ENTROPY OF SPACETIME AND GRAVITY

T. Padmanabhan

(IUCAA, Pune, India)

IAGRG 07, FEB 2007

IAGRG, Feb 07
 PLAN OF THE TALK

• GRAVITY AND THERMODYNAMICS: HINTS OF A

DEEPER CONNECTION

• THE SAKHAROV PARADIGM

• GRAVITY AS AN EMERGENT PHENOMENON

• ENTROPY OF NULL SURFACES AND SPACETIME

DYNAMICS

• HORIZON ENTROPY

• PERSPECTIVE AND CONCLUSIONS

IAGRG, Feb 07
THERMODYNAMICS OF THE HORIZON

IAGRG, Feb 07
 THERMODYNAMICS OF THE HORIZON

• Null surfaces =⇒ Causal Horizons for a class of observers

=⇒ observer dependent temperature and entropy.

IAGRG, Feb 07
 THERMODYNAMICS OF THE HORIZON

• Null surfaces =⇒ Causal Horizons for a class of observers

=⇒ observer dependent temperature and entropy.

• All this is very generic.

IAGRG, Feb 07
 THERMODYNAMICS OF THE HORIZON

• Null surfaces =⇒ Causal Horizons for a class of observers

=⇒ observer dependent temperature and entropy.

• All this is very generic.

• Uses only the kinematics — special relativity and

principle of equivalence — of gravity [“Spacetime tells
matter how to move”]

IAGRG, Feb 07
 THERMODYNAMICS OF THE HORIZON

• Null surfaces =⇒ Causal Horizons for a class of observers

=⇒ observer dependent temperature and entropy.

• All this is very generic.

• Uses only the kinematics — special relativity and

principle of equivalence — of gravity [“Spacetime tells
matter how to move”]
Question:
• Is there a deeper connection between these results and
spacetime dynamics [“Matter tells spacetime how to
curve”]? What is this telling us?

IAGRG, Feb 07
 HINT OF A DEEPER CONNECTION
Spherically symmetric spacetimes with horizon(s)

IAGRG, Feb 07
 HINT OF A DEEPER CONNECTION
Spherically symmetric spacetimes with horizon(s)

• Metric:

ds2 = f (r)dt2 − f (r)−1dr2 − r 2(dθ 2 + sin2 θdφ2 )

≡ f (r)dt2 − f (r)−1dr2 − dL2⊥

• Source:
(r) µ(r)
Ttt = Trr = ; Tθθ = Tφφ =
8π 8π
• Einstein’s equations:
1 f0
2
(1 − f ) − = ; ∇2 f = −2µ
r r

IAGRG, Feb 07
 HINT OF A DEEPER CONNECTION
Spherically symmetric spacetimes with horizon(s)

• Metric:

ds2 = f (r)dt2 − f (r)−1dr2 − r 2(dθ 2 + sin2 θdφ2 )

≡ f (r)dt2 − f (r)−1dr2 − dL2⊥

• Source:
(r) µ(r)
Ttt = Trr = ; Tθθ = Tφφ =
8π 8π
• Einstein’s equations:
1 f0
2
(1 − f ) − = ; ∇2 f = −2µ
r r
• Consider the case with horizon at r = a; that is, f = 0 at r = a with
f 0 (a) = B > 0.

IAGRG, Feb 07
 HINT OF A DEEPER CONNECTION
Spherically symmetric spacetimes with horizon(s)

• Metric:

ds2 = f (r)dt2 − f (r)−1dr2 − r 2(dθ 2 + sin2 θdφ2 )

≡ f (r)dt2 − f (r)−1dr2 − dL2⊥

• Source:
(r) µ(r)
Ttt = Trr = ; Tθθ = Tφφ =
8π 8π
• Einstein’s equations:
1 f0
2
(1 − f ) − = ; ∇2 f = −2µ
r r
• Consider the case with horizon at r = a; that is, f = 0 at r = a with
f 0 (a) = B > 0.

• Temperature of horizon: kB T = ~cB/4π.

IAGRG, Feb 07
AN INTERPRETATION OF EINSTEIN’S EQUATIONS
T.P., CQG 19, 5387 (2002). [gr-qc/0204019]

IAGRG, Feb 07
 AN INTERPRETATION OF EINSTEIN’S EQUATIONS
T.P., CQG 19, 5387 (2002). [gr-qc/0204019]

• At r = a, f (a) = 0, f 0 (a) ≡ B. Einstein’s equation gives:

 
c4 1 1
Ba − = −4πTrr a2
G 2 2

IAGRG, Feb 07
 AN INTERPRETATION OF EINSTEIN’S EQUATIONS
T.P., CQG 19, 5387 (2002). [gr-qc/0204019]

• At r = a, f (a) = 0, f 0 (a) ≡ B. Einstein’s equation gives:

 
c4 1 1
Ba − = −4πTrr a2
G 2 2

• Multiply da to write:

