-
Ghost Condensation
and
Horizon Entropy
Ref. arXiv: 1602.06511 w/ S.Jazayeri, R.Saitou and
Y.Watanabe
Shinji Mukohyama
(YITP, Kyoto)
Based also on other works with
N. Arkani-Hamed, H.-C. Cheng, P. Creminelli, T. Furukawa, K.
Ichiki, K.
Izumi, M. Luty, N. Sugiyama, J. Thaler, T. Wiseman, M.
Zaldarriaga
-
Can we change gravity in IR?
Change Theory?Massive gravity Fierz-Pauli 1939DGP model
Dvali-Gabadadze-Porrati 2000
Change State?Higgs phase of gravityThe simplest: Ghost
condensationArkani-Hamed, Cheng, Luty and Mukohyama, JHEP
0405:074,2004.
-
Ghost condensation
Suppose
2)( LSMgrav LLL L
gravity
SMEnergy
Vacuum
Any coupling instability
Arkani-Hamed, Cheng, Luty and
Mukohyama, JHEP 0405:074,2004
-
In analogy with Higgs potential, can this be stabilized?
424 )()( ML ?Clearly are higher dim ops. Really we
should consider
4)(
()(~)( 22 QPPL )
Naively no sensible EFT.
NOT THE CASE.
There is a sensible EFT + derivative expansion.
(Shift symmetry is assumed.)
-
For simplicity
2)( PL
Eq of motion
0 PClearly any constant is a solution!
Suppose constant + timelike.
Go to frame where
ctc i 0,
Solution for any c!
P
in flat background.
-
Look at small fluctuations
ct
2222224 )()()(2)( cPcPccPxdS
So for 0)(2)( 222 cPccP
0)( 2 cP
NORMAL sign kinetic terms.
STABLE stable stable
-
In this language, we can have a good EFT because
the background sits at some value c.
doesn’t sample entire function .
Taylor expansion of around
makes perfect sense.
2)( P
Small fluctuations controlled by small # of
parameters at low energies.
2)( P 22)( c
-
So far, possible backgrounds are labeled by
continuous parameter c.
Situation changes in presence of gravity, in
expanding universe.2 2 2 2
4
( )
( )
ds dt a t dx
S d x gP g
T
~ PPg
E.O.M.
0][ 3 Pat 0P as a
0 or 0)( 2 P(unstable ghost
background)
-
P
2M
Gravity (Hubble friction)
drives you here!
In this language, we can have a good EFT because
the background sits at the special value.
doesn’t sample entire function .
Taylor expansion of around
makes perfect sense.
2)( P
Small fluctuations controlled by small # of
parameters at low energies.
2)( P 42)( M
Attractor
-
Look at small perturbations
2424444 ))(()()( MPMPMMPxd
No spatial kinetic term for !
Other terms ()(~ 2 QP )do contain spatial kinetic terms but at
least
22 )(
22
2
24 )(2
1
MxdS
Low
energy
22
2
23 )(2
1
MxdPositive definite Hamiltonian
STABLE!
2
42
M
k
tM 2
-
Of course there are higher terms in
effective theory.
2
2224 )(
2
1
MxdS
Because of k4, spatial fluctuations are less
suppressed than we are used to…
To analyze low-energy theory, we must do
proper analysis of scaling dimensions of
operators!
2
2
~)(
M
-
2
2223 )(
2
1
Mxdtd
4/1
2/1
1
r
dxrdx
dtrdt
rEE
Make
invariant
Leading operator in infrared
2
23
~)(
Mxdtd
scaling dimension 1/4. (Barely) irrelevant
Good low-E effective theory.
HEALTHY SECTOR IN ABSENCE
OF GRAVITY
-
What makes the ghost condensate special is that,
unlike other cosmological fluid, IT DOES NOT
DILUTE AS UNIVERSE EXPANDS.
Normally:
In deep future, flat or dS (maximal symmetry)
EMPTY OF FLUID
NOTE: Background breaks Lorentz invariance
spontaneously – prefered frame where is spatially
isotropic. No different from CMBR or any
cosmological fluid.
