2D is Fire Free
Ahmed Almheiri UCSB
w/ James Sully; 2013
Black Holes in String Theory Workshop UMich, 10/16/13 Mo va on
• Need a controllable se ng to a empt to answer long-standing QG ques ons: – Informa on Loss? – remnants? – Firewall? • 2D provides such a se ng: It has completely solvable models, – CGHS [Callan, Giddings, Harvey, Strominger] – RST [Russo, Susskind, Thorlacius] Mo va on
• Ashtekar et al simulate CGHS in the Mean Field Approxima on (Large-N semiclassical analysis) and find: – Unitary Evapora on* – No Firewall [Ashtekar, Pretorius, Ramazanola, Taveras, Varadarajan]
• This looks like it contradicts the Firewall proposal. Outline
• The Model – Classical CGHS – CGHS in MFA • Entanglement Entropy of 1+1 CFTs and Moving Mirrors
• CGHS Remnants
• Upli
• Discussion The Model – Classical CGHS
• The CGHS model is given by the ac on
1 2 2φ a 2 S = d x√ge− R +4 φ φ +4κ G G ∇ ∇a coupled to N scalar fields via 1 S = d2x√g af i f i M −2 ∇ ∇a [Callan et al] • It describes the near horizon physics of magne cally charged extremal dilatonic black holes in 4 and 5 dimensions [Giddings, Strominger] The Model – Classical CGHS
• Solu on fixed by scalar field
+ κ(z z−) Θ(z±)=e − + + z z i 2 + + G + κz + κz ∂f+ Φ(z±)=Θ(z±) dz e dz e− − 2 ∂z + i −∞ −∞ ab 1 ab 2φ Where g = ΦΘ− η , Φ = e− • Form a black hole by sending in a scalar shockwave
1 i 2 MADM + (∂ + f ) = δ(z ) 2 z + N The Model – Classical CGHS
• Singularity is at Φ =0
+ • and are R L− Icomplete I w.r.t. the metric gab
+ • Past of covers only IR part of − IL y− = ln(z− z−) − h − = Informa on Loss! ⇒ • Next: Include backreac on of Hawking Radia on The Model – CGHS in MFA
• Do so by working in a Mean Field Approxima on: – Replace geometric fields by expecta on values – Include scalar field quantum contribu ons
– Requires large N [callan et al, Ashtekar et al] [Christensen, Fullings] • Einstein’s equa ons are sourced by the one-loop trace anomaly G = T µν µν The Model – CGHS in MFA
+ • The MFA equa ons give a balance law at from R which one defines a Bondi Mass and Flux: I
dM ATV Bondi = F ATV dy− −
2 NG d dy− where F ATV = ln 48 dy dz − − + and are the affine parameters on and z− y− − IL IR respec vely. The metric is flat in these coordinates. The Model – CGHS in MFA
• Comments on the model: NG – Quantum corrected singularity Φ = 12 – Scaling Symmetry: ATV ATV ATV ATV (M ADM,F ,M ,N) λ(MADM,F ,M ,N) Bondi → Bondi • Physical parameters are ra os, eg: MADM M ∗ = N – A black hole is considered Planckian if it evaporates in a Planck me when assuming the standard hawking flux M ∗ GMPl The Model – CGHS in MFA
• Again, form a black hole by sending in a coherent ˆi state of scalar fields, , peaked at the classical f+ o profile f+ s.t.:
1 o 2 MADM + (∂ + f ) = δ(z ) 2 z + N
• The right movers, , are ini ally in their ground fˆi state, so the classical profiles are − f i =0 −
The Model – Results The Model – Results
0.9
0.85
0.8 * m
0.75
0.7 w=0 fitted curve
0.65 [Ashtekar et al] 0 5 10 15 M* • Universal small Bondi mass at the last ray
m∗ = M ∗ Last ray 0.86GMPl Bondi| ∼ The Model – Results
• Small Bondi mass at the last ray, m∗ The Model – Results
6 0.1
0 5 ï0.1
) ï0.2 4 6 ) ï ï0.3 /dz
3 ))/d(ln ï
ï ï0.4
/dz ï0.5 ( dy 2 ï 10 ï0.6 log 1 ï0.7 d(ln(dy
ï0.8 0 ï0.9
ï1 ï1 [Ashtekar et al] ï10 ï5 0 ï10 ï5 0 log log 106 106 • is finite at the last ray, and goes as y− y− = kz− dy 1 • Intermediate region: (like classical case)− = dz −κ(z− z ) − h − − + • is not complete w.r.t. gab IR The Model – Results
• Small Bondi mass at the last ray, m∗
+ • is not complete IR w.r.t. gab y− = kz− The Model – Results
8 ï5 0 5
6
ï8 ï10 4
2
ï7 ï10
ï sing 0 1/5 z R ï
ï ï2 z ï6 ï10 ï4
1eï5 ï6 1eï6 ï5 ï10 1eï8 ï8 2 4 8 16 8 12 16 20 [Ashtekar et al] z+ z+ • No thunderbolt singularity. The last ray is regular. • No singularity in the horizon region.
