Hamilton-Jacobi counterterms in quantum mechanical and black hole partition functions
Daniel Grumiller
Center for Theoretical Physics Massachusetts Institute of Technology
7th Workshop on Quantization, Dualities and Integrable Systems, Anadolu University,April 2008
hep-th/0703230, 0711.4115
Appetizer: Action Principles And... Action!
I = Z
M
dd+1x
(∂φ)2−V(φ)
First variation δI= 0 yields bulk EOM and boundary conditions: boundary :π δφ= 0 → chooseφ= const.at∂M
Problem: If π→ ∞ or φ→ ∞ at boundary thenπ δφ6= 0 possible Solution: subtractboundary counterterm
Γ =I−ICT , ICT = Z
∂M
ddxL(φ, ∂kφ) Γ has same EOM and same boundary value problem asI
Question: How to construct the countertermICT?
Another question: Why is this relevant?
Appetizer: Action Principles And... Action!
I = Z
M
dd+1x
(∂φ)2−V(φ)
First variation δI= 0 yields bulk EOM and boundary conditions:
boundary :π δφ= 0 → chooseφ= const.at∂M
Problem: If π→ ∞ or φ→ ∞ at boundary thenπ δφ6= 0 possible Solution: subtractboundary counterterm
Γ =I−ICT , ICT = Z
∂M
ddxL(φ, ∂kφ) Γ has same EOM and same boundary value problem asI
Question: How to construct the countertermICT?
Another question: Why is this relevant?
D. Grumiller — Hamilton-Jacobi counterterms 2/26
Appetizer: Action Principles And... Action!
I = Z
M
dd+1x
(∂φ)2−V(φ)
First variation δI= 0 yields bulk EOM and boundary conditions:
boundary :π δφ= 0 → chooseφ= const.at∂M Problem: If π→ ∞ or φ→ ∞ at boundary thenπ δφ6= 0 possible
Solution: subtractboundary counterterm Γ =I−ICT , ICT =
Z
∂M
ddxL(φ, ∂kφ) Γ has same EOM and same boundary value problem asI
Question: How to construct the countertermICT?
Another question: Why is this relevant?
Appetizer: Action Principles And... Action!
I = Z
M
dd+1x
(∂φ)2−V(φ)
First variation δI= 0 yields bulk EOM and boundary conditions:
boundary :π δφ= 0 → chooseφ= const.at∂M
Problem: If π→ ∞ or φ→ ∞ at boundary thenπ δφ6= 0 possible Solution: subtractboundary counterterm
Γ =I−ICT , ICT = Z
∂M
ddxL(φ, ∂kφ) Γ has same EOM and same boundary value problem asI
Question: How to construct the countertermICT?
Another question: Why is this relevant?
D. Grumiller — Hamilton-Jacobi counterterms 2/26
Appetizer: Action Principles And... Action!
I = Z
M
dd+1x
(∂φ)2−V(φ)
First variation δI= 0 yields bulk EOM and boundary conditions:
boundary :π δφ= 0 → chooseφ= const.at∂M
Problem: If π→ ∞ or φ→ ∞ at boundary thenπ δφ6= 0 possible Solution: subtractboundary counterterm
Γ =I−ICT , ICT = Z
∂M
ddxL(φ, ∂kφ) Γ has same EOM and same boundary value problem asI
Question: How to construct the countertermICT?
Another question: Why is this relevant?
Appetizer: Action Principles And... Action!
I = Z
M
dd+1x
(∂φ)2−V(φ)
First variation δI= 0 yields bulk EOM and boundary conditions:
boundary :π δφ= 0 → chooseφ= const.at∂M
Problem: If π→ ∞ or φ→ ∞ at boundary thenπ δφ6= 0 possible Solution: subtractboundary counterterm
Γ =I−ICT , ICT = Z
∂M
ddxL(φ, ∂kφ) Γ has same EOM and same boundary value problem asI
Question: How to construct the countertermICT? Another question: Why is this relevant?
D. Grumiller — Hamilton-Jacobi counterterms 2/26
Outline
Statement of problem: divergences and variational principle
Solution: Hamilton-Jacobi counterterm
Applications: Black Hole Thermodynamics, AdS/CFT, ...
Summary and Literature
Outline
Statement of problem: divergences and variational principle
Solution: Hamilton-Jacobi counterterm
Applications: Black Hole Thermodynamics, AdS/CFT, ...