3
  4
 
~cB c 1 1 c da 4π 3
d 4πa2 − = −Trr d a
4π G~ 4 2 G 3

IAGRG, Feb 07
 AN INTERPRETATION OF EINSTEIN’S EQUATIONS
T.P., CQG 19, 5387 (2002). [gr-qc/0204019]

• At r = a, f (a) = 0, f 0 (a) ≡ B. Einstein’s equation gives:

 
c4 1 1
Ba − = −4πTrr a2
G 2 2

• Multiply da to write:

3
  4
 
~cB c 1 1 c da 4π 3
d 4πa2 − = −Trr d a
4π G~ 4 2 G 3
| {z }
P dV

IAGRG, Feb 07
 AN INTERPRETATION OF EINSTEIN’S EQUATIONS
T.P., CQG 19, 5387 (2002). [gr-qc/0204019]

• At r = a, f (a) = 0, f 0 (a) ≡ B. Einstein’s equation gives:

 
c4 1 1
Ba − = −4πTrr a2
G 2 2

• Multiply da to write:

3
  4
 
~cB c 1 1 c da 4π 3
d 4πa2 − = −Trr d a
|4π
{z } G~ 4 2 G
| {z
3
}
kB T P dV

IAGRG, Feb 07
 AN INTERPRETATION OF EINSTEIN’S EQUATIONS
T.P., CQG 19, 5387 (2002). [gr-qc/0204019]

• At r = a, f (a) = 0, f 0 (a) ≡ B. Einstein’s equation gives:

 
c4 1 1
Ba − = −4πTrr a2
G 2 2

• Multiply da to write:

3
  4
 
~cB c 1 1 c da 4π 3
d 4πa2 − = −Trr d a
|4π
{z } |G~ 4
{z
2 G
} | {z } | {z
3
}
kB T k−1dS −dE P dV
B

IAGRG, Feb 07
 AN INTERPRETATION OF EINSTEIN’S EQUATIONS
T.P., CQG 19, 5387 (2002). [gr-qc/0204019]

• At r = a, f (a) = 0, f 0 (a) ≡ B. Einstein’s equation gives:

 
c4 1 1
Ba − = −4πTrr a2
G 2 2

• Multiply da to write:

3
  4
 
~cB c 1 1 c da 4π 3
d 4πa2 − = −Trr d a
|4π
{z } |G~ 4
{z
2 G
} | {z } | {z
3
}
kB T k−1dS −dE P dV
B

• Read off (L2P ≡ G~/c3 ):

 1/2
−1 1 2 1 AH c4 c4 AH
kB S= (4πa ) = ; E= a=
4L2P 4 L2P 2G G 16π

IAGRG, Feb 07
 AN INTERPRETATION OF EINSTEIN’S EQUATIONS
T.P., CQG 19, 5387 (2002). [gr-qc/0204019]

• At r = a, f (a) = 0, f 0 (a) ≡ B. Einstein’s equation gives:

 
c4 1 1
Ba − = −4πTrr a2
G 2 2

• Multiply da to write:

3
  4
 
~cB c 1 1 c da 4π 3
d 4πa2 − = −Trr d a
|4π
{z } |G~ 4
{z
2 G
} | {z } | {z
3
}
kB T k−1dS −dE P dV
B

• Read off (L2P ≡ G~/c3 ):

 1/2
−1 1 2 1 AH c4 c4 AH
kB S= (4πa ) = ; E= a=
4L2P 4 L2P 2G G 16π

• In normal units, there is no ~ in T dS! (Compare: No kB in thermodynamics)

IAGRG, Feb 07
 AN INTERPRETATION OF EINSTEIN’S EQUATIONS
T.P., CQG 19, 5387 (2002). [gr-qc/0204019]

• At r = a, f (a) = 0, f 0 (a) ≡ B. Einstein’s equation gives:

 
c4 1 1
Ba − = −4πTrr a2
G 2 2

• Multiply da to write:

3
  4
 
~cB c 1 1 c da 4π 3
d 4πa2 − = −Trr d a
|4π
{z } |G~ 4
{z
2 G
} | {z } | {z
3
}
kB T k−1dS −dE P dV
B

• Read off (L2P ≡ G~/c3 ):

 1/2
−1 1 2 1 AH c4 c4 AH
kB S= (4πa ) = ; E= a=
4L2P 4 L2P 2G G 16π

• In normal units, there is no ~ in T dS! (Compare: No kB in thermodynamics)

• Works for time-dependent horizons, Kerr, Lanczos-Lovelock theory ......; close to

Membrane Paradigm in spirit

IAGRG, Feb 07
 SAKHAROV PARADIGM
Gravity as an Emergent Phenomenon

IAGRG, Feb 07
 SAKHAROV PARADIGM
Gravity as an Emergent Phenomenon

SOLIDS SPACETIME

Mechanics; Elasticity (ρ, v ...) Einstein’s Theory (gab ...)

Statistical Mechanics Statistical mechanics

of atoms/molecules of “atoms of spacetime”

IAGRG, Feb 07
 SAKHAROV PARADIGM
Gravity as an Emergent Phenomenon

SOLIDS SPACETIME

Mechanics; Elasticity (ρ, v ...) Einstein’s Theory (gab ...)

