r
X
i
v
:
0
8
0
7
.
0
9
7
7
v
1
[
h
e
p

t
h
]
7
J
u
l
2
0
0
8
ChernSimons gravity based
on a nonsemisimple group
C. DUVAL (
1
), Z. HORV
´
ATH (
2
) and P. A. HORV
´
ATHY (
3
)
Abstract. The gauge theoryformulation of stringmotivated lineal gravity proposed by Cangemi and Jackiw
is obtained by dimensional reduction from a (2 + 1) dimensional gravity with ChernSimons Lagrangian.
The ‘stringinspired’ gravity in (1 + 1) dimensions [1] admits a gaugetheory formulation [2]: starting
with the centrally extended Poincar´e algebra with 4 generators,
(0) [P
a
, J] = ǫ
b
a
P
b
, [P
a
, P
b
] = ǫ
ab
iΛI
(a, b = 1, 2), where I is the central element, the ‘Poincar´e’ gauge theory with connection form
(1) A = e
a
P
a
+ ωJ + a
1
2
iΛI
and Lagrangian
(2) L = η
A
F
A
= η
a
(De)
a
+ η
2
dω + η
3
(da +
1
2
e
a
ǫ
ab
e
b
), η
A
= (η
a
, η
2
, η
3
),
where the η’s are Lagrange multipliers, leads to the (1 + 1) dimensional gravitational equation [1]
(3) R = 0 and ∂
µ
∂
ν
η =
1
2
Λh
µν
(where h
µν
= diag(1, −1) and Λ is the cosmological constant), supplemented with the additional equation
(4) da +
1
2
e
a
ǫ
ab
e
b
= 0.
The algebra (1) has been subsequently extensively studied in connection of a recently proposed Wess
ZuminoWitten model [36].
In the conslusion of their paper [2], Cangemi and Jackiw wonder whether their model can be obtained
from a (2+1) dimensional model by reduction. The aim of this Letter is to show that this is indeed possible.
Using recent results on nonsemisimple algebras [6,7], we construct a (2 + 1)dimensional ChernSimons
Lagrangian whose reduction gives precisely the CangemiJackiw model. Other models are found in Ref.
[89].
Consider the extended Poincar´e algebra in 2 + 1 dimensions with nonvanishing commutators
(5)
[J
µ
, J
ν
] = ǫ
ρ
µν
J
ρ
, [J
µ
, P
ν
] = ǫ
ρ
µν
P
ρ
,
[J
µ
, T
ν
] = ǫ
ρ
µν
T
ρ
, [P
µ
, P
ν
] = ǫ
ρ
µν
T
ρ
,
(
1
) D´epartement de Physique, Universit´e d’AixMarseille II, and Centre de Physique Th´eorique CNRS–
Luminy, Case 907, F–13288 MARSEILLE Cedex 09 (France); email: duval@cpt.univmrs.fr
(
2
) Institute for Theoretical Physics, E¨ otv¨os University, H–1088 BUDAPEST, Puskin u. 57 (Hungary);
email: zalanh@ludens.elte.hu
(
3
) D´epartement de Math´ematiques, Universit´e de Tours, Parc de Grandmont, F–37200 TOURS (France);
email: horvathy@univtours.fr
2 Duval, Horv´ ath, Horv´ athy
(µ, ν, ρ = 0, 1, 2). Note that the T
µ
’s commute with the P
µ
’s and with each other. In these formulae,
the indices are raised and lowered through the Lorentz metric η
µν
= diag(−1, 1, 1). This algebra carries a
symmetric, nondegenerate bilinear form, namely
(6) Ω
AB
M
A
M
B
= P
µ
P
µ
+ J
µ
T
µ
+ T
µ
J
µ
.
(The other Casimir invariant, T
µ
T
ν
is degenerate since the group is non semisimple). The group generated
by the algebra (1.1) can be represented by upper triangular matrices [10]
(7)
_
_
_
U Uω U(
1
2
ω
2
+ a)
U Uω
U
_
_
_,
_
¸
¸
¸
¸
¸
¸
¸
_
¸
¸
¸
¸
¸
¸
¸
_
U =
_
u v
v
∗
u
∗
_
,
a =
_
a
2
i(a
0
−a
1
)
i(a
0
+ a
1
) −a
2
_
,
ω =
_
ω
2
i(ω
0
−ω
1
)
i(ω
0
+ ω
1
) −ω
2
_
.
Note that U ∈ SU(1, 1).
Consider the gauge theory in 2 + 1 dimensions given by the connection form
(8) A
α
= e
µ
α
P
µ
+ ω
µ
α
J
µ
+ a
µ
α
T
µ
,
where µ = 0, 1, 2 is an algebra index and α = 0, 1, 2 denote spacetime indices. Its curvature form is found
as
(9)
F
αβ
=
_
∂
α
e
ρ
β
−∂
β
e
ρ
α
+ ǫ
ρ
µν
ω
µ
α
e
ν
β
−ǫ
ρ
µν
ω
µ
β
e
ν
α
_
P
ρ
+
_
∂
α
ω
ρ
β
−∂
β
ω
ρ
α
+ ǫ
ρ
µν
ω
µ
α
ω
ν
β
_
J
ρ
+
_
∂
α
a
ρ
β
+ ǫ
ρ
µν
ω
µ
α
a
ν
β
−∂
β
a
ρ
α
−ǫ
ρ
µν
ω
µ
β
a
ν
α
+ ǫ
ρ
µν
e
µ
α
e
ν
β
_
T
ρ
.