Summary and Literature
D. Grumiller — Hamilton-Jacobi counterterms Statement of problem: divergences and variational principle 4/26
Quantum Mechanical Toy Model
Step 1: Boundary value problem
Let us start with an “extraordinary” Hamiltonianaction IEH =
tf
Z
ti
dt[−pq˙ −H(q, p)]
Want to set up a Dirichlet boundary value problemq= fixed atti, tf Problem:
δIEH = 0 requiresq δp= 0 at boundary Solution: add“great, helpful (yellowish)”boundary term
IE =IEH +IGHY , IGHY = pq|ttf
i
As expectedIE =
tf
R
ti
[pq˙−H(q, p)] is standard Hamiltonian action
Quantum Mechanical Toy Model
Step 1: Boundary value problem
Let us start with an “extraordinary” Hamiltonianaction IEH =
tf
Z
ti
dt[−pq˙ −H(q, p)]
Want to set up a Dirichlet boundary value problemq= fixed atti, tf
Problem:
δIEH = 0 requiresq δp= 0 at boundary Solution: add“great, helpful (yellowish)”boundary term
IE =IEH +IGHY , IGHY = pq|ttf
i
As expectedIE =
tf
R
ti
[pq˙−H(q, p)] is standard Hamiltonian action
D. Grumiller — Hamilton-Jacobi counterterms Statement of problem: divergences and variational principle 5/26
Quantum Mechanical Toy Model
Step 1: Boundary value problem
Let us start with an “extraordinary” Hamiltonianaction IEH =
tf
Z
ti
dt[−pq˙ −H(q, p)]
Want to set up a Dirichlet boundary value problemq= fixed atti, tf Problem:
δIEH = 0 requiresq δp= 0 at boundary
Solution: add“great, helpful (yellowish)”boundary term IE =IEH +IGHY , IGHY = pq|ttf
i
As expectedIE =
tf
R
ti
[pq˙−H(q, p)] is standard Hamiltonian action
Quantum Mechanical Toy Model
Step 1: Boundary value problem
Let us start with an “extraordinary” Hamiltonianaction IEH =
tf
Z
ti
dt[−pq˙ −H(q, p)]
Want to set up a Dirichlet boundary value problemq= fixed atti, tf Problem:
δIEH = 0 requiresq δp= 0 at boundary Solution: add“great, helpful (yellowish)”boundary term
IE =IEH +IGHY , IGHY = pq|ttf
i
As expectedIE =
tf
R
ti
[pq˙−H(q, p)] is standard Hamiltonian action
D. Grumiller — Hamilton-Jacobi counterterms Statement of problem: divergences and variational principle 5/26
Quantum Mechanical Toy Model
Step 1: Boundary value problem
Let us start with an “extraordinary” Hamiltonianaction IEH =
tf
Z
ti
dt[−pq˙ −H(q, p)]
Want to set up a Dirichlet boundary value problemq= fixed atti, tf Problem:
δIEH = 0 requiresq δp= 0 at boundary Solution: add“great, helpful (yellowish)”boundary term
IE =IEH +IGHY , IGHY = pq|ttf
i
As expectedIE =
tf
R[pq˙−H(q, p)] is standard Hamiltonian action
Half-binding potentials
Step 2: Impose asymptotic boundary conditions
Specific Hamiltonian with half-binding potential:
H(q, p) = p2
2 +V(q), V(q) = 1 q2
1 2 3 4 5 q
0.2 0.4 0.6 0.8 1 V
I Of interest is asymptotic limittf → ∞
I In that case almost free propagation
I Appropriate boundary condition: qf =∞
D. Grumiller — Hamilton-Jacobi counterterms Statement of problem: divergences and variational principle 6/26
Half-binding potentials
Step 2: Impose asymptotic boundary conditions
Specific Hamiltonian with half-binding potential:
H(q, p) = p2
2 +V(q), V(q) = 1 q2
1 2 3 4 5 q
0.2 0.4 0.6 0.8 1 V
I Of interest is asymptotic limittf → ∞
I In that case almost free propagation
I Appropriate boundary condition: qf =∞
Half-binding potentials
Step 2: Impose asymptotic boundary conditions
Specific Hamiltonian with half-binding potential:
H(q, p) = p2
2 +V(q), V(q) = 1 q2
1 2 3 4 5 q
0.2 0.4 0.6 0.8 1 V
I Of interest is asymptotic limittf → ∞
I In that case almost free propagation
I Appropriate boundary condition: qf =∞
D. Grumiller — Hamilton-Jacobi counterterms Statement of problem: divergences and variational principle 6/26
Half-binding potentials
Step 2: Impose asymptotic boundary conditions
Specific Hamiltonian with half-binding potential:
H(q, p) = p2
2 +V(q), V(q) = 1 q2
1 2 3 4 5 q
0.2 0.4 0.6 0.8 1 V
I Of interest is asymptotic limittf → ∞
I In that case almost free propagation
A Problem Arises...
Step 3: Variational principle
Consider the first variation of the action δIE =δIEH +δIGHY =
tf
Z
ti
dt[ ˙q δp+p δq˙−δH(q, p)]
I Bulk: Hamilton EOM
I Boundary (recallq|tf =∞):
δIE =p δq|tf −p δq|ti =p δq
|{z}
finite!
|tf 6= 0.
Not all variations preserving the boundary conditionqf =∞ lead to vanishing first variation of the actionIE!
SERIOUS PROBLEM!
D. Grumiller — Hamilton-Jacobi counterterms Statement of problem: divergences and variational principle 7/26
A Problem Arises...
Step 3: Variational principle
Consider the first variation of the action δIE =δIEH +δIGHY =
tf
Z
ti
dt[ ˙q δp+p δq˙−δH(q, p)]
I Bulk: Hamilton EOM
I Boundary (recallq|tf =∞):
δIE =p δq|tf −p δq|ti =p δq
|{z}
finite!
|tf 6= 0.
Not all variations preserving the boundary conditionqf =∞ lead to vanishing first variation of the actionIE!
SERIOUS PROBLEM!
A Problem Arises...
Step 3: Variational principle
Consider the first variation of the action δIE =δIEH +δIGHY =
tf
Z
ti
dt[ ˙q δp+p δq˙−δH(q, p)]
I Bulk: Hamilton EOM
I Boundary (recallq|tf =∞):
δIE =p δq|tf −p δq|ti =p δq
|{z}
finite!
|tf 6= 0.
Not all variations preserving the boundary conditionqf =∞ lead to vanishing first variation of the actionIE!
SERIOUS PROBLEM!
