g

r
-
q
c
/
9
7
0
3
0
7
4



2
6

M
a
r

9
7
A GENERAL EXPRESSION FOR THE QUARTIC LOVELOCK TENSOR
C. C. Briggs
Center for Academic Computing, Penn State University, University Park, PA 16802
Tuesday, March 25, 1997, 7:37:49 PM
Abstract. A general expression is given for the quartic Lovelock tensor in terms of the Riemann-Christoffel and Ricci curvature tensors and the Riemann curvature scalar for
n-dimensional differentiable manifolds having a general linear connection. In addition, expressions are given (in the appendix) for the coefficient of the quartic Lovelock
Lagrangian as well as for lower-order Lovelock tensors and Lovelock Lagrangian coefficients.
PACS numbers:02.40.-k, 04.20.Cv, 04.20.Fy
Since the quartic part of the Lovelock action involves 25 terms, each containing the product of four curvature terms, the writing of the corresponding
field equations using the usual method of variation of the Lagrangian with respect to the metric tensor is a formidable task.
Demaret, J., Y. De Rop, P. Tombal, and A. Moussiaux, Qualitative Analysis of
Ten-dimensional Lovelock Cosmological Models, Gen. Rel. Grav., 24 (1992) 1169.
Writing explicitly the field equations for general space-times in high dimensions, taking into account all the relevant terms of the Lovelock action, is a
very complex task. In the case of, say, ten-dimensional models, it can take weeks to write the equations by hand, and the absolute correctness of the
results is not at all guaranteed. As an example of the formidable complexity of this type of calculation, let us note that the quartic part of the Lovelock
action involves 25 terms, each of which constituted by the contracted product of four curvature tensors, and, moreover, the resulting field equations,
obtained after variation with respect to the metric tensor, are not explicitly known.
Demaret, J., H. Caprasse, A. Moussiaux, P. Tombal, and D. Papadopoulos,
Ten-dimensional Lovelock-type space-times, Phys. Rev. D, 41 (1990) 1163.
The general expression of the variation of the fourth-order contribution to the Lovelock Lagrangian density has even not yet been obtained.
Ibid.
This letter provides the interested reader with a general expression for the
quartic Lovelock tensor G
(4)a
b
in terms of the Riemann-Christoffel curvature
tensor
1
R
abc
d
2 (
[a

b]
d
c
+
[a
d
|e|

b]
e
c
+
a
e
b

e
d
c
), (1)
the Ricci curvature tensor
R
a
b
R
ca
bc
= R
ac
bc
= R
ca
cb
+ 2 (
[c
Q
a]
bc
+ S
ca
d
Q
d
bc
), (2)
and the Riemann curvature scalar
R R
a
a
= R
ba
ab
= R
ab
ab
= R
ab
ba
= R
ba
ba
(3)
using anholonomic coordinates for n-dimensional differentiable manifolds
having a general linear connection, where
a
is the Pfaffian derivative,
a
b
c
the connection coefficient,
a
b
c
the object of anholonomity, Q
a
bc
the
non-metricity tensor, and S
ab
c
the torsion tensor. (An expression for the
coefficient L
(4)
of the quartic Lovelock Lagrangian appears in the appendix.)
Such an expression for G
(4)a
b
has evidently not appeared in the open
literature hitherto, even though analogues of some terms of G
(4)a
b
per Eq. (7)
below have already appeared, viz. analogues of the 103
rd
, 104
th
, 107
th
, 109
th
,
111
th
, 112
th
, and 115
th
terms in Grisaru and Zanon
2
, of the 109
th
and 111
th
in
Freeman et al.,
3
and of the 111
th
in Grisaru et al.
4
In accordance with various general definitions given by Mller-Hoissen
5
and Verwimp,
6
L
(p)
and G
(p)a
b
for p 1 are given by the formulas
L
(p)
=
(2p)!
2
p
R
[i
1
i
2
i
1
i
2
R
i
3
i
4
i
3
i
4
R
i
2 p 1
i
2p
]
i
2p 1
i
2p
, (4)
and
G
(p)a
b
=
(2p + 1)!
2
p + 1
p

