Integrating out geometry in 2D dilaton gravity coupled to fermions
René Meyer
<rene.meyer@itp.uni-leipzig.de>
Outline
1. Introduction 2. The Action
3. Constraint algebra 4. BRST gauge xing
5. Integrating out Geometry
6. Outlook
1. Introduction
• GOAL: Use Path Integral to quantize gravity in D=2 + scalar eld X (Dilaton) + Fermions
S (dil) = Z
d 2 x √
−g
X R
2 − U (X )
2 (∇X ) 2 + V (X )
+ S m
• Arise from compactications, spherical reduction and string theory
• 1980s: Jackiw, Teitelboim, D'Hoker, Katanaev, Volovich & others
• Exact path integration of geometry: Pioneering work by W. Kummer, H. Liebl
& D. Vassilevich 1996-1998
Literature
• Quantum gravity: S. Carlip gr-qc/0108040
• Review on Dilaton Gravity: D. Grumiller, W. Kummer, D. Vassilevich hep-th/0204253
• W. Kummer, H. Liebl, D. Vassilevich: gr-qc/9612012, hep-th/9707115, hep-th/9809168
• D. Grumiller's PhD Thesis: gr-qc/0105078
2. Action
S = S F OG + S kin + S SI S F OG =
Z
M
2X a (De) a + Xdω + (U (X )X + X − + V (X ))
S kin = i Z
M
2f (X ) e a ∧ (χγ a dχ) + h.c.c.
S SI = 1 2
Z
M
2e b ∧ e a ab h(X )g(χχ)
= Z
dx 2 (e)h(X )g(χχ)
Remarks
• f (X ) , h(X ) and g(χχ) are real functions (hermiticity of S )
• Functions U , V (and f , h , g ) specify the model, e.g. D=4 SRG:
U (X ) = −1/2X , V (X ) = −λ 2
• χ i , χ † i are anticommuting, thus the general form of a polynomial self interaction is
g(χχ) = c + mχχ + e(χχ) 2
and c can be absorbed into V(X).
3. Constraint Algebra - Primary constraints
1st class constraints: p i := ∂ L
∂ q ˙
i= (0, 0, 0) The P i := ∂
LL
∂ Q ˙
igive rise to 2nd class constraints:
Φ 0 = P 0 + i
√ 2 f (p 1 )q 3 Q 2 Φ 1 = P 1 − i
√ 2 f (p 1 )q 2 Q 3 Φ 2 = P 2 + i
√ 2 f (p 1 )q 3 Q 0 Φ 3 = P 3 − i
√ f (p 1 )q 2 Q 1
Secondary Constraints
We pass to the Dirac bracket formalism. The time evolution of the primary 1st class constraints G i := ˙ p i = {p i , H 0 } ∗ = {p i , H 0 } s ≈ 0 gives rise to secondary 1st class constraints
G 1 = G g 1
G 2 = G g 2 + i
√ 2 f (X )(χ † 1 ← →
∂ 1 χ 1 ) + e + 1 h(X )g(χχ) G 3 = G g 3 − i
√ 2 f (X )(χ † 0 ← →
∂ 1 χ 0 ) − e − 1 h(X )g(χχ) G g 1 = ∂ 1 X + X − e + 1 − X + e − 1
G g 2 = ∂ 1 X + + ω 1 X + − e + 1 V
G g 3 = ∂ 1 X − − ω 1 X − + e − 1 V
The Hamiltonian
The Hamiltonian can be written as a sum over the secondary constraints
H = −q i G i ≈ 0
and thus to check the absence of ternary constraints one has to explore the algebra of the G i
{G i , H 0 } ∗ = − q j 0 {G i , G 0 j } ∗ ≈ 0
Algebra of secondary constraints
{G i , G 0 i } ∗ = 0 i = 1, 2, 3 {G 1 , G 0 2 } ∗ = −G 2 δ
{G 1 , G 0 3 } ∗ = G 3 δ {G 2 , G 0 3 } ∗ =
"
−
3
X
i=1
dV
dp i G i +
gh 0 − h
f f 0 g 0 · (χχ)
G 1
#
δ
Remarks on the additional term
• Closes with δ -functions → Lie algebra with structure functions
• It vanishes for minimal coupling, i.e. for h = f = const.