D. Grumiller — Hamilton-Jacobi counterterms Statement of problem: divergences and variational principle 7/26
A Problem Arises...
Step 3: Variational principle
Consider the first variation of the action δIE =δIEH +δIGHY =
tf
Z
ti
dt[ ˙q δp+p δq˙−δH(q, p)]
I Bulk: Hamilton EOM
I Boundary (recallq|tf =∞):
δIE =p δq|tf −p δq|ti =p δq
|{z}
finite!
|tf 6= 0.
Not all variations preserving the boundary conditionqf =∞ lead to vanishing first variation of the actionIE!
SERIOUS PROBLEM!
A Problem Arises...
Step 3: Variational principle
Consider the first variation of the action δIE =δIEH +δIGHY =
tf
Z
ti
dt[ ˙q δp+p δq˙−δH(q, p)]
I Bulk: Hamilton EOM
I Boundary (recallq|tf =∞):
δIE =p δq|tf −p δq|ti =p δq
|{z}
finite!
|tf 6= 0.
Not all variations preserving the boundary conditionqf =∞ lead to vanishing first variation of the actionIE!
SERIOUS PROBLEM!
D. Grumiller — Hamilton-Jacobi counterterms Statement of problem: divergences and variational principle 7/26
Another Problem Arises...
Step 4: Check on-shell action
Classical approximation to partition function:
Z = Z
DqDp exp
−1
~IE[q, p]
∼exp
−1
~IE[qcl, pcl]
To leading order determined by on-shell action IE[qcl, pcl] :=IE|EOM.
In our case (p(tf → ∞)→v= const.): IE|EOM=
∞
Z
ti
dt
pq˙−p2 2 − 1
q2
= v2 2
∞
Z
ti
dt+O(1)→ ∞
The on-shell action IE|EOM diverges!
PROBLEM IN THERMODYNAMICAL APPLICATIONS!
Another Problem Arises...
Step 4: Check on-shell action
Classical approximation to partition function:
Z = Z
DqDp exp
−1
~IE[q, p]
∼exp
−1
~IE[qcl, pcl]
To leading order determined by on-shell action IE[qcl, pcl] :=IE|EOM. In our case (p(tf → ∞)→v= const.):
IE|EOM=
∞
Z
ti
dt
pq˙−p2 2 − 1
q2
= v2 2
∞
Z
ti
dt+O(1)→ ∞
The on-shell action IE|EOM diverges!
PROBLEM IN THERMODYNAMICAL APPLICATIONS!
D. Grumiller — Hamilton-Jacobi counterterms Statement of problem: divergences and variational principle 8/26
Another Problem Arises...
Step 4: Check on-shell action
Classical approximation to partition function:
Z = Z
DqDp exp
−1
~IE[q, p]
∼exp
−1
~IE[qcl, pcl]
To leading order determined by on-shell action IE[qcl, pcl] :=IE|EOM. In our case (p(tf → ∞)→v= const.):
IE|EOM=
∞
Z
ti
dt
pq˙−p2 2 − 1
q2
= v2 2
∞
Z
ti
dt+O(1)→ ∞
The on-shell action IE|EOM diverges!
PROBLEM IN THERMODYNAMICAL APPLICATIONS!
Another Problem Arises...
Step 4: Check on-shell action
Classical approximation to partition function:
Z = Z
DqDp exp
−1
~IE[q, p]
∼exp
−1
~IE[qcl, pcl]
To leading order determined by on-shell action IE[qcl, pcl] :=IE|EOM. In our case (p(tf → ∞)→v= const.):
IE|EOM=
∞
Z
ti
dt
pq˙−p2 2 − 1
q2
= v2 2
∞
Z
ti
dt+O(1)→ ∞
The on-shell action IE|EOM diverges!
PROBLEM IN THERMODYNAMICAL APPLICATIONS!
D. Grumiller — Hamilton-Jacobi counterterms Statement of problem: divergences and variational principle 8/26
Outline
Statement of problem: divergences and variational principle
Solution: Hamilton-Jacobi counterterm
Applications: Black Hole Thermodynamics, AdS/CFT, ...
Summary and Literature
Exploit Freedom to Subtract Further Boundary Terms
Step 5: Ansatz for counterterm
Key observation: Dirichlet boundary problem not changed under IE →Γ =IE−ICT =IEH+IGHY −ICT with
ICT = S(q, t)|tf
Improved action: Γ =
tf
Z
ti
dt
−pq˙ −p2 2 − 1
q2
+ pq|ttf
i − S(q, t)|tf
First variation:
δΓ =
p−∂S(q, t)
∂q
δq
tf
= 0?
D. Grumiller — Hamilton-Jacobi counterterms Solution: Hamilton-Jacobi counterterm 10/26
Exploit Freedom to Subtract Further Boundary Terms
Step 5: Ansatz for counterterm
Key observation: Dirichlet boundary problem not changed under IE →Γ =IE−ICT =IEH+IGHY −ICT with
ICT = S(q, t)|tf Improved action:
Γ =
tf
Z
ti
dt
−pq˙ −p2 2 − 1
q2
+ pq|ttf
i −S(q, t)|tf
First variation:
δΓ =
p−∂S(q, t)
∂q
δq
tf
= 0?
Exploit Freedom to Subtract Further Boundary Terms
Step 5: Ansatz for counterterm
Key observation: Dirichlet boundary problem not changed under IE →Γ =IE−ICT =IEH+IGHY −ICT with
ICT = S(q, t)|tf Improved action:
Γ =
tf
Z
ti
dt
−pq˙ −p2 2 − 1
q2
+ pq|ttf
i −S(q, t)|tf
First variation:
δΓ =
p−∂S(q, t)
∂q
δq
tf
= 0?