-
NOT HERE!2M even as universe expands.
)()(44 MPgMPT
Exactly that of c.c. !)( 4MP
dS solution!
0 0
Flat!
Present acceleration can be due to ghost condensate
even with =0 (in the symmetric phase). w=p/r=-1
exactly. BUT NOT A C.C.
PHYSICAL FLUID WITH PHYSICAL
EXCITATIONS.
P P
-
and timelike
Background metric is maximally
symmetric, either Minkowski or dS.
Systematic construction of
Low-energy effective theory
Backgrounds characterized by
0
Arkani-Hamed, Cheng, Luty and Mukohyama, JHEP 0405:074,2004
224 21
00 22eff EHL L M h K
M
22ij i j
ij i jK KM
-
Gauge choice: .),( txt
(Unitary gauge)
Residual symmetry: ),( xtxx
Write down most general action invariant under
this residual symmetry.
( Action for : undo unitary gauge!)
Start with flat background hg
h
Under residuali
ijjiijii hhh ,,0 0000
0
-
Action invariant under i
2
00h
2
0ih2 , ij ijK K K
0 0 01
2ij ij j i i jK h h h
OK
OK
24 21 2
00 2 2
ij
eff EH ijL L M h K K KM M
Action for
ij ij i jK K 00 00 02h h 0
224 21
00 22eff EHL L M h K
M
22ij i j
ij i jK KM
Beginning at quadratic order,
since we are assuming flat
space is good background.
-
2 23 2
2
1 ( )
2dtd x
M
4/1
2/1
1
r
dxrdx
dtrdt
rEE
Make
invariant
Leading nonlinear operator in infrared
2
23
~)(
Mxdtd
has scaling dimension 1/4. (Barely) irrelevant
Good low-E effective theory
Robust prediction
-
Higgs mechanism Ghost condensate
Order
parameter
Instability Tachyon Ghost
Condensate V’=0, V’’>0 P’=0, P’’>0
Broken
symmetry
Gauge symmetry Time translational
symmetry
Force to be
modified
Gauge force Gravity
New force
law
Yukawa type Newton+Oscillation
2 2 2
2( )P
(| |)V
Arkani-Hamed, Cheng, Luty and Mukohyama 2004
-
Bounds on symmetry breaking scale M
M
Jeans Instability
(sun)
100GeV 1TeV
Supernova time-delay
ruled out
ruled out
ruled out
0
allowed
Twinkling from Lensing
(CMB)
So far, there is no conflict with experiments
and observations if M < 100GeV.
Arkani-Hamed, Cheng, Luty and Mukohyama and Wiseman, JHEP
0701:036,2007
-
Ghost condensation• Ghost condensation is the simplest Higgs
phase of gravity.
• The low-E EFT is determined by the symmetry breaking pattern.
No ghost in the EFT.
• Gravity is modified in IR.
• Consistent with experiments and observations if M <
100GeV.
-
Holography and GSL• Do holographic dual descriptions always
exist?
PROBABLY NO. e.g.) A de Sitter space is only meta-
stable and a unitary holographic dual is not known.
• How about ghost condensate?
• Let’s look for violation of GSL in ghost condensate,
since violation of GSL would indicate absence of
holographic dual. (GSL is expected to be dual to
ordinary 2nd law.)
• Three proposals: (i) semi-classical heat flow; (ii)
analogue of Penrose process; (iii) negative energy.
• The generalized 2nd law holds in the presence of
ghost condensate. (Mukohyama 2009, 2010)
-
Ghost inflation and de Sitter
entropy bound
• Black holes & cosmology in gravity theories are
as important as Hydrogen atoms & blackbody
radiation in quantum mechanics
• Provide non-trivial tests for theories of gravity
e.g. black-hole entropy in string theory
• Does the theory of ghost condensation pass
those tests?
• Ghost condensation is known to be consistent
with BH thermodynamics (Mukohyama 2009, 2010)
• How about de Sitter thermodynamics?