Thermodynamics of solids

Statistical Mechanics Statistical mechanics

of atoms/molecules of “atoms of spacetime”

IAGRG, Feb 07
 SAKHAROV PARADIGM
Gravity as an Emergent Phenomenon

SOLIDS SPACETIME

Mechanics; Elasticity (ρ, v ...) Einstein’s Theory (gab ...)

Thermodynamics of solids Thermodynamics of spacetime

Statistical Mechanics Statistical mechanics

of atoms/molecules of “atoms of spacetime”

IAGRG, Feb 07
 THE SPACETIME SOLID!

• We will take this analogy seriously.

IAGRG, Feb 07
 THE SPACETIME SOLID!

• We will take this analogy seriously.

• The gab (t, x) is non-dynamical and is more like density ρ(t, x) of a

solid.

IAGRG, Feb 07
 THE SPACETIME SOLID!

• We will take this analogy seriously.

• The gab (t, x) is non-dynamical and is more like density ρ(t, x) of a

solid.

• Long wavelength dynamics should come from a different variational

principle.

IAGRG, Feb 07
 THE SPACETIME SOLID!

• We will take this analogy seriously.

• The gab (t, x) is non-dynamical and is more like density ρ(t, x) of a

solid.

• Long wavelength dynamics should come from a different variational

principle.

• Strategy: Associate an entropy with a null surfaces. Demand

maximisation of entropy of all null surfaces to get the dynamics.

IAGRG, Feb 07
 THE SPACETIME SOLID!

• We will take this analogy seriously.

• The gab (t, x) is non-dynamical and is more like density ρ(t, x) of a

solid.

• Long wavelength dynamics should come from a different variational

principle.

• Strategy: Associate an entropy with a null surfaces. Demand

maximisation of entropy of all null surfaces to get the dynamics.

• Note: There is no ‘quantum thermodynamics’; for example,

T dS = dE + P dV is valid with quantum corrections as well.

IAGRG, Feb 07
A PRIMER ON ELASTICITY

IAGRG, Feb 07
 A PRIMER ON ELASTICITY

• Deformation field : xµ → xµ + v µ (xν ).

IAGRG, Feb 07
 A PRIMER ON ELASTICITY

• Deformation field : xµ → xµ + v µ (xν ).

• The entropy is a quadratic functional of Qµν ≡ ∇µ vν . Decompose

Qµν into shear, expansion and rotation. Choose different
coefficients (‘elastic moduli’) to describe different situations.

IAGRG, Feb 07
 A PRIMER ON ELASTICITY

• Deformation field : xµ → xµ + v µ (xν ).

• The entropy is a quadratic functional of Qµν ≡ ∇µ vν . Decompose

Qµν into shear, expansion and rotation. Choose different
coefficients (‘elastic moduli’) to describe different situations.

• The form of entropy functional is phenomenological. Only a

microscopic theory can predict it.

IAGRG, Feb 07
 A PRIMER ON ELASTICITY

• Deformation field : xµ → xµ + v µ (xν ).

• The entropy is a quadratic functional of Qµν ≡ ∇µ vν . Decompose

Qµν into shear, expansion and rotation. Choose different
coefficients (‘elastic moduli’) to describe different situations.

• The form of entropy functional is phenomenological. Only a

microscopic theory can predict it.

• Our state of knowledge of quantum gravity is worse that that of

18th century physicists about microstructure of solids! So we
should not hesitate to be phenomenological.

IAGRG, Feb 07
 ENTROPY OF SPACETIME AND GRAVITY
T.P, Gen.Rel.Grav., (2006), T.P.,A.Paranjape, (2007)

• Deformation: xa → x̄a = xa + ξ a (x).

IAGRG, Feb 07
 ENTROPY OF SPACETIME AND GRAVITY
T.P, Gen.Rel.Grav., (2006), T.P.,A.Paranjape, (2007)

• Deformation: xa → x̄a = xa + ξ a (x).

• Associate with every vector field an entropy which is quadratic in

deformation field:
Z
D √ abcd ab


S[ξ] = d x −g 4P ∇ c ξa ∇ d ξb − T ξa ξb
V

IAGRG, Feb 07
 ENTROPY OF SPACETIME AND GRAVITY
T.P, Gen.Rel.Grav., (2006), T.P.,A.Paranjape, (2007)

• Deformation: xa → x̄a = xa + ξ a (x).

• Associate with every vector field an entropy which is quadratic in

deformation field:
Z
D √ abcd ab


S[ξ] = d x −g 4P ∇ c ξa ∇ d ξb − T ξa ξb
V

• Demand that: (i) P abcd has the symmetries of Riemann tensor; (ii)
Covariant divergence on all indices are zero: ∇a P abcd = 0; ∇a T ab = 0.
[Analogue of elastic “constants” ]

IAGRG, Feb 07
 ENTROPY OF SPACETIME AND GRAVITY
T.P, Gen.Rel.Grav., (2006), T.P.,A.Paranjape, (2007)

• Deformation: xa → x̄a = xa + ξ a (x).