D. Grumiller — Hamilton-Jacobi counterterms Solution: Hamilton-Jacobi counterterm 10/26
Constructing the Counterterm
Step 6: Solve Hamilton-Jacobi equation
Need a function S(q, t) of the boundary data qf,tf that obeys asymptotically
p= ∂S(q, t)
∂q
Hamilton’s principal function in the Hamilton-Jacobi equation
∂S(q, t)
∂t +H
q,∂S(q, t)
∂q
= 0 has precisely this property!
Thus we postulate:
The counterterm S(q, t) is a solution of the Hamilton-Jacobi equation!
Does it work?
Constructing the Counterterm
Step 6: Solve Hamilton-Jacobi equation
Need a function S(q, t) of the boundary data qf,tf that obeys asymptotically
p= ∂S(q, t)
∂q
Hamilton’s principal function in the Hamilton-Jacobi equation
∂S(q, t)
∂t +H
q,∂S(q, t)
∂q
= 0 has precisely this property!
Thus we postulate:
The counterterm S(q, t) is a solution of the Hamilton-Jacobi equation!
Does it work?
D. Grumiller — Hamilton-Jacobi counterterms Solution: Hamilton-Jacobi counterterm 11/26
Constructing the Counterterm
Step 6: Solve Hamilton-Jacobi equation
Need a function S(q, t) of the boundary data qf,tf that obeys asymptotically
p= ∂S(q, t)
∂q
Hamilton’s principal function in the Hamilton-Jacobi equation
∂S(q, t)
∂t +H
q,∂S(q, t)
∂q
= 0 has precisely this property!
Thus we postulate:
The counterterm S(q, t) is a solution of the Hamilton-Jacobi equation!
Does it work?
Constructing the Counterterm
Step 6: Solve Hamilton-Jacobi equation
Need a function S(q, t) of the boundary data qf,tf that obeys asymptotically
p= ∂S(q, t)
∂q
Hamilton’s principal function in the Hamilton-Jacobi equation
∂S(q, t)
∂t +H
q,∂S(q, t)
∂q
= 0 has precisely this property!
Thus we postulate:
The counterterm S(q, t) is a solution of the Hamilton-Jacobi equation!
Does it work?
D. Grumiller — Hamilton-Jacobi counterterms Solution: Hamilton-Jacobi counterterm 11/26
Final Result for the Improved Action
Step 7: Check that everything works
For our exampleH(q, p) = p22 +q12 we obtain (∆+:= 12(1 +p
1−8t2/q4))
S(q, t) = q2 2t
p4∆+−8t2/q4−∆+
+√
2 arctan 1
pq4∆+/(2t2)−1
Asymptotic expansion: S(q, t) = q2
2t +O(t/q2) = v2
2 t+O(1/t) Check on-shell action:
Γ|EOM= lim
tf→∞
v2 2
tf
Z
ti
dt− v2
2 tf +O(1) =O(1)
→OK!
Check first variation: δΓ|EOM =
p−∂S(q, t)
∂q
δq
tf
= O(1/tf)δq|tf = 0
→OK!
Final Result for the Improved Action
Step 7: Check that everything works
For our exampleH(q, p) = p22 +q12 we obtain (∆+:= 12(1 +p
1−8t2/q4))
S(q, t) = q2 2t
p4∆+−8t2/q4−∆+
+√
2 arctan 1
pq4∆+/(2t2)−1 Asymptotic expansion:
S(q, t) = q2
2t +O(t/q2) = v2
2 t+O(1/t)
Check on-shell action: Γ|EOM= lim
tf→∞
v2 2
tf
Z
ti
dt− v2
2 tf +O(1) =O(1)
→OK!
Check first variation: δΓ|EOM =
p−∂S(q, t)
∂q
δq
tf
= O(1/tf)δq|tf = 0
→OK!
D. Grumiller — Hamilton-Jacobi counterterms Solution: Hamilton-Jacobi counterterm 12/26
Final Result for the Improved Action
Step 7: Check that everything works
For our exampleH(q, p) = p22 +q12 we obtain (∆+:= 12(1 +p
1−8t2/q4))
S(q, t) = q2 2t
p4∆+−8t2/q4−∆+
+√
2 arctan 1
pq4∆+/(2t2)−1 Asymptotic expansion:
S(q, t) = q2
2t +O(t/q2) = v2
2 t+O(1/t) Check on-shell action:
Γ|EOM= lim
tf→∞
v2 2
tf
Z
ti
dt−v2
2 tf +O(1) =O(1)
→OK!
Check first variation: δΓ|EOM =
p−∂S(q, t)
∂q
δq
tf
= O(1/tf)δq|tf = 0
→OK!
Final Result for the Improved Action
Step 7: Check that everything works
For our exampleH(q, p) = p22 +q12 we obtain (∆+:= 12(1 +p
1−8t2/q4))
S(q, t) = q2 2t
p4∆+−8t2/q4−∆+
+√
2 arctan 1
pq4∆+/(2t2)−1 Asymptotic expansion:
S(q, t) = q2
2t +O(t/q2) = v2
2 t+O(1/t) Check on-shell action:
Γ|EOM= lim
tf→∞
v2 2
tf
Z
ti
dt−v2
2 tf +O(1) =O(1)→OK!
