Free Gravitons Break de Sitter Invariance ( arXiv:0907.4930 , - - PowerPoint PPT Presentation

Free Gravitons Break de Sitter Invariance (arXiv:0907.4930, 1002.4037)

• S. P. Miao (Utrecht)
• N. C. Tsamis (Crete)
• R. P. Woodard (Florida)

Spacetime Exp. Strengthens QFT

Why?

Loops classical physics of virtuals Expansion holds virtuals apart longer

Maximum Effect for

Inflation M=0 No conformal invariance (classically)

Two Particles

MMC scalars gravitons

Primordial Inflation was nearly de Sitter with small GH

²

ds2 = 9dt2 + a²(t) dx.dx

H(t) = á/a & ε(t) = 9Ḣ/H²

For single9scalar inflation with k = H(tk)a(tk)

@²

R(k) ≈ GH²(tk)/πε(tk) & @ ² h(k) ≈ 16GH²(tk)/π

WMAP data for k = .002/Mpc

@²

R = 2.441 x 1099

& r = @²h/@²

R < 0.22

Hence

ε ≈ r/16 < 0.014 (even smaller before tk!) GH² ≈ π/16 x r x @²

R < 10910

MMC Scalar Models

1.

λφ4 (Brunier, Kahya, Onemli)

• M²(x;x') @u(t,k) & <Tν>
• Growing scalar mass & pos. vac. Energy

2.

SQED (Kahya, Prokopec, Tornkvist, Tsamis)

• M²(x;x') @u & [Πν](x;x') @ε
• <φ*φ>, <(Dφ)*Dνφ>, <FνFρσ> & <Tν>
• Growing photon mass & neg. vac. Energy

3.

Yukawa (Duffy, Prokopec, Miao)

• M²(x;x') @u, Σ(x;x') @u & <φψ†γ0ψ>
• Growing fermion mass & neg. vac. Energy

Quantum Gravity Models

1.

QG + Dirac (Miao)

• [iΣj](x;x') @u(t,k)
• Growing fermion field strength

2.

QG + MMC Scalar (Kahya,Park)

• M²(x;x') @u(t,k)
• [νΣρσ](x;x’) @εν & force of gravity
• Possible tilt in Power Spectrum

3.

QG (Tsamis,Mora)

• [νΣρσ](x;x') & <hZν>
• Consistent with relaxation of Λ

Enhanced QFT as IR Logs

What? factors of ln(a) = Ht

Eg ρ = λ(H/2π)4 x ⅛ ln²(a) + O(λ)

Through propagators

i@(x;x’) = (dS inv) + H²/8π² ln(aa’) i[ij@kℓ](x;x’) = [2δi(kδℓ)j – 2δijδkℓ] x same

Also from vertex integrations

∫t dt’ 1 = t = ln(a)/H NB occur even if no dS breaking in i@!

Math Guys IR logs

Reluctantly accept in i@(x;x’)

But struggle to avoid consequences

But deny in i[ij@kℓ](x;x’) NB vertex integrations still break dS

∫d4x’ √9g(x’) θ(x09x’0) θ[9ℓ2(x;x’)] = ∫t dt’ a3(t’) x 4π/3H3 (1/a’ – 1/a)3 = 4π/3H4 [ln(a) + O(1)]

But simplest IR logs come from props

dS Inv Eqns Don’t Always Have Invariant Solutions

MMC φ: □i@(x;x’) = iδ4(x9x’)/√9g

Allen & Folacci, PRD35 (1987) 3371.

ds2 = 9dt2 + a2(t) dx—dx

a(t) = eHt

i@(x;x’) = ∫d3k/(2π)3 ek—(x9x’)

x[θ(t9t’)u(t,k)u*(t’,k) + θ(t’9t)u*(t,k)u(t’,k)]

u(t,k) = H/(2 k3)½ [1 – ik/Ha] Exp[ik/Ha] IR problem: uu* ~ H2/2k3

What about i[ν@ρσ](x;x’)? Cosmologists: not invariant

Grishchuk (Sov. Phys. JETP 40 (1975) 409)

Gravitons have same u(t,k) as MMC φ

This IS observable!

@²h = k3/2π ∫d3x eik—x <hij(t,x) hij(t,0)> = k3/2π × 32πG × 2 × |u(t,k)|2 = 16/π GH2 (a.k.a. SCALE INVARIANCE)

Kleepe (PLB 317B (1993) 305)

• Comp. trans. does not restore invariance

What about i[ν@ρσ](x;x’)? Math Physicists: Yes it is!

Add α(Dνhν + βDhνν)2 Solve in Euclidean space & continue Ok except few “singular” choices of α and β

Burden of my Talk: Math Physicists are wrong

Obstacle to adding gauge fixing term Obstacle to analytic continuation Origin of “singular” gauges

“Exact” vs “Average” Gauges

Illustrate with EM in flat space

Exact: ∂iAi= 0 (Coulomb) Average: L L 9 ½ (∂A)2 (Feynman)

Derive Average from Exact

Start in canonical functional formalism

∫ [dET] [dAT] eiSfixed

• S. Coleman, Erice 1973

Coleman’s Seven Steps

1.

Integrate out ET

2.