• Associate with every vector field an entropy which is quadratic in

deformation field:
Z
D √ abcd ab


S[ξ] = d x −g 4P ∇ c ξa ∇ d ξb − T ξa ξb
V

• Demand that: (i) P abcd has the symmetries of Riemann tensor; (ii)
Covariant divergence on all indices are zero: ∇a P abcd = 0; ∇a T ab = 0.
[Analogue of elastic “constants” ]

• We expect P abcd to have a (RG-like) derivative expansion in powers

of number of derivatives of the metric:
(1) (2)
abcd abcd
P (gij , Rijkl ) = c1 P (gij ) + c2 P abcd (gij , Rijkl ) + · · · ,

IAGRG, Feb 07
 ab
THE DERIVATIVE EXPANSION FOR Pcd

IAGRG, Feb 07
 ab
THE DERIVATIVE EXPANSION FOR Pcd

• If we do not use the curvature tensor:

(1) 1 1 1 ab
P ab
cd = (δca δdb − δda δcb ) = δcd .
32π 16π 2

IAGRG, Feb 07
 ab
THE DERIVATIVE EXPANSION FOR Pcd

• If we do not use the curvature tensor:

(1) 1 1 1 ab
P ab
cd = (δca δdb − δda δcb ) = δcd .
32π 16π 2
• Linear on curvature:
(2) 1
1 1 aba3 a4 b3 b4
P ab
cd = ab a b b a a b b a
Rcd − Gc δd + Gc δd + Rd δc − Rd δc = δ Ra a .
8π 16π 2 cd b3 b4 3 4

IAGRG, Feb 07
 ab
THE DERIVATIVE EXPANSION FOR Pcd

• If we do not use the curvature tensor:

(1) 1 1 1 ab
P ab
cd = (δca δdb − δda δcb ) = δcd .
32π 16π 2
• Linear on curvature:
(2) 1
1 1 aba3 a4 b3 b4
P ab
cd = ab a b b a a b b a
Rcd − Gc δd + Gc δd + Rd δc − Rd δc = δ Ra a .
8π 16π 2 cd b3 b4 3 4
• The m-th order term is unique:
(m)
aba3 ...a2m b3 b4
P ab
cd ∝ δcdb 3 ...b2m
Ra3 a3 · · · Rab2m−1 b2m
2m−1 a2m
.

IAGRG, Feb 07
 ab
THE DERIVATIVE EXPANSION FOR Pcd

• If we do not use the curvature tensor:

(1) 1 1 1 ab
P ab
cd = (δca δdb − δda δcb ) = δcd .
32π 16π 2
• Linear on curvature:
(2) 1
1 1 aba3 a4 b3 b4
P ab
cd = ab a b b a a b b a
Rcd − Gc δd + Gc δd + Rd δc − Rd δc = δ Ra a .
8π 16π 2 cd b3 b4 3 4
• The m-th order term is unique:
(m)
aba3 ...a2m b3 b4
P ab
cd ∝ δcdb 3 ...b2m
Ra3 a3 · · · Rab2m−1 b2m
2m−1 a2m
.

• Explicit form:
Z
dD x

S1 [ξ] = (∇c ξ c )2 − ∇a ξ b ∇b ξ a
V 8π
Z


S2 [ξ] = c2 dD x Rab
cd
∇c ξ a ∇d ξ b − (Gca + Rac )(∇c ξ a ∇b ξ b − ∇c ξ b ∇b ξ a )
V

IAGRG, Feb 07
 ab
THE DERIVATIVE EXPANSION FOR Pcd

• If we do not use the curvature tensor:

(1) 1 1 1 ab
P ab
cd = (δca δdb − δda δcb ) = δcd .
32π 16π 2
• Linear on curvature:
(2) 1
1 1 aba3 a4 b3 b4
P ab
cd = ab a b b a a b b a
Rcd − Gc δd + Gc δd + Rd δc − Rd δc = δ Ra a .
8π 16π 2 cd b3 b4 3 4
• The m-th order term is unique:
(m)
aba3 ...a2m b3 b4
P ab
cd ∝ δcdb 3 ...b2m
Ra3 a3 · · · Rab2m−1 b2m
2m−1 a2m
.

• Explicit form:
Z
dD x c 2 b a


S1 [ξ] = (∇c ξ ) − ∇a ξ ∇b ξ
V 8π
Z


S2 [ξ] = c2 dD x Rab
cd
∇c ξ a ∇d ξ b − (Gca + Rac )(∇c ξ a ∇b ξ b − ∇c ξ b ∇b ξ a )
V

• Closely related to the Lanczos-Lovelock Lagrangian:

∂Lm
Pa ijk = .
∂Raijk

IAGRG, Feb 07
PRIMER ON LANCZOS-LOVELOCK GRAVITY
T.P (2006); A.Mukhopadhyay and T.P (2006)

IAGRG, Feb 07
 PRIMER ON LANCZOS-LOVELOCK GRAVITY
T.P (2006); A.Mukhopadhyay and T.P (2006)

• A very natural, geometrical generalization of Einstein’s theory in D-dimensions.