Check first variation: δΓ|EOM =
p−∂S(q, t)
∂q
δq
tf
= O(1/tf)δq|tf = 0
→OK!
D. Grumiller — Hamilton-Jacobi counterterms Solution: Hamilton-Jacobi counterterm 12/26
Final Result for the Improved Action
Step 7: Check that everything works
For our exampleH(q, p) = p22 +q12 we obtain (∆+:= 12(1 +p
1−8t2/q4))
S(q, t) = q2 2t
p4∆+−8t2/q4−∆+
+√
2 arctan 1
pq4∆+/(2t2)−1 Asymptotic expansion:
S(q, t) = q2
2t +O(t/q2) = v2
2 t+O(1/t) Check on-shell action:
Γ|EOM= lim
tf→∞
v2 2
tf
Z
ti
dt−v2
2 tf +O(1) =O(1)→OK!
Check first variation:
δΓ|EOM =
p−∂S(q, t) δq
tf
= O(1/tf)δq|tf = 0
→OK!
Final Result for the Improved Action
Step 7: Check that everything works
For our exampleH(q, p) = p22 +q12 we obtain (∆+:= 12(1 +p
1−8t2/q4))
S(q, t) = q2 2t
p4∆+−8t2/q4−∆+
+√
2 arctan 1
pq4∆+/(2t2)−1 Asymptotic expansion:
S(q, t) = q2
2t +O(t/q2) = v2
2 t+O(1/t) Check on-shell action:
Γ|EOM= lim
tf→∞
v2 2
tf
Z
ti
dt−v2
2 tf +O(1) =O(1)→OK!
Check first variation:
δΓ|EOM =
p−∂S(q, t)
∂q
δq
tf
= O(1/tf)δq|tf = 0→OK!
D. Grumiller — Hamilton-Jacobi counterterms Solution: Hamilton-Jacobi counterterm 12/26
Summary and Recipe
What we have learned so far...
I Start with bulk action IEH
→ Einstein-Hilbert
I Check consistency of boundary value problem
I If necessary, add boundary term IGHY
→ Gibbons-Hawking-York
I Check consistency of variational principle
I If necessary, subtract Hamilton-Jacobi counterterm ICT
→ ???
I Use improved action
Γ =IEH +IGHY −ICT for applications!
Examples where this is of relevance?
I Black hole thermodynamics from Euclidean path integral
I Holographic renormalization in AdS/CFT
Summary and Recipe
What we have learned so far...
I Start with bulk action IEH
→ Einstein-Hilbert
I Check consistency of boundary value problem
I If necessary, add boundary term IGHY
→ Gibbons-Hawking-York
I Check consistency of variational principle
I If necessary, subtract Hamilton-Jacobi counterterm ICT
→ ???
I Use improved action
Γ =IEH +IGHY −ICT for applications!
Examples where this is of relevance?
I Black hole thermodynamics from Euclidean path integral
I Holographic renormalization in AdS/CFT
D. Grumiller — Hamilton-Jacobi counterterms Solution: Hamilton-Jacobi counterterm 13/26
Summary and Recipe
What we have learned so far...
I Start with bulk action IEH
→ Einstein-Hilbert
I Check consistency of boundary value problem
I If necessary, add boundary term IGHY
→ Gibbons-Hawking-York
I Check consistency of variational principle
I If necessary, subtract Hamilton-Jacobi counterterm ICT
→ ???
I Use improved action
Γ =IEH +IGHY −ICT for applications!
Examples where this is of relevance?
I Black hole thermodynamics from Euclidean path integral
I Holographic renormalization in AdS/CFT
Summary and Recipe
What we have learned so far...
I Start with bulk action IEH
→ Einstein-Hilbert
I Check consistency of boundary value problem
I If necessary, add boundary term IGHY
→ Gibbons-Hawking-York
I Check consistency of variational principle
I If necessary, subtract Hamilton-Jacobi counterterm ICT
→ ???
I Use improved action
Γ =IEH +IGHY −ICT for applications!
Examples where this is of relevance?
I Black hole thermodynamics from Euclidean path integral
I Holographic renormalization in AdS/CFT
D. Grumiller — Hamilton-Jacobi counterterms Solution: Hamilton-Jacobi counterterm 13/26
Summary and Recipe
What we have learned so far...
I Start with bulk action IEH
→ Einstein-Hilbert
I Check consistency of boundary value problem
I If necessary, add boundary term IGHY
→ Gibbons-Hawking-York
I Check consistency of variational principle
I If necessary, subtract Hamilton-Jacobi counterterm ICT
→ ???
I Use improved action
Γ =IEH +IGHY −ICT for applications!
Examples where this is of relevance?
I Black hole thermodynamics from Euclidean path integral
I Holographic renormalization in AdS/CFT
Summary and Recipe
What we have learned so far...
I Start with bulk action IEH
→ Einstein-Hilbert
I Check consistency of boundary value problem
I If necessary, add boundary term IGHY
→ Gibbons-Hawking-York
I Check consistency of variational principle
I If necessary, subtract Hamilton-Jacobi counterterm ICT
→ ???
I Use improved action
Γ =IEH +IGHY −ICT for applications!
Examples where this is of relevance?
I Black hole thermodynamics from Euclidean path integral
I Holographic renormalization in AdS/CFT
D. Grumiller — Hamilton-Jacobi counterterms Solution: Hamilton-Jacobi counterterm 13/26
Summary and Recipe
What we have learned so far...