b
[a
R
i
1
i
2
i
1
i
2
R
i
3
i
4
i
3
i
4
R
i
2 p 1
i
2p
]
i
2p 1
i
2p
, (5)
which comprise (2p)! and (2p + 1)! tensor index permutations, respectively.
Numerical properties of L
(p)
and G
(p)a
b
for 0 p 4 are tabulated below.
Expressions for G
(p)a
b
and L
(p)
for 0 p 3 appear in the appendix.
TABLE 1. SOME NUMERICAL PROPERTIES OF L
(p)
FOR 0 p 4. TABLE 2. SOME NUMERICAL PROPERTIES OF G
(p)a
b
FOR 0 p 4.
Order (p) Quantity
Number of
Terms
Number of Permuta-
tions Comprehended
Sum of Numerical
Factors
Order (p) Quantity
Number of
Terms
Number of Permuta-
tions Comprehended
Sum of Numerical
Factors
0 L
(0)
1 1 1 0 G
(0)a
b
1 1 1
1 L
(1)
1 2 1 1 G
(1)a
b
2 6 1
1
/2
2 L
(2)
3 24 6 2 G
(2)a
b
7 120 7
1
/2
3 L
(3)
8 720 90 3 G
(3)a
b
2 6 5040 105
4 L
(4)
2 5 40,320 2520 4 G
(4)a
b
115 362,880 2835
The quartic Lovelock tensor G
(4)a
b
is given by the formula
G
(4)a
b
=
9!
2
5
4

[a
b
R
i
1
i
2
i
1
i
2
R
i
3
i
4
i
3
i
4
R
i
5
i
6
i
5
i
6
R
i
7
i
8
]
i
7
i
8
, (6)
which comprises 362,880 unique covariant index permutations, of which but 115together with numerical coefficientssuffice for rendering a general
expression for G
(4)a
b
, the final result (after substituting contractions and re-labeling indices) being given by
G
(4)a
b
=
1
8
(
a
b
R
4
24
a
b
R
2
R
d
c
R
c
d
+ 6
a
b
R
2
R
ef
cd
R
cd
ef
+ 64
a
b
R R
d
c
R
e
d
R
c
e
96
a
b
R R
e
c
R
f
d
R
cd
ef
96
a
b
R R
d
c
R
fg
de
R
ce
fg
(7)
8
a
b
R R
ef
cd
R
gh
ef
R
cd
gh
+ 32
a
b
R R
eg
cd
R
ch
ef
R
df
gh
+ 48
a
b
R
d
c
R
c
d
R
f
e
R
e
f
96
a
b
R
d
c
R
e
d
R
f
e
R
c
f
+ 384
a
b
R
d
c
R
f
d
R
g
e
R
ce
fg

24
a
b
R
d
c
R
c
d
R
gh
ef
R
ef
gh
+ 192
a
b
R
d
c
R
e
d
R
gh
ef
R
cf
gh
+ 96
a
b
R
e
c
R
f
d
R
gh
ef
R
cd
gh
192
a
b
R
e
c
R
g
d
R
ch
ef
R
df
gh
+
+ 192
a
b
R
e
c
R
g
d
R
dh
ef
R
cf
gh
192
a
b
R
d
c
R
cf
de
R
hi
fg
R
eg
hi
+ 96
a
b
R
d
c
R
fg
de
R
hi
fg
R
ce
hi
384
a
b
R
d
c
R
fh
de
R
ci
fg
R
eg
hi
+
+ 3
a
b
R
ef
cd
R
cd
ef
R
ij
gh
R
gh
ij
48
a
b
R
ef
cd
R
cg
ef
R
ij
gh
R
dh
ij
+ 6
a
b
R
ef
cd
R
gh
ef
R
ij
gh
R
cd
ij
96
a
b
R
ef
cd
R
gi
ef
R
cj
gh
R
dh
ij
+
+ 48
a
b
R
eg
cd
R
ci
ef
R
dj
gh
R
fh
ij
96
a
b
R
eg
cd
R
ci
ef
R
fj
gh
R
dh
ij
8 R
a
b
R
3
+ 48 R
a
c
R
2
R
c
b
+ 96 R
a
b
R R
d
c
R
c
d
192 R
a
c
R R
d
b
R
c
d
+
+ 192 R
a
c
R R
e
d
R
cd
be
24 R
a
b
R R
ef
cd
R
cd
ef
+ 96 R
a
c
R R
ef
bd
R
cd
ef
128 R
a
b
R
d
c
R
e
d
R
c
e
192 R
a
c
R
c
b
R
e
d
R
d
e
+ 384 R
a
c
R
d
b
R
e
d
R
c
e
+
+ 192 R
a
b
R
e
c
R
f
d
R
cd
ef
384 R
a
c
R
e
b
R
f
d
R
cd
ef
384 R
a
c
R
e
d
R
f
e
R
cd
bf
+ 384 R
a
c
R
f
d
R
c
e
R
de
bf
+ 192 R
a
b
R
d
c
R
fg
de
R
ce
fg
+
1
Schouten, J. A., Ricci Calculus, Springer-Verlag, Berlin (1954), p. 171.
2
Grisaru, M. T., and D. Zanon, Sigma-Model Superstring Corrections to the Einstein-Hilbert Action, Phys. Lett., 177B (1986) 347.
3
Freeman, M. D., C. N. Pope, M. F. Sohnius, and K. S. Stelle, Higher-Order -Model Counterterms and the Effective Action for Superstrings, Phys. Lett., 178B (1986) 199.
4
Grisaru, M. T., A. E. M. van de Ven, and D. Zanon, Four-Loop -Function for the N = 1 and N = 2 Supersymmetric Non-Linear Sigma Model in Two Dimensions, Phys. Lett., 173B
(1986) 423.
5
Mller-Hoissen, F., Spontaneous Compactification with Quadratic and Cubic Curvature Terms, Phys. Lett., 163B (1985) 106; From Chern-Simons to Gauss-Bonnet, Nucl. Phys.,
B346 (1990) 235.
6
Verwimp, T., On higher dimensional gravity: the Lagrangian, its dimensional reduction and a cosmological model, Class. Quantum Grav., 6 (1989) 1655.
2 Tuesday, March 25, 1997, 7:37:49 PM
+ 48 R
a
c
R
c
b
R
fg
de
R
de
fg
192 R
a
c
R
d
b
R
fg
de
R
ce
fg
192 R
a
c
R
c
d
R
fg
be
R
de
fg
192 R
a
c
R
e
d
R
fg
be
R
cd
fg
+ 384 R
a
c
R
f
d
R
cg
be
R
de
fg