IAGRG, Feb 07
 PRIMER ON LANCZOS-LOVELOCK GRAVITY
T.P (2006); A.Mukhopadhyay and T.P (2006)

• A very natural, geometrical generalization of Einstein’s theory in D-dimensions.

• The D-dimensional Lanczos-Lovelock Lagrangian is a polynomial in the curvature

tensor (with D ≥ (2K + 1)):
K
X 1 −m a1 a2 ...a2m b1 b2 b2m−1 b2m
L = Qabcd Rabcd = cm L m ; L m = 2 δb1 b2 ...b2m Ra1 a2 Ra2m−1 a2m ,
m=1
16π

IAGRG, Feb 07
 PRIMER ON LANCZOS-LOVELOCK GRAVITY
T.P (2006); A.Mukhopadhyay and T.P (2006)

• A very natural, geometrical generalization of Einstein’s theory in D-dimensions.

• The D-dimensional Lanczos-Lovelock Lagrangian is a polynomial in the curvature

tensor (with D ≥ (2K + 1)):
K
X 1 −m a1 a2 ...a2m b1 b2 b2m−1 b2m
L = Qabcd Rabcd = cm L m ; L m = 2 δb1 b2 ...b2m Ra1 a2 Ra2m−1 a2m ,
m=1
16π

• The Lanczos-Lovelock Lagrangian separates to a bulk and surface terms

√ √  √
−gL = 2∂c −gQabcd Γabd + 2 −gQabcd Γadk Γkbc ≡ Lsur + Lbulk

and is ‘holographic’:
 
δLbulk ∂Lbulk
[(D/2) − m]Lsur = −∂i gab + ∂j gab
δ(∂i gab ) ∂(∂i ∂j gab )

IAGRG, Feb 07
 PRIMER ON LANCZOS-LOVELOCK GRAVITY
T.P (2006); A.Mukhopadhyay and T.P (2006)

• A very natural, geometrical generalization of Einstein’s theory in D-dimensions.

• The D-dimensional Lanczos-Lovelock Lagrangian is a polynomial in the curvature

tensor (with D ≥ (2K + 1)):
K
X 1 −m a1 a2 ...a2m b1 b2 b2m−1 b2m
L = Qabcd Rabcd = cm L m ; L m = 2 δb1 b2 ...b2m Ra1 a2 Ra2m−1 a2m ,
m=1
16π

• The Lanczos-Lovelock Lagrangian separates to a bulk and surface terms

√ √  √
−gL = 2∂c −gQabcd Γabd + 2 −gQabcd Γadk Γkbc ≡ Lsur + Lbulk

and is ‘holographic’:
 
δLbulk ∂Lbulk
[(D/2) − m]Lsur = −∂i gab + ∂j gab
δ(∂i gab ) ∂(∂i ∂j gab )

• The surface term is closely related to horizon entropy in Lanczos-Lovelock theory.

IAGRG, Feb 07
EXTREMISING ENTROPY TO GET DYNAMICS

IAGRG, Feb 07
 EXTREMISING ENTROPY TO GET DYNAMICS

• Demand that δS = 0 for variations of all null vectors:

2Pab cd (∇c ∇d − ∇d ∇c ) ξ a − T̄ab ξ a = 0 ,

IAGRG, Feb 07
 EXTREMISING ENTROPY TO GET DYNAMICS

• Demand that δS = 0 for variations of all null vectors:

2Pab cd (∇c ∇d − ∇d ∇c ) ξ a − T̄ab ξ a = 0 ,

• This leads to Lanczos-Lovelock theory with an arbitrary

cosmological constant:
 
1
16π Pb ijk Raijk − δba Lm = 8πT ab + Λδba ,
2

IAGRG, Feb 07
 EXTREMISING ENTROPY TO GET DYNAMICS

• Demand that δS = 0 for variations of all null vectors:

2Pab cd (∇c ∇d − ∇d ∇c ) ξ a − T̄ab ξ a = 0 ,

• This leads to Lanczos-Lovelock theory with an arbitrary

cosmological constant:
 
1
16π Pb ijk Raijk − δba Lm = 8πT ab + Λδba ,
2
• To the lowest order we get Einstein’s theory with cosmological
constant as integration constant. Equivalent to

(Gab − 8πTab )ξ a ξ b = 0

IAGRG, Feb 07
 EXTREMISING ENTROPY TO GET DYNAMICS

• Demand that δS = 0 for variations of all null vectors:

2Pab cd (∇c ∇d − ∇d ∇c ) ξ a − T̄ab ξ a = 0 ,

• This leads to Lanczos-Lovelock theory with an arbitrary

cosmological constant:
 
1
16π Pb ijk Raijk − δba Lm = 8πT ab + Λδba ,
2
• To the lowest order we get Einstein’s theory with cosmological
constant as integration constant. Equivalent to

(Gab − 8πTab )ξ a ξ b = 0

• In a derivative coupling expansion, Lanczos-Lovelock terms are

corrections.