I Start with bulk action IEH
→ Einstein-Hilbert
I Check consistency of boundary value problem
I If necessary, add boundary term IGHY
→ Gibbons-Hawking-York
I Check consistency of variational principle
I If necessary, subtract Hamilton-Jacobi counterterm ICT
→ ???
I Use improved action
Γ =IEH +IGHY −ICT for applications!
Examples where this is of relevance?
I Black hole thermodynamics from Euclidean path integral
I Holographic renormalization in AdS/CFT
Summary and Recipe
What we have learned so far...
I Start with bulk action IEH → Einstein-Hilbert
I Check consistency of boundary value problem
I If necessary, add boundary term IGHY → Gibbons-Hawking-York
I Check consistency of variational principle
I If necessary, subtract Hamilton-Jacobi counterterm ICT → ???
I Use improved action
Γ =IEH +IGHY −ICT for applications!
Examples where this is of relevance?
I Black hole thermodynamics from Euclidean path integral
I Holographic renormalization in AdS/CFT
D. Grumiller — Hamilton-Jacobi counterterms Solution: Hamilton-Jacobi counterterm 13/26
Outline
Statement of problem: divergences and variational principle
Solution: Hamilton-Jacobi counterterm
Applications: Black Hole Thermodynamics, AdS/CFT, ...
Summary and Literature
Black Hole Thermodynamics
Consider Euclidean path integral (Gibbons, Hawking, 1977)
Z = Z
Dg exp
−1
~IE[g]
I g: metric; IE[g]: Euclidean action
I Semiclassical limit (~→0): dominated by classical solutions
I Exploit relationship between Z and Euclidean partition function Z ∼e−βΩ
I Ω: thermodynamic potential for appropriate ensemble
I β: periodicity in Euclidean time
I Derive thermodynamical quantities from Ω(β, . . .) Requires periodicity in Euclidean time and accessibility of (semi-)classical approximation
D. Grumiller — Hamilton-Jacobi counterterms Applications: Black Hole Thermodynamics, AdS/CFT, ... 15/26
Black Hole Thermodynamics
Consider Euclidean path integral (Gibbons, Hawking, 1977)
Z = Z
Dg exp
−1
~IE[g]
∼exp
−1
~IE[gcl]
I g: metric; IE[g]: Euclidean action
I Semiclassical limit (~→0): dominated by classical solutions
I Exploit relationship between Z and Euclidean partition function Z ∼e−βΩ
I Ω: thermodynamic potential for appropriate ensemble
I β: periodicity in Euclidean time
I Derive thermodynamical quantities from Ω(β, . . .) Requires periodicity in Euclidean time and
Black Hole Thermodynamics
Consider Euclidean path integral (Gibbons, Hawking, 1977)
Z = Z
Dg exp
−1
~IE[g]
∼exp
−1
~IE[gcl]
Semi-classical approximation typically is not accessible for bulk action!
First variation of action not zero for all field configurations contributing to path integral due to boundary terms
δIE EOM∼
Z
∂M
ddx√ γ h
πabδγabi 6= 0
SAME PROBLEM AS IN QUANTUM MECHANICS EXAMPLE!
D. Grumiller — Hamilton-Jacobi counterterms Applications: Black Hole Thermodynamics, AdS/CFT, ... 15/26
Black Hole Thermodynamics
Consider Euclidean path integral (Gibbons, Hawking, 1977)
Z = Z
Dg exp
−1
~IE[g]
∼exp
−1
~IE[gcl]
In view of quantum mechanics example: Which action to take forIE?
I Naive attempt: IE[g] =IEH
Problem: IEH = 0 on classical solutions! ThusZ = 1 ???
I Less naive attempt: IE[g] =IEH +IGHY
Problem: typicallyIGHY → −∞ asymptotically! ThusZ → ∞ ???
I More sophisticated attempt: IE[g] = Γ =IEH+IGHY −ICT, ICT = “∞00
Problem 1: Finite ambiguity remains!
Problem 2: Sometimes IE → ∞ correct physical result! Need systematic way to constructICT!
Black Hole Thermodynamics
Consider Euclidean path integral (Gibbons, Hawking, 1977)
Z = Z
Dg exp
−1
~IE[g]
∼exp
−1
~IE[gcl]
In view of quantum mechanics example: Which action to take forIE?
I Naive attempt: IE[g] =IEH
Problem: IEH = 0 on classical solutions! ThusZ = 1 ???
I Less naive attempt: IE[g] =IEH +IGHY
Problem: typicallyIGHY → −∞ asymptotically! ThusZ → ∞ ???
I More sophisticated attempt: IE[g] = Γ =IEH+IGHY −ICT, ICT = “∞00
Problem 1: Finite ambiguity remains!
Problem 2: Sometimes IE → ∞ correct physical result! Need systematic way to constructICT!
D. Grumiller — Hamilton-Jacobi counterterms Applications: Black Hole Thermodynamics, AdS/CFT, ... 15/26
Black Hole Thermodynamics
Consider Euclidean path integral (Gibbons, Hawking, 1977)
Z = Z
Dg exp
−1
~IE[g]
∼exp
−1
~IE[gcl]
In view of quantum mechanics example: Which action to take forIE?
I Naive attempt: IE[g] =IEH
Problem: IEH = 0 on classical solutions! ThusZ = 1 ???
I Less naive attempt: IE[g] =IEH +IGHY
Problem: typicallyIGHY → −∞asymptotically! Thus Z → ∞ ???
I More sophisticated attempt: IE[g] = Γ =IEH+IGHY −ICT, ICT = “∞00
Problem 1: Finite ambiguity remains!
