 0 [ρ + p]0 = 0 [ρ + p]i+1 = e−3bi { [ρ + p]i − ǫ ∆Xi f ′[−ǫ Xi ] } ⊲ all quantities of interest and their initial values can be determined from the above 28 • numerical results ⊲ discretized evolution equation: 3 2 ∆ bi = [ δ − (∆bi)2 ] − 12πδ 2ǫ f [−ǫ Xi ] 2 2 ⊲ results are for the exponential model: f (x) = ex − 1 f −1(x) = ln(1 + x) ⇒ , f ′ (x) = ex for the following choice of input parameters and step range: δ = 1 1000 , ǫ = 1 200 ; i ∈ [0, 350000] ⊲ numerical integration used MATHEMATICA 29 • the source behaviour: growth → oscillations X 50 000 100 000 150 000 200 000 250 000 300 000 t 350 000 -100 -200 -300 -400 X -437.5 -438.0 -438.5 200 000 250 000 300 000 t 350 000 30 • the curvature behaviour: decrease → oscillations R R 0.000012 6. ´ 10-8 0.00001 4. ´ 10-8 8. ´ 10-6 2. ´ 10-8 6. ´ 10-6 4. ´ 10-6 200 000 2. ´ 10-6 250 000 300 000 t 350 000 -2. ´ 10-8 50 000 t 100 000 150 000 200 000 250 000 300 000 350 000 -4. ´ 10-8 ⊲ oscillations are centered around R = 0 ⊲ oscillations have an envelope falling like t−1 31 • the expansion rate behaviour: decrease → oscillations H H 0.0010 0.00006 0.0008 0.00005 0.00004 0.0006 0.00003 0.0004 0.00002 0.0002 0.00001 50 000 t 100 000 150 000 200 000 250 000 300 000 350 000 200 000 250 000 300 000 t 350 000 ⊲ there is net expansion ⊲ there are short periods of H < 0 (novel feature) 32 • the Ḣ(t) behaviour: increase → oscillations è è H H 5. ´ 10-9 5. ´ 10-9 t 50 000 100 000 150 000 200 000 250 000 300 000 350 000 200 000 -5. ´ 10-9 -5. ´ 10-9 -1. ´ 10-8 -1. ´ 10-8 250 000 300 000 t 350 000 ⊲ during the oscillations era we see that: R(t) and 6Ḣ(t) are almost equal ⇒ R = 12H 2 + 6Ḣ ≃ 6Ḣ 33 • the end of inflation: q(t) goes to positive values q 300 200 100 200 000 250 000 300 000 t 350 000 -100 -200 -300 q 1.0 0.5 110 000 120 000 130 000 140 000 150 000 160 000 t 170 000 -0.5 -1.0 34 • the scale factor: 30 25 20 15 10 5 200 000 250 000 300 000 350 000 ⊲ the evolution of the scale factor ratio [a(t)/a(150000)] during the oscillation regime versus a linear interpolation 35 • analytical results • criticality: ⊲ f is monotonically unbounded, therefore ∃ Xcr : 1 − 8πǫ f [−ǫ Xcr ] = 0 ⇒ Xcr = − 1 −1 1 f ( ) ǫ 8πǫ ⊲ inflationary evolution dominates roughly until criticality ⊲ close to criticality the induced pressure p is small ⇒ can expand f around Xcr and perturbatively solve for the subsequent evolution: 2Ḣ + 3H 2 = 3H02 { 1 − 8πǫ f [−ǫ Xcr − ǫ(X − Xcr )] } ≃ 24πǫ2 H02 (X − Xcr ) f ′[−ǫ Xcr ] ⊲ resulting evolution equation is that of a damped oscillator: 2 R̈ + 2H Ṙ + (ω − Ḣ ) R ≃ 0 , ω ≡ ǫH0 q ′ 72π fcr 36 • asymptotic solution: lim a(t) ≃ t≫1 lim H(t) ≃ t≫1 lim Ḣ(t) ≃ t≫1 lim R(t) ≃ t≫1 K1 1 K3 + K2 t − sin(ωt + ϕ) + O( ) 2 6ω t 1 K1 cos(ωt + ϕ) 1 − + O( 2 ) t 6K2 ω t t K1 sin(ωt + ϕ) 1 + O( 2 ) 6K2 t t 1 K1 sin(ωt + ϕ) + O( 2 ) K2 t t • analytical predictions: ⊲ as inflation exit approaches, oscillations become significant 1 ⊲ inflation ended before criticality was reached: qcr = + 2 ⊲ at criticality the system is underdamped ⊲ during this evolution: H 2 < |Ḣ| < ω 2 37 • consistency with numerical analysis: criticality ⊲ numerical side: Criticality 0.030 0.025 0.020 0.015 0.010 0.005 155 000 160 000 165 000 170 000 175 000 t 180 000 -0.005 Determining the critical point 1 − 8πǫf [−ǫXi ] = 0 for the exponential model at the critical point (step i = 160942): X[160942] = −438.50 , q[160942] = 0.50 ⊲ analytical side: complete agreement 1 1 Xcr = − ln (1 + ) ∼ − 438.50 ǫ 8πǫ , 1 qcr = + 2 38 • consistency with numerical analysis: frequency ⊲ numerical side: R -8 6. ´ 10 4. ´ 10-8 2. ´ 10-8 200 000 250 000 300 000 t 350 000 -2. ´ 10-8 -4. ´ 10-8 Six oscillations have occured between steps i = 174291 and i = 342478 342478 − 174291 2π T = ∆t = 6 ω ⇒ 2.24 × 10−4 ω = ∆t ⊲ analytical side: complete agreement ǫδ q ǫδ h 1 i 21 2.25 × 10−4 ′ ω = 72π fcr = 72π (1 + ∼ ) ∆t ∆t 8πǫ ∆t 39 • scalar perturbations • tensor perturbations • long-range force • reheating 40 An Improved Model for Cosmology • post-inflationary evolution must consider constant ε spacetimes: Ḣ ε ≡ − 2 H ⇒ εdS = 0 , εrad = 2 , a(t) = ain [ 1 + ε Hin(t − 3 εmat = 2 1 tin) ] ε Hin H(t) = 1 + ε Hin(t − tin) R(t) = 6 (2 − ε) H 2(t) R(t) = +36ε (1 − ε)(2 − ε) H 4(t) R00(t) = −3 (1 − ε) H 2(t) 41 • simple model inadequacy ⊲ oscillations after the end of inflation are not a problem ⊲ average expansion a(t) ∼ t after the end of inflation ⇒ no reheating ⇒ a problem ⊲ assume, as usual, that energy flows from the gravitational to the matter sector • transitions ⊲ inflation → radiation, say at t = tr ⊲ radiation → matter, say at t = tm 42 • the total pressure ⊲ recall simple induced pressure ansatz: 2 p[g](x) = Λ f [−ǫ X](x) , X ≡ 1 R f [−ǫ X] = −ǫ X + O(ǫ2) ⊲ the total pressure equals: Λ + p[g](x) 8πG o Λ n 1 − 8πGΛ f [ − GΛ (Xcr + ∆X) ] = − 8πG Λ ′ ≃ − × (GΛ)2 fcr ∆X G ptot = − ⊲ source change after transition to matter domination: 3 4 ∆X(t) ≡ X(t) − Xcr = − ln[ 1 + Hm(t − tm) ] + O(1) 3 2 43 • the sign problem ⊲ since f is monotonically increasing and unbounded ⇒ ptot > 0 when X(t) < Xcr ≪ 0 observation implies the reverse ⊲ need f to be monotonically increasing and unbounded to cancel an arbitrary bare Λ • the magnitude problem ⊲ total pressure magnitude produced is unacceptably large: ptot ≃ pnow GΛ H0 2 ′ ′ × ∆X fcr ∆X ≃ 1086 × fcr Hnow 3 H2 ⊲ have used: pnow ≃ − 8πG now H0 ∼ 1013GeV , Hnow ∼ 10−33eV ′ ≥1 ⊲ for all models analyzed: fcr 44 • decreasing the magnitude ⊲ in our induced pressure ansatz Λ = 3H02 H0 can be 55 orders larger than Hnow ⊲ replace a Λ with a dynamical scalar S without disturbing the relaxation mechanism: p[g](x) = Λ2 f [−GΛ X](x) : −GΛ X = −GΛ 1 R −→ −G 1 (R × S ) = − GΛ S R× Λ 45 • changing the sign ⊲ the curvature scalar R is positive during both inflation and matter domination ⊲ simple quantities behaviour: R R R00 INFLATION RADIATION MATTER +12H 2 0 + 3H 2 0 0 −3H 2 3H 2 27 4 − H 2 3 2 + H 2 ⊲ simplest choice that does the job seems to be R00 46 • improved ansatz ⊲ the scalar S must be evaluated far back in the past ⇒ use integral curves χµ [g](x) of a timelike 4-velocity field V µ(x) ⇒ ... ⊲ we shall consider p[g](x) = Λ2 f [−GΛ Y ](x) where: 1 1 n Y [g](x) ≡ R(x) × Λ o µ ∗ ν ∗ ∗ − V (χ(τ , x)) V (χ(τ , x)) Rµν (χ(τ , x)) ⊲ for F RW spacetimes and for τ ∗[g](x) associated with, 9 of the time from xµ back to the IVS: e.g. 10 1 1 1 R(t) × R00( 10 t) ≡ Xcr + ∆Y YF RW [g](t) = − Λ ( V µ → δ µ0 ) 47 • late time acceleration ⊲ compute total pressure in the improved ansatz: ′ ′ 2 ptot ≃ − GΛ3 fcr ∆Y ≃ − 200 GΛ2 fcr Hm 1 ′ = ⊲ for the exponential model: fcr 8πGΛ ⊲ the pressure ratio is: t ≫ tm ⇒ 2 H 200 ptot m ′ × 8π(GΛ)2 × fcr ≃ pnow 3 Hnow 200 ′ ≃ 8π(GΛ)2 × fcr × 1010 3 2 ≃ × 1012 × GΛ 3 ⊲ physical values of GΛ = M 4 MP−4 l easily achieve equality 48 • prospects, etc 49