• If f ∝ h , a mass term mχχ doesn't contribute, but a Thirring term does
∝ h 0 (g − g 0 · (χχ)) = −h 0 e(χχ) 2
• The new contribution does not contain derivatives of χ , opposing to the scalar
case, where ∝ f f
0L scalar .
4. BRST gauge xing
• BRST charge: Ω 2 . := {Ω, {Ω, .} ∗ } ∗ = 0 ⇐⇒ {Ω, J.I. Ω} ∗ = 0
• Ghosts for the G i : (c i , p c i ) s.t. {c i (x), p c j (x 0 )} s = −δ j i δ(x − x 0 )
• Result:
Ω = Ω (1) + Ω (2) Ω (1) = c i G i
Ω (2) = 1
2 c i c j C ij k p c k
Gauge xed Hamiltonian
Gauge xing fermion: Ψ = p c 2 Gauge xed Hamiltonian:
H gf =
=0
z }| {
H BRST +{Ω, Ψ} ∗
= −G 2 − C 2i k c i p c k
= −G 2 − c 1 p c 2 − C 23 k c 3 p c k
⇒ Eddington-Finkelstein gauge (ω 0 , e − 0 , e + 0 ) = (0, 1, 0)
5. Integrating out Geometry
Path integral: Z [J, j, S ] = N R
Dµ[Q, P, q, p, c, p c ]e i
R d
2x(L
gf+L
src)
Measure: Dµ[Q, P, q, p, c, p c ] =
3
Q
i=0
DQ i Dµ 0 [q, p]
3
Q
i=0
DP i δ(Φ i )
3
Q
j=1
Dc j Dp c j
Lagrangian:
L gf = p i q ˙ i + ˙ Q i P i + p c i c i − H gf
= p i q ˙ i + ˙ Q i P i + G 2 + p c k M l k c l
L src = J i p i + j i q i + S i Q i
Ghost integration
M =
∂ 0 0 ∂p ∂V
1
−
gh 0 − h f f 0 g 0 · (χχ)
−1 ∂ 0 ∂p ∂V
2
0 0 ∂ 0 + ∂p ∂ V
3
Z 3 Y
j =1
Dc j Dp c j exp{i Z
d 2 xp c k M l k c l } ∝ DetM = Det(∂ 0 2 (∂ 0 + U (p 1 )p 2 ))
will be cancelled later
Matter Momenta
Φ i linear in P , Q & δ(Φ i ) in the path integral ⇒ P i -Integration trivial
⇒ L (1) ef f = p i q ˙ i + G 2 + i
√ 2 f (p 1 ) h
q 3 (Q 0 ← →
∂ 0 Q 2 ) − q 2 (Q 1 ← →
∂ 0 Q 3 ) i
+ L src G 2 = G g 2 + i
√ 2 f (p 1 )(Q 3 ← →
∂ 1 Q 3 ) + q 3 h(p 1 )g(χχ) G g 2 = ∂ 1 p 2 + q 1 p 2 − q 3 V
→ linear in q i , thus their integration is trivial too, giving Delta functionals
Covariant path integral measure for fermions (1)
2nd class constraints → local measure in path integral [Henneaux/Teitelboim Ch. 16;
Henneaux/Slavnov hep-th/9406161]
q
sdet{Φ i , Φ j } s = 2f 2 (p 1 )|q 2 q 3 |
Dirac fermions on a curved, but xed background (in D=2) [D.J. Toms Phys. Rev. D 35, 3796 (1987); M. Basler Fortsch.Phys.41:1-43,1993]