Problem 2: Sometimes IE → ∞ correct physical result! Need systematic way to constructICT!
Black Hole Thermodynamics
Consider Euclidean path integral (Gibbons, Hawking, 1977)
Z = Z
Dg exp
−1
~IE[g]
∼exp
−1
~IE[gcl]
In view of quantum mechanics example: Which action to take forIE?
I Naive attempt: IE[g] =IEH
Problem: IEH = 0 on classical solutions! ThusZ = 1 ???
I Less naive attempt: IE[g] =IEH +IGHY
Problem: typicallyIGHY → −∞asymptotically! Thus Z → ∞ ???
I More sophisticated attempt: IE[g] = Γ =IEH+IGHY −ICT, ICT = “∞00
Problem 1: Finite ambiguity remains!
Problem 2: Sometimes IE → ∞ correct physical result!
Need systematic way to constructICT!
D. Grumiller — Hamilton-Jacobi counterterms Applications: Black Hole Thermodynamics, AdS/CFT, ... 15/26
Black Hole Thermodynamics
Consider Euclidean path integral (Gibbons, Hawking, 1977)
Z = Z
Dg exp
−1
~IE[g]
In view of quantum mechanics example: Which action to take forIE?
I Naive attempt: IE[g] =IEH
Problem: IEH = 0 on classical solutions! ThusZ = 1 ???
I Less naive attempt: IE[g] =IEH +IGHY
Problem: typicallyIGHY → −∞asymptotically! Thus Z → ∞ ???
I More sophisticated attempt: IE[g] = Γ =IEH+IGHY −ICT, ICT = “∞00
Problem 1: Finite ambiguity remains!
Problem 2: Sometimes IE → ∞ correct physical result!
Need systematic way to constructI !
Example: 2D dilaton gravity
Same problem – same solution!
Bulk action of 2D dilaton gravity:
Γ =− 1 16πG2
Z
M
d2x√ g
X R−U(X) (∇X)2−2V(X)
− 1 8πG2
Z
∂M
dx√
γ X K− 1 8πG2
Z
∂M
dx√
γ S(X,∇kX, γ,∇kγ)
D. Grumiller — Hamilton-Jacobi counterterms Applications: Black Hole Thermodynamics, AdS/CFT, ... 16/26
Example: 2D dilaton gravity
Same problem – same solution!
Bulk action of 2D dilaton gravity:
Γ =− 1 16πG2
Z
M
d2x√ g
X R−U(X) (∇X)2−2V(X)
− 1 8πG2
Z
∂M
dx√
γ X K− 1 8πG2
Z
∂M
dx√
γ S(X,∇kX, γ,∇kγ)
I Dilaton X defined via coupling to Ricci scalar R
I Model specified by kinetic and potential functions for dilaton
I Dilaton gravity analog of Gibbons-Hawking-Yorkboundary term: coupling of X to trace of extrinsic curvature K of (∂M, γ)
I Boundary value problem: fix X and induced metric γ at∂M For later: Q(X) :=RX
U, w(X) :=RX
eQV
Example: 2D dilaton gravity
Same problem – same solution!
Bulk action of 2D dilaton gravity:
Γ =− 1 16πG2
Z
M
d2x√ g
X R−U(X)(∇X)2−2V(X)
− 1 8πG2
Z
∂M
dx√
γ X K− 1 8πG2
Z
∂M
dx√
γ S(X,∇kX, γ,∇kγ)
I
I Dilaton X defined via coupling to Ricci scalar R
I Model specified by kineticandpotential functions for dilaton
I Dilaton gravity analog of Gibbons-Hawking-Yorkboundary term: coupling of X to trace of extrinsic curvature K of (∂M, γ)
I Boundary value problem: fix X and induced metric γ at∂M
For later: Q(X) :=RX
U, w(X) :=RX
eQV
D. Grumiller — Hamilton-Jacobi counterterms Applications: Black Hole Thermodynamics, AdS/CFT, ... 16/26
Example: 2D dilaton gravity
Same problem – same solution!
Classical action of 2D dilaton gravity:
Γ =− 1 16πG2
Z
M
d2x√ g
X R−U(X) (∇X)2−2V(X)
− 1 8πG2
Z
∂M
dx√ γX K
− 1 8πG2
Z
∂M
dx√
γ S(X,∇kX, γ,∇kγ)
I
I Dilaton X defined via coupling to Ricci scalar R
I Model specified by kinetic and potential functions for dilaton
I Dilaton gravity analog of Gibbons-Hawking-Yorkboundary term:
coupling of X to trace of extrinsic curvature K of (∂M, γ)
I Boundary value problem: fix X and induced metric γ at∂M
X X
Example: 2D dilaton gravity
Same problem – same solution!
Improved action of 2D dilaton gravity:
Γ =− 1 16πG2
Z
M
d2x√ g
X R−U(X) (∇X)2−2V(X)
− 1 8πG2
Z
∂M
dx√
γ XK− 1 8πG2
Z
∂M
dx√
γ S(X,∇kX, γ,∇kγ)
I
I Dilaton X defined via coupling to Ricci scalar R
I Model specified by kinetic and potential functions for dilaton
I Dilaton gravity analog of Gibbons-Hawking-Yorkboundary term:
coupling of X to trace of extrinsic curvature K of (∂M, γ)
I Boundary value problem: fix X and induced metric γ at∂M For later: Q(X) :=RX
U, w(X) :=RX
eQV
D. Grumiller — Hamilton-Jacobi counterterms Applications: Black Hole Thermodynamics, AdS/CFT, ... 16/26
Example: 2D dilaton gravity
Same problem – same solution!