The ﬁrst bracket here is the torsion and the second bracket is the curvature of the spin connection.
Now we want to postulate a ChernSimons gauge theory. The ChernSimons 3form is conveniently
found by considering the closed 4form
(10) (F, ∗F) = ǫ
αβγδ
Ω
AB
F
A
αβ
F
B
γδ
which is locally the divergence of the ChernSimons 3form, ∂
α
(Chern −Simons)
α
. Our ChernSimons gauge
theory is therefore given by the Lagrangian
(11) L =
1
2
ǫ
αβγ
_
_
∂
α
e
µ
β
e
ν
γ
+ ∂
α
ω
µ
β
a
ν
γ
+ ∂
α
a
µ
β
ω
ν
γ
_
η
µν
+ ǫ
µνρ
_
ω
µ
α
e
ν
β
e
ρ
γ
+ ω
µ
α
ω
ν
β
a
ρ
γ
_
_
.
The corresponding ﬁeld equations read
(12)
ǫ
αβγ
_
∂
α
e
ν
β
+ ǫ
ν
µρ
ω
µ
α
e
ρ
β
_
= 0,
ǫ
αβγ
_
∂
α
ω
ν
β
+
1
2
ǫ
ν
µρ
ω
µ
α
ω
ρ
β
_
= 0,
ǫ
αβγ
_
∂
α
a
ν
β
+
1
2
ǫ
ν
µρ
e
µ
α
e
ρ
β
+ ǫ
ν
µρ
ω
µ
α
a
ρ
β
_
= 0.
The ﬁrst two equations say that the torsion and the curvature has to vanish.
ChernSimons gravity. . . 3
Dimensional reduction of our theory reproduces the formalism of Cangemi and Jackiw [2]. Let us in
fact assume that none of the ﬁelds depends on the second coordinate x
2
. Using the indices µ = (k, 2) and
α = (a, 2),
(13)
e
k
b
→ e
k
b
, e
k
2
→ η
k
,
a
2
2
→ η
2
, ω
2
2
→ η
3
,
ω
2
a
→ ω
a
, a
2
a
→ a
a
,
a
k
α
= ω
k
α
= 0.
Then the ChernSimons Lagrangian (12) reduces precisely to the Lagrangian (1.3) of Cangemi and Jackiw.
Note added: This paper has been written in 1993, but remained unpublished because of its overlap with
Cangemi’s Salamanca talk [11]. The structure studied here has been applied to various physical problems
and in particular to string theory [3,12]. Similar conclusions to those presented here were reached later,
independently, by de Montigny at al [13].
References
[1] H. Verlinde, in The Sixth Marcel Grossmann Meeting on General Relativity, edited by M. Sato (World
Scientiﬁc, Singapore, 1992); C. Callan, S. Giddings, A. Harvey and A. Strominger, Phys. Rev. D45, 1005
(1992).
[2] D. Cangemi and R. Jackiw, Phys. Rev. Lett. 69, 233 (1992); Phys. Lett. B299, 24 (1993); Ann. Phys.
225, 229 (1993).
[3] A. Ach´ ucarro and P. K. Townsend, Phys. Lett. B180, 89 (1986); E. Witten, Nucl. Phys. B311, 47
(1988/89).
[4] C. Nappi and E. Witten, Phys. Rev. Lett. 71, 3751 (1993).
[5] E. Witten, Commun. Math. Phys. 117, 353 (1988).
[6] A. Medina and Ph. Revoy, Ann. Sci.
´
Ec. Norm. Sup. 18, 553 (1985); J. M. FigueroaO’Farrill and S.
Stanciu, hepth/9402035
[7] D. I. Olive, E. Rabinovici and A. Schwimmer, Phys. Lett. 321B, 355 (1994)
[8] C. Klimˇcik and A. A. Tseytlin, Phys. Lett. B323, 305 (1994); K. Steftsos, Phys. Rev. D50, 2784 (1994)
N. Mohammedi, Phys. Lett. B331, 93 (1994); E. Kiritsis and C. Kounnas Phys. Lett. B320, 264 (1994)
[9] A. Ach´ ucarro, Phys. Rev. Lett. 70, 1037 (1993); A. Ach´ ucarro and M. E. Ortiz, Phys. Rev. D48, 3600
(1993).
[10] N. Ja. Vilenkin and A. U. Klimyk, Representation of Lie Groups and Special Functions. Recent Advances
(Mathematics and Its Applications. (1994).
4 Duval, Horv´ ath, Horv´ athy
[11] D. Cangemi, Phys. Lett. B297, 261 (1992); in Proc. Group Theoretical Methods in Phys., Salamanca’92,
del Olmo, Santander, Guilarte (eds.), Anales de Fisica Monograﬁas, Vol. II, p. 72 (1993). See also Cangemi,
Phys. Lett. B297, 261 (1992).
[12] pp waves were ﬁrst discussed by H. W. Brinkmann, Math. Ann. 94, 119145 (1925), and have been
rediscovered and applied to strings in C. Duval, G. Gibbons and P. A. Horv´ athy, Phys. Rev. D43, 3907
(1991). See A. A. Tseytlin, Nucl. Phys. B390, 153 (1993); C. Duval, Z. Horv´ ath and P. A. Horv´ athy, Phys.
Lett. B313, 10 (1993); Mod. Phys. Lett. A8, 3749 (1993).
[13] M. de Montigny, F. C. Khanna, A. E. Santana, E. S. Santos and J. D. M. Vianna, Annals Phys. 277
144 (1999).