IAGRG, Feb 07
EXTREMUM VALUE OF ENTROPY

IAGRG, Feb 07
 EXTREMUM VALUE OF ENTROPY

• The extremum value of entropy can be computed on-shell on a

solution.

IAGRG, Feb 07
 EXTREMUM VALUE OF ENTROPY

• The extremum value of entropy can be computed on-shell on a

solution.

• On any solution with horizon it gives the standard result (Wald

entropy):
I K Z
 abcd
X
D−2
√
S H [on − shell] = 2π P nab ncd˜
= 4πmcm d x⊥ σL(m−1) ,
H m=1 H

IAGRG, Feb 07
 EXTREMUM VALUE OF ENTROPY

• The extremum value of entropy can be computed on-shell on a

solution.

• On any solution with horizon it gives the standard result (Wald

entropy):
I K Z
 abcd
X
D−2
√
S H [on − shell] = 2π P nab ncd˜
= 4πmcm d x⊥ σL(m−1) ,
H m=1 H

• At the lowest order, we get quarter of transverse area as entropy.

IAGRG, Feb 07
 EXTREMUM VALUE OF ENTROPY

• The extremum value of entropy can be computed on-shell on a

solution.

• On any solution with horizon it gives the standard result (Wald

entropy):
I K Z
 abcd
X
D−2
√
S H [on − shell] = 2π P nab ncd˜
= 4πmcm d x⊥ σL(m−1) ,
H m=1 H

• At the lowest order, we get quarter of transverse area as entropy.

• For any solution, in a local Rindler frame, the causal horizons have
the correct entropy.

IAGRG, Feb 07
THE REAL TROUBLE COSMOLOGICAL CONSTANT

IAGRG, Feb 07
THE REAL TROUBLE COSMOLOGICAL CONSTANT
Gravity seems to be immune to bulk vacuum energy

IAGRG, Feb 07
 THE REAL TROUBLE COSMOLOGICAL CONSTANT
Gravity seems to be immune to bulk vacuum energy

• The matter sector and its equations are invariant under the shift of
the Lagrangian by a constant: Lmatter → Lmatter − ρ.

IAGRG, Feb 07
 THE REAL TROUBLE COSMOLOGICAL CONSTANT
Gravity seems to be immune to bulk vacuum energy

• The matter sector and its equations are invariant under the shift of
the Lagrangian by a constant: Lmatter → Lmatter − ρ.

• But this changes energy momentum tensor by Tab → Tab + ρgab and
gravity sector is not invariant under this transformation.

IAGRG, Feb 07
 THE REAL TROUBLE COSMOLOGICAL CONSTANT
Gravity seems to be immune to bulk vacuum energy

• The matter sector and its equations are invariant under the shift of
the Lagrangian by a constant: Lmatter → Lmatter − ρ.

• But this changes energy momentum tensor by Tab → Tab + ρgab and
gravity sector is not invariant under this transformation.

• So after you have “solved” the cosmological constant problem, if

someone introduces Lmatter → Lmatter − ρ, you are in trouble again!

IAGRG, Feb 07
 THE REAL TROUBLE COSMOLOGICAL CONSTANT
Gravity seems to be immune to bulk vacuum energy

• The matter sector and its equations are invariant under the shift of
the Lagrangian by a constant: Lmatter → Lmatter − ρ.

• But this changes energy momentum tensor by Tab → Tab + ρgab and
gravity sector is not invariant under this transformation.

• So after you have “solved” the cosmological constant problem, if

someone introduces Lmatter → Lmatter − ρ, you are in trouble again!

• The only way out is to have a formalism for gravity which is

invariant under Tab → Tab + ρgab .

IAGRG, Feb 07
 THE REAL TROUBLE COSMOLOGICAL CONSTANT
Gravity seems to be immune to bulk vacuum energy

• The matter sector and its equations are invariant under the shift of
the Lagrangian by a constant: Lmatter → Lmatter − ρ.

• But this changes energy momentum tensor by Tab → Tab + ρgab and
gravity sector is not invariant under this transformation.

• So after you have “solved” the cosmological constant problem, if

someone introduces Lmatter → Lmatter − ρ, you are in trouble again!

• The only way out is to have a formalism for gravity which is

invariant under Tab → Tab + ρgab .

• This will also make gravity immune to bulk vacuum energy density.

IAGRG, Feb 07
GRAVITY IS IMMUNE TO BULK ENERGY

IAGRG, Feb 07
 GRAVITY IS IMMUNE TO BULK ENERGY

• The entropy functional for null vectors is invariant under

the shift Tab → Tab + ρgab !

IAGRG, Feb 07
 GRAVITY IS IMMUNE TO BULK ENERGY

• The entropy functional for null vectors is invariant under

the shift Tab → Tab + ρgab !