Y
i=0,1
Dχ † i Dχ i Y
x
[−g(x)] −1 = Y
i=0,1
Dχ † i Dχ i Y
x
q 3 −2
→ change the measure by hand [Kummer, Liebl, Vassilevich hep-th/9707115 for scalar elds]
Covariant path integral measure for fermions (2)
→ Trick to get rid of the measure factor [Kummer, Liebl, Vassilevich hep-th/9809168]
Z [J, j, S ] = N Z
Df f −2 δ
f − 1 i
δ δj 3
Z ˜ [J, j, S ]
Z ˜ [J, j, S ] =
Z Y
i=0,1
Dχ † i χ i
3
Y
i=1
Dp i Dq i detM e i
R d
2xL
(1)ef f
Integrating the q i
→ Delta functionals give system of PDEs
0 = ∂ 0 p 1 − p 2 − j 1 (1)
0 = ∂ 0 p 2 − j 2 + i
√ 2 f (p 1 )(Q 1 ← →
∂ 0 Q 3 ) (2)
0 = (∂ 0 + U (p 1 )p 2 )p 3 − j 3 − i
√ 2 f (p 1 )(Q 0 ← →
∂ 0 Q 2 ) − h(p 1 )g(χχ) + V (p 1 ) (3)
→ det M cancels during p i integration
Solving the PDEs (1)
Minimal coupling f (p 1 ) ≡ f = const.
p 1 = B 1 = ∇ −1 0 (j 1 + B 2 ) + ˜ p 1 (x 1 ) p 2 = B 2 = ∇ −1 0
j 2 − f i
√ 2 (Q 1 ← →
∂ 0 Q 3 )
+ ˜ p 2 (x 1 ) p 3 = B 3 = e −Q
∇ −1 0 e Q
j 3 − V (B 1 ) + h(B 1 )g(χχ) + f i
√ 2 (Q 0 ← →
∂ 0 Q 2 )
+ ˜ p 3 (x 1 )
Q := ∇ −1 0 (U (B 1 )B 2 )
Solving the PDEs (2)
General case → Weak matter approximation, i.e. p i =
∞
P
n=0
p (n) i , i = 1, 2 and p (n) 1 (Q 1 ← →
∂ 0 Q 3 )
| {z }
(n+1)th order
→ recursion for e.g. f (p 1 ) = p 1
p (0) 2 = ∇ −1 0 j 2 + ˜ p (0) 2 (x 1 )
p (0) 1 = ∇ −1 0 (j 1 + p (0) 2 ) + ˜ p (0) 1 (x 1 ) p (n) 2 = − i
√ 2 ∇ −1 0
p (n−1) 1 (Q 1 ← →
∂ 0 Q 3 )
+ ˜ p (n) 2 (x 1 ) n ≥ 1
p (n) 1 = ∇ −1 0 p (n) 2 + ˜ p (n) 1 (x 1 )
Eective action
L (2) ef f = J i B ˆ i + S i Q i + i
√ 2 f ( ˆ B 1 )(Q 3 ← →
∂ 1 Q 1 ) + L amb L amb = g ˜ 3 e Q ˆ
j 3 − V ( ˆ B 1 ) + h( ˆ B 1 )g(χχ) + i
√ 2 f ( ˆ B 1 )(Q 0 ← →
∂ 0 Q 2 )
Z [J, j = 0; S ] = N R Q 3
i=0
DQ i [˜ g 3 e Q ˆ ] −2 e i
R d
2xL
(2)ef f
J,j=0
→ propagators, vertices, perturbation theory, S-Matrix elements
6. Outlook
• "Bosonization": g = mχχ + e(χχ) 2 , f ∝ h
→ {G 2 , G 3 } ∗ = − ∂p ∂ V
i
G i − h 0 e(χχ) 2 G 1
→ {G 2 , G 2 } scalar p = − ∂p ∂V
i