Improved action of 2D dilaton gravity:
Γ =− 1 16πG2
Z
M
d2x√ g
X R−U(X) (∇X)2−2V(X)
− 1 8πG2
Z
∂M
dx√
γ X K− 1 8πG2
Z
∂M
dx√
γ S(X,∇kX, γ,∇kγ)
I
I Dilaton X defined via coupling to Ricci scalar R
I Model specified by kinetic and potential functions for dilaton
I Dilaton gravity analog of Gibbons-Hawking-Yorkboundary term:
coupling of X to trace of extrinsic curvature K of (∂M, γ)
I Boundary value problem: fix X and induced metric γ at∂M
I Generic Ansatz for countertermS
X X
Example: 2D dilaton gravity
Same problem – same solution!
Improved action of 2D dilaton gravity:
Γ =− 1 16πG2
Z
M
d2x√ g
X R−U(X) (∇X)2−2V(X)
− 1 8πG2
Z
∂M
dx√
γ X K− 1 8πG2
Z
∂M
dx√
γ S(X,∇kX)
I Dilaton X defined via coupling to Ricci scalar R
I Model specified by kinetic and potential functions for dilaton
I Dilaton gravity analog of Gibbons-Hawking-Yorkboundary term:
coupling of X to trace of extrinsic curvature K of (∂M, γ)
I Boundary value problem: fix X and induced metric γ at∂M
I Diffeomorphism invariance along boundary simplifies countertermS For later: Q(X) :=RX
U, w(X) :=RX
eQV
D. Grumiller — Hamilton-Jacobi counterterms Applications: Black Hole Thermodynamics, AdS/CFT, ... 16/26
Example: 2D dilaton gravity
Same problem – same solution!
Improved action of 2D dilaton gravity:
Γ =− 1 16πG2
Z
M
d2x√ g
X R−U(X) (∇X)2−2V(X)
− 1 8πG2
Z
∂M
dx√
γ X K− 1 8πG2
Z
∂M
dx√ γ S(X)
I Dilaton X defined via coupling to Ricci scalar R
I Model specified by kinetic and potential functions for dilaton
I Dilaton gravity analog of Gibbons-Hawking-Yorkboundary term:
coupling of X to trace of extrinsic curvature K of (∂M, γ)
I Boundary value problem: fix X and induced metric γ at∂M
I Assumption about boundary ∂Mfurther simplifies counterterm S
X X
Example: 2D dilaton gravity
Same problem – same solution!
Improved action of 2D dilaton gravity:
Γ =− 1 16πG2
Z
M
d2x√ g
X R−U(X) (∇X)2−2V(X)
− 1 8πG2
Z
∂M
dx√
γ X K+ 1 8πG2
Z
∂M
dx√ γ
q
w(X)e−Q(X)
I Dilaton X defined via coupling to Ricci scalar R
I Model specified by kinetic and potential functions for dilaton
I Dilaton gravity analog of Gibbons-Hawking-Yorkboundary term:
coupling of X to trace of extrinsic curvature K of (∂M, γ)
I Boundary value problem: fix X and induced metric γ at∂M
I Solving Hamilton-Jacobi equation yields simple universal result!
For later: Q(X) :=RX
U, w(X) :=RX
eQV
D. Grumiller — Hamilton-Jacobi counterterms Applications: Black Hole Thermodynamics, AdS/CFT, ... 16/26
Example: 2D dilaton gravity
Same problem – same solution!
Improved action of 2D dilaton gravity:
Γ =− 1 16πG2
Z
M
d2x√ g
X R−U(X) (∇X)2−2V(X)
− 1 8πG2
Z
∂M
dx√
γ X K+ 1 8πG2
Z
∂M
dx√ γ
q
w(X)e−Q(X) Properties of improved action:
I On-shell actionΓ|EOM finite
I δΓ = 0 for all variations that preserve boundary conditions
I Counter-termis (essentially) unique
I Counter-termis universal and applies to different asymptotics! Prime application: Black hole thermodynamics!
Example: 2D dilaton gravity
Same problem – same solution!
Improved action of 2D dilaton gravity:
Γ =− 1 16πG2
Z
M
d2x√ g
X R−U(X) (∇X)2−2V(X)
− 1 8πG2
Z
∂M
dx√
γ X K+ 1 8πG2
Z
∂M
dx√ γ
q
w(X)e−Q(X) Properties of improved action:
I On-shell actionΓ|EOM finite
I δΓ = 0 for all variations that preserve boundary conditions
I Counter-termis (essentially) unique
I Counter-termis universal and applies to different asymptotics!
Prime application: Black hole thermodynamics!
D. Grumiller — Hamilton-Jacobi counterterms Applications: Black Hole Thermodynamics, AdS/CFT, ... 16/26
Example: 2D dilaton gravity
Same problem – same solution!
Improved action of 2D dilaton gravity:
Γ =− 1 16πG2
Z
M
d2x√ g
X R−U(X) (∇X)2−2V(X)
− 1 8πG2
Z
∂M
dx√
γ X K+ 1 8πG2
Z
∂M
dx√ γ
q
w(X)e−Q(X) Properties of improved action:
I On-shell actionΓ|EOM finite
I δΓ = 0 for all variations that preserve boundary conditions
I Counter-termis (essentially) unique
I Counter-termis universal and applies to different asymptotics!
Prime application: Black hole thermodynamics!