S.Jazayeri, S.Mukohyama, R.Saitou, Y.Watanabe 2016
-
de Sitter thermodynamics
• de Sitter (dS) spacetime is one of the three
spacetimes with maximal symmetry
• dS horizon has temperature TH = H/(2)
• In quantum gravity, a dS space is probably
unstable (e.g. KKLT, Susskind, …). So, let’s
consider a dS space as a part of inflation
• Friedmann equation
1st law with entropy S = A/(4GN) = /(GNH2)
(This is in contrast with analogue gravity systems.)
-
de Sitter entropy bound
• Slow roll inflation (non-eternal)
Arkani-Hamed, et.al. 2007
for non-eternal inflation
-
de Sitter entropy bound
• Eternal inflation
• Fluctuation generated during eternal epoch
would collapse to form BH unobservable!
• This bound holds for a large class of models
of inflation
• Does ghost inflation satisfy the bound?
Arkani-Hamed, et.al. 2007
-
!0
NOT SLOW ROLL
Scale-invariant perturbationscf. tilted ghost inflation,
Senatore (2004)
r
r
H~ 4/1)/( MHM ~
4/5
M
H~
~ 2M
scaling dim of
[compare ]Pl
H
M
Similar to
hybrid inflation but
Arkani-Hamed, Creminelli, Mukohyama and Zaldarriaga, JHEP
0404:001,2004
Ghost inflation
V
-
Prediction of Large non-Gauss.
Leading non-linear interaction
non-G of ~H
M
1/4 scaling dim of op.
1/5
r
r
~
2
2
( )
M
2 23 2
2
1 ( )
2dtd x
M
[Really “0.1” ~ 10-2. VISIBLE.
In usual inflation, non-G ~ ~ 10-5 too small.]
5/1/ rr rr /
fNL ~ 82 -4/5, equilateral type
Planck 2015 constraint (equilateral type)
fNL = -4±43 (68% CL statistical) 4/50.6 0.5
-
de Sitter entropy bound
• Eternal inflation
• Fluctuation generated during eternal epoch
would collapse to form BH unobservable!
• This bound holds for a large class of models
of inflation
• Does ghost inflation satisfy the bound?
The answer appears to be “no” since Ntot can
be arbitrarily large. Swampland?
Arkani-Hamed, et.al. 2007
-
Lower bound on ?
• Tiny prevents earlier inflationary modes
from being observed.
• Nobs ~ ln(kmax/kmin) S = /(GNH2)
• In our universe, W=O(1) and thus the
bound is well satisfied.
with
S.Jazayeri, S.Mukohyama, R.Saitou, Y.Watanabe 2016
-
Summary• Ghost condensation is the simplest Higgs
phase of gravity.
• The low-E EFT is determined by the symmetry breaking pattern.
No ghost in the EFT.
• Gravity is modified in IR.
• Consistent with experiments and observations if M <
100GeV.
• It appears easy but is actually difficult to violate
the generalized 2nd law by ghost condensate. (Mukohyama 2009,
2010)
• Ghost inflation predicts large non-Gaussianitythat can be
tested.
• de Sitter entropy bound appears to be violated but is actually
satisfied by ghost inflation.
-
Higgs mechanism Ghost condensate
Order
parameter
Instability Tachyon Ghost
Condensate V’=0, V’’>0 P’=0, P’’>0
Broken
symmetry
Gauge symmetry Time translational
symmetry
Force to be
modified
Gauge force Gravity
New force
law
Yukawa type Newton+Oscillation
2 2 2
2( )P
(| |)V
Thank you
very much!
Arkani-Hamed, Cheng, Luty and Mukohyama 2004
-
Cosmological Page time
• Hawking rad from BH Srad = Sent increases
but SBH (≧ Sent) decreases semi-classical description should
break down @ Page time,
when SBH ~ half of SBH,init
• After inflation, we expect to see O(1) deviation
from semi-classical description @ Page time,
when Nobs ~ Send
• For example, if decays at a=adecay then
S.Jazayeri, S.Mukohyama, R.Saitou, Y.Watanabe 2016