• The field equations have a new ‘gauge freedom’ and has

the form:
ijk 1 a
Pb Raijk − Lδb − κTba = (constant)δba
2

IAGRG, Feb 07
 GRAVITY IS IMMUNE TO BULK ENERGY

• The entropy functional for null vectors is invariant under

the shift Tab → Tab + ρgab !

• The field equations have a new ‘gauge freedom’ and has

the form:
ijk 1 a
Pb Raijk − Lδb − κTba = (constant)δba
2
• The classical bulk energy density of the form Tba = ρδba
can be gauged away.

IAGRG, Feb 07
 GRAVITY IS IMMUNE TO BULK ENERGY

• The entropy functional for null vectors is invariant under

the shift Tab → Tab + ρgab !

• The field equations have a new ‘gauge freedom’ and has

the form:
ijk 1 a
Pb Raijk − Lδb − κTba = (constant)δba
2
• The classical bulk energy density of the form Tba = ρδba
can be gauged away.

• But not the quantum fluctuations in the energy density.

IAGRG, Feb 07
 VACUUM FLUCTUATIONS AND ρvac
T.P, CQG, 22, L107-L110, (2005) hep-th/0406060

• The coupling between gravity and Tab allows gravity to act as

“detector” of large scale vacuum fluctuations in energy density.

IAGRG, Feb 07
 VACUUM FLUCTUATIONS AND ρvac
T.P, CQG, 22, L107-L110, (2005) hep-th/0406060

• The coupling between gravity and Tab allows gravity to act as

“detector” of large scale vacuum fluctuations in energy density.

• Given LP and LH we have ρUV = 1/L4P and ρIR = 1/L4H .

IAGRG, Feb 07
 VACUUM FLUCTUATIONS AND ρvac
T.P, CQG, 22, L107-L110, (2005) hep-th/0406060

• The coupling between gravity and Tab allows gravity to act as

“detector” of large scale vacuum fluctuations in energy density.

• Given LP and LH we have ρUV = 1/L4P and ρIR = 1/L4H . The observed
values is:
√ 1 H2
ρDE ≈ ρUV ρIR ≈ 2 2 ≈
LP LH G

IAGRG, Feb 07
 VACUUM FLUCTUATIONS AND ρvac
T.P, CQG, 22, L107-L110, (2005) hep-th/0406060

• The coupling between gravity and Tab allows gravity to act as

“detector” of large scale vacuum fluctuations in energy density.

• Given LP and LH we have ρUV = 1/L4P and ρIR = 1/L4H . The observed
values is:
√ 1 H2
ρDE ≈ ρUV ρIR ≈ 2 2 ≈
LP LH G
• The hierarchy:
 2  4
1 1 LP 1 LP
ρvac = + + 4 + ···
L4P L4P LH LP LH

IAGRG, Feb 07
 VACUUM FLUCTUATIONS AND ρvac
T.P, CQG, 22, L107-L110, (2005) hep-th/0406060

• The coupling between gravity and Tab allows gravity to act as

“detector” of large scale vacuum fluctuations in energy density.

• Given LP and LH we have ρUV = 1/L4P and ρIR = 1/L4H . The observed
values is:
√ 1 H2
ρDE ≈ ρUV ρIR ≈ 2 2 ≈
LP LH G
• The hierarchy:
 2  4
1 1 LP 1 LP
ρvac = + + 4 + ···
L4P L4P LH LP LH

Important: Bulk term

is now ignored by gravity

IAGRG, Feb 07
 VACUUM FLUCTUATIONS AND ρvac
T.P, CQG, 22, L107-L110, (2005) hep-th/0406060

• The coupling between gravity and Tab allows gravity to act as

“detector” of large scale vacuum fluctuations in energy density.

• Given LP and LH we have ρUV = 1/L4P and ρIR = 1/L4H . The observed
values is:
√ 1 H2
ρDE ≈ ρUV ρIR ≈ 2 2 ≈
LP LH G
• The hierarchy:
 2  4
1 1 LP 1 LP
ρvac = + + 4 + ···
L4P L4P LH LP LH

observed value; comes

Important: Bulk term
directly from the
is now ignored by gravity
quantum fluctuations

IAGRG, Feb 07
 VACUUM FLUCTUATIONS AND ρvac
T.P, CQG, 22, L107-L110, (2005) hep-th/0406060

• The coupling between gravity and Tab allows gravity to act as

“detector” of large scale vacuum fluctuations in energy density.

• Given LP and LH we have ρUV = 1/L4P and ρIR = 1/L4H . The observed
values is:
√ 1 H2
ρDE ≈ ρUV ρIR ≈ 2 2 ≈
LP LH G
• The hierarchy:
 2  4
1 1 LP 1 LP
ρvac = + + 4 + ···
L4P L4P LH LP LH

observed value; comes

Important: Bulk term thermal energy density
directly from the 4
is now ignored by gravity of de Sitter ρrad ∝ TGH
quantum fluctuations

IAGRG, Feb 07
SUMMARY

IAGRG, Feb 07
 SUMMARY

• Gravity described by is an emergent, long-wavelength phenomenon

like elasticity. The gab (t, x) is like ρ(t, x).