Use ∫[dAT] = ∫ [dA] δ[∂iAi] √det[∂i∂i]

3.

Undo A0 constraint

4.

Write integrand as invariant

5.

δ[∂A] w field dependent gauge trans

6.

δ[∂A–f(x)] w C9number gauge trans

7.

Multiply by ∫[df] Exp[9½i∫f2]

Obstacle on T3 × R

Invariant: ∂iFi0 = J0 Q = 0 Feynman: [9∂t2+∂i∂i]A0 = J0 Q ≠ 0 ok Problems at Coleman’s steps 2 & 3

No 09modes for δ[∂iAi] and A0 Hence no 09mode for gauge fixing term

Same Obstacle on de Sitter

IR ∞ of φφ* self9energy (gr9qc/0508015)

Analytic Continuation Sees Only Logarithmic IR Divergences

[□9M2]i@(x;x’) = iδ4(x9x’)/√9g

i@(x;x’) =∫d3k/(2π)3 eik—(x9x’) x[θ(t9t’)u(t,k)u*(t’,k)+θ(t’9t)u*(t,k)u(t’,k)]

u(t,k) = [π/(4Ha3)]½ Hν(1)[k/Ha]

• ν = [9/4 – M2/H2]½

uu* ~ k92ν [1 + O(k2)]

IR ∞’s for 2ν ≥ 3 M2 ≤ 0

But only logarithmic for M2 = 9N(3+N) H2

IR ∞’s Signal Wrong Physics

DON’T subtract them, fix the physics

Exclusive Inclusive in flat QED, QCD & QG

Physical Problem:

Can’t enforce Bunch9Davies for k < Hainitial

Standard Fixes

Vilenkin (NPB:226,527,1983)

Change Bunch9Davies for k < Hainitial

NCT and RPW (CQG:11,2969,1994)

Keep Bunch9Davies on T3xR with no k < Hainitial

How It Works In Practice

i@naïve(x;x’) = ∫d3k/(2π)3 ek—(x9x’)

x[θ(t9t’)u(t,k)u*(t’,k) + θ(t’9t)u*(t,k)u(t’,k)]

Just cut off IR ∫d3k/(2π)3 ek—(x9x’) θ(k9k0) x (Same) Resolves old problem of Ford & Parker (1977)

Scalar9driven FRW

Iliopoulos,Tomaras,NCT,RPW (NPB:534,419,1998)

MMC scalars on FRW with constant ε

Janssen,SPM,Prokopec,RPW (CQG:25,245013,2008)

de Sitter with M

² S = 9N(N+3)H ² and M ² V = 9(N+1)(N+3)H ²

SPM,NCT,PRW (JMP:51,072503,2010)

Why Not Use the Subtracted Solutions for Power Law IR ∞’s?

[□9M²] i@(x;x’) = iδ4(x9x’)/√9g

but i@(x;x’) ≠ <ψ|T[φ(x)φ(x’) ]|ψ>

Eg i@(x;x’) i Gret(x;x’)

Vanishes for x’=x vs

<ψ|φ²|ψ> ≠ 0

SHO: 9m[(d/dt)²+ω²] i@(t;t’) = iδ(t9t’)

9i sin[ω|t9t’|]/2mω + α cos(ωt)cos(ωt’) + β sin[ω(t+t’)] + γ sin(ωt)sin(ωt’)

Solves for any α, β & γ, but QM requires

α + γ ≥ 1/2mω and αγ ≥ ¼β²

• Math: Reflection Positivity fails

Exact de Donder Gauge

mtrcν = gν+ hν with Dνhν = ½Dhν

ν

Dν

αβ i[αβ@ρσ](x;x’) ≠ g(ρgσ)ν iδ4(x9x’)

Not consistent with gauge condition

rhs = [g(ρgσ)ν 9 ½ gνgρσ] iδ4(x9x’)

+ 2 √9g Sym{DDρ’ i[ν@σ](x;x’)}

[□+3H2] i[@ν](x;x’) = gν iδ4(x9x’)/√9g

Corresponds to MV2 = 96H2 IR ∞’s MV2≤ 0, Log MV2 = 9(N+2)(N+3)H2

Scalar Structure Functions in i[ν@ρσ](x;x’)

Spin 0 Part: Pν(x) Pρσ(x’) F0(x;x’)

• Pν = DDν + ½ [□+6H²] gν

¾[□+4H²][□+6H²]² F0(x;x’) = iδ4(x9x’)/√9g M² = 94H² is Log ∞, M² = 96H² is power ∞

Spin 2 Part: Pνβ δ(x) Pρσκθ(x’) [TβκTδθF2]

Cαβγδ = Pν

αβγδ hν + O(h2)

Pν

βδ = 91/2H² Pν αβγδ DαDγ

Tβκ = 91/2H² ∂²y/∂xβ∂x’κ

y= aa’H²(x9x’)²

□³[□92H²]² F2 = 64 H4 iδ4(x9x’)

Conclusion: Graviton Propagator Is NOT de Sitter Invariant

Plausible arguments each way

Pro: Inv. solns w some gauge fixing terms Con: Dynamically same as MMC scalars + IR

divergences in some gauges

Long controversy resolved

Obstacle to adding gauge fixing terms Obstacle to Euclidean continuation

De Donder projection operator not invariant