The Model – Results
• Small Bondi mass at the last ray, m∗
+ • is not complete IR w.r.t. gab y− = kz−
• No Thunderbolt Singularity
• Horizon region is regular = No Firewalls ⇒ The Model – Results
• Comments: + – might be extendable s.t. it becomes unitarily IR equivalent to L− – Assuming the singularity is alleviated by quantum I correc ons. • So we have unitary black hole evapora on with semiclassical physics in the horizon region with no drama---!....Relax • Will show that this is a Remnant scenario. The small bondi mass object is entangled with the hawking radia on. The Model – Analysis
• Compute entropy of an interval containing the Hawking radia on
• The radia on is due to having two different asympto c killing vectors with which to define posi ve energy
• This is now exactly like compu ng entanglement entropy of radia on from moving mirrors [Holzhey, Larsen, Wilczek] The Model – Analysis
=
• We obtain from f(z−) Ashtekar et al’s plots. The Model – Analysis
6 0.1
0 5 ï0.1
) ï0.2 4 6 ) ï ï0.3 /dz
3 ))/d(ln ï
ï ï0.4
/dz ï0.5 ( dy 2 ï 10 ï0.6 log 1 ï0.7 d(ln(dy
ï0.8 0 ï0.9
ï1 ï1 [Ashtekar et al] ï10 ï5 0 ï10 ï5 0 log log 106 106 • Mirror Trajectory: – Ini al y− = z− dy 1 – Intermediate − = dz −κ(z− z ) − h − − – Near the last ray and past y− = kz− EE in 1+1 CFT
• We resort to 1+1 CFT EE techniques. • EE of an interval in the vacuum is given by
c +¯c Σ Sent = ln 6 √ 1 2
ε2 Σ ε1
[Callan, Wilczek, Srednicki] EE in 1+1 CFT
• The mirror transforms the vacuum state to an excited state
• EE for excited states: – These states relate to the ground state by a conformal transforma on f(x) – This transforma on preserves the no on of inside and outside And so, Exc c +¯c f(x2) f(x1) Sent = ln − 6 f (x2)2f (x1)1
[Holzhey, Larsen, Wilczek] EE in 1+1 CFT
• We are interested in compu ng the excess entropy on top of the vacuum. – This gives the entropy of the new excita ons produced by the mirror (or black hole). • This defines a renormalized EE, c +¯c (f(x ) f(x ))2 SRen = ln 2 − 1 ent 12 (x x )2f (x )f (x ) 2 − 1 2 1 – By subtrac ng off the vacuum piece we got rid of the
dependence on the UV regulator. [Holzhey, Larsen, Wilczek] EE in 1+1 CFT and Moving Mirrors
• Let’s apply the formula for the mirror trajectory that models the black hole evapora on 1 • Mirror path given by y − = f − (z−) 1 – Far past: f − (z−)=z− 1 – Near the last ray and beyond: f − (z−)=kz− • We wish to compute the entropy of an interval with y in the far past, and near the last ray. Then: 1− y2−
f(y2−) f(y1−) f (y1−) 1 − 1 → y− y− → 2 − 1 as y− 1 →−∞ EE in 1+1 CFT and Moving Mirrors
• The Ren. Entropy simplifies to Ren c S = ln(f (y−)) where f (y−)=1/k ent −12 2 2
• The entropy is independent of and thus all the excess y1− entanglement is with modes past the last ray • This excess comes from pulling modes below the cut-off EE in 1+1 CFT and CGHS Remnants
• We can do be er than just plugging in . We can k Ren obtain the scaling of with SEnt MADM • From the formulas of the Bondi Mass and Flux we have ATV 2 dMBondi ATV NG d = F = ln(f (y)) − dy 48 −dy − − NG d κ ln(f (y)) ∼ 48 −dy − dy− 1 Used the intermediate region approx: = dz− −κ(zh− z−) • Now integrate from to the last ray: − −∞ ATV NG M + M = κ ln(f (y−)) + ln(f (y−)) − Bondi|last ray ADM 48 − 2 1 EE in 1+1 CFT and CGHS Remnants
• We can do be er than just plugging in . We can k Ren obtain the scaling of with SEnt MADM • From the formulas of the Bondi Mass and Flux we have ATV 2 dMBondi ATV NG d = F = ln(f (y)) − dy 48 −dy − − NG d κ ln(f (y)) ∼ 48 −dy − dy− 1 Used the intermediate region approx: = dz− −κ(zh− z−) • Now integrate from to the last ray: − −∞ ATV NG M + M = κ ln(f (y−)) + ln(f (y−)) − Bondi|last ray ADM 48 − 2 1 Small 0 EE in 1+1 CFT and CGHS Remnants
• To leading order we find
Ren 4 SEnt = MADM κG • We found very large entanglement between the Hawking radia on and the region past the last ray, where the Bondi Mass is very small (Planckian). This is a Remnant!
• Long lived. The evapora on is energe cally constrained to be 2 slower than MADM Upli
• The solu on can be upli ed to describe the process of throwing in excita ons inside an extremal dilatonic black holes in 4d or 5d • Before the pulse we had an infinite throat with a linear dilaton • The infalling pulse pushes the black hole away from extremality • The black hole then proceeds to evaporate away to extremality Upli
• This suggests an old version of the remnant scenario where the remnants are given by these infinite throats – cornucopions [Banks, Giddings, Loughlin, Strominger] • These are ruled out in higher dimensions by the usual arguments: AdS/CFT and infinite pair produc on [Giddings] • But didn’t we just show that remnants are indeed what you get? • No: Reduc on and Quan za on do not commute. Discussion
• Why didn’t we get anything new? • Gravity in 2d is a local renormalizable field theory • We do not expect a complementary descrip on to emerge for such theories.