IAGRG, Feb 07
 SUMMARY

• Gravity described by is an emergent, long-wavelength phenomenon

like elasticity. The gab (t, x) is like ρ(t, x).

• Maximizing the entropy associated with all null surfaces gives

Einstein’s theory with Lanczos-Lovelock corrections [but not, e.g.,
f (R) gravity].

IAGRG, Feb 07
 SUMMARY

• Gravity described by is an emergent, long-wavelength phenomenon

like elasticity. The gab (t, x) is like ρ(t, x).

• Maximizing the entropy associated with all null surfaces gives

Einstein’s theory with Lanczos-Lovelock corrections [but not, e.g.,
f (R) gravity].

• Since metric is not dynamical, usual problems of surface term etc.

in the action do not arise.

IAGRG, Feb 07
 SUMMARY

• Gravity described by is an emergent, long-wavelength phenomenon

like elasticity. The gab (t, x) is like ρ(t, x).

• Maximizing the entropy associated with all null surfaces gives

Einstein’s theory with Lanczos-Lovelock corrections [but not, e.g.,
f (R) gravity].

• Since metric is not dynamical, usual problems of surface term etc.

in the action do not arise.

• Connects with the radial displacements of horizons and

T dS = dE + P dV as the key to obtaining a thermodynamic
interpretation of gravitational theories.

IAGRG, Feb 07
 SUMMARY

• Gravity described by is an emergent, long-wavelength phenomenon

like elasticity. The gab (t, x) is like ρ(t, x).

• Maximizing the entropy associated with all null surfaces gives

Einstein’s theory with Lanczos-Lovelock corrections [but not, e.g.,
f (R) gravity].

• Since metric is not dynamical, usual problems of surface term etc.

in the action do not arise.

• Connects with the radial displacements of horizons and

T dS = dE + P dV as the key to obtaining a thermodynamic
interpretation of gravitational theories.

• The deep connection between gravity and thermodynamics goes

well beyond Einstein’s theory. Closely related to the holographic
structure of Lanczos-Lovelock theories.

IAGRG, Feb 07
OPEN QUESTIONS

IAGRG, Feb 07
 OPEN QUESTIONS

• Why does all this work ?

ab
– Conditions on Pcd , derivatives of curvature tensor ......
– Thermodynamics goes beyond Einstein’s theory!
– Does not work in gauge theories, f (R) gravity, ....

IAGRG, Feb 07
 OPEN QUESTIONS

• Why does all this work ?

ab
– Conditions on Pcd , derivatives of curvature tensor ......
– Thermodynamics goes beyond Einstein’s theory!
– Does not work in gauge theories, f (R) gravity, ....

• Matter sector is horribly inelegant. Possible clue in

Z  
D ∂L a b 1 ∂L a b
S= d x 4 ∇ ξ ∇d ξ +
ab c ab
ξ ξ
∂Rcd 2 ∂g

IAGRG, Feb 07
 OPEN QUESTIONS

• Why does all this work ?

ab
– Conditions on Pcd , derivatives of curvature tensor ......
– Thermodynamics goes beyond Einstein’s theory!
– Does not work in gauge theories, f (R) gravity, ....

• Matter sector is horribly inelegant. Possible clue in

Z  
D ∂L a b 1 ∂L a b
S= d x 4 ∇ ξ ∇d ξ +
ab c ab
ξ ξ
∂Rcd 2 ∂g

• Analogy: horizons in euclidean spacetime ↔ defects in solids. Clue

for a new path integral prescription ?

IAGRG, Feb 07
 REFERENCES

1. Original ideas were developed in:

• T. Padmanabhan, Class.Quan.Grav. 19, 5387 (2002). [gr-qc/0204019]
• T. Padmanabhan, Gen.Rel.Grav., 34 2029-2035 (2002) [gr-qc/0205090] [Second
Prize essay; Gravity Research Foundation Essay Contest, 2002]
• T. Padmanabhan, Gen.Rel.Grav., 35, 2097-2103 (2003) [Fifth Prize essay; Gravity
Research Foundation Essay Contest, 2003]
• T. Padmanabhan, Gen.Rel.Grav., 38, 1547-1552 (2006) [Third Prize essay; Gravity
Research Foundation Essay Contest, 2006]

2. Summary of the basic approach is in:

• T. Padmanabhan Phys. Reports, 406, 49 (2005) [gr-qc/0311036]

3. Current results are based on:

• Aseem Paranjape, Sudipta Sarkar, T. Padmanabhan, Phys.Rev., D 74, 104015
(2006) [hep-th/0607240]
• A. Mukhopadhyay, T. Padmanabhan, Phys.Rev., D 74, 124023 (2006)
[hep-th/0608120]
• Dawood Kothawala, Sudipta Sarkar, T. Padmanabhan, [gr-qc/0701002]
• T. Padmanabhan, Aseem Paranjape, [gr-qc/0701003]

IAGRG, Feb 07