384 R
a
c
R
f
d
R
dg
be
R
ce
fg
+ 16 R
a
b
R
ef
cd
R
gh
ef
R
cd
gh
64 R
a
b
R
eg
cd
R
ch
ef
R
df
gh
+ 192 R
a
c
R
ce
bd
R
gh
ef
R
df
gh
96 R
a
c
R
ef
bd
R
gh
ef
R
cd
gh

384 R
a
c
R
eg
bd
R
dh
ef
R
cf
gh
48 R
ad
bc
R
2
R
c
d
24 R
ae
cd
R
2
R
cd
be
+ 192 R
ad
bc
R R
e
d
R
c
e
+ 192 R
ae
cd
R R
c
b
R
d
e
192 R
ae
bc
R R
f
d
R
cd
ef

96 R
ae
cd
R R
f
b
R
cd
ef
+ 96 R
ae
cd
R R
f
e
R
cd
bf
192 R
af
cd
R R
c
e
R
de
bf
96 R
ad
bc
R R
fg
de
R
ce
fg
+ 48 R
ae
cd
R R
fg
be
R
cd
fg
192 R
af
cd
R R
cg
be
R
de
fg
+
+ 192 R
ad
bc
R
c
d
R
f
e
R
e
f
384 R
ad
bc
R
e
d
R
f
e
R
c
f
384 R
ae
cd
R
c
b
R
f
e
R
d
f
384 R
af
cd
R
e
b
R
c
e
R
d
f
+ 384 R
ad
bc
R
f
d
R
g
e
R
ce
fg
+
+ 384 R
af
bc
R
c
d
R
g
e
R
de
fg
+ 384 R
af
bc
R
e
d
R
g
e
R
cd
fg
192 R
ae
cd
R
f
b
R
g
e
R
cd
fg
+ 384 R
af
cd
R
c
b
R
g
e
R
de
fg
+ 192 R
af
cd
R
e
b
R
g
e
R
cd
fg

384 R
af
cd
R
g
b
R
c
e
R
de
fg
192 R
ae
cd
R
f
e
R
g
f
R
cd
bg
+ 384 R
af
cd
R
c
e
R
g
f
R
de
bg
384 R
af
cd
R
g
e
R
c
f
R
de
bg
192 R
ag
cd
R
c
e
R
d
f
R
ef
bg
+
+ 384 R
ag
cd
R
f
e
R
c
f
R
de
bg
+ 96 R
ag
cd
R
f
e
R
e
f
R
cd
bg
48 R
ad
bc
R
c
d
R
gh
ef
R
ef
gh
+ 192 R
ad
bc
R
e
d
R
gh
ef
R
cf
gh
+ 192 R
ae
bc
R
c
d
R
gh
ef
R
df
gh
+
+ 192 R
ae
bc
R
f
d
R
gh
ef
R
cd
gh
384 R
ae
bc
R
g
d
R
ch
ef
R
df
gh
+ 384 R
ae
bc
R
g
d
R
dh
ef
R
cf
gh
+ 192 R
ae
cd
R
c
b
R
gh
ef
R
df
gh
+ 96 R
ae
cd
R
f
b
R
gh
ef
R
cd
gh
+
+ 384 R
ag
cd
R
e
b
R
ch
ef
R
df
gh
192 R
ae
cd
R
c
e
R
gh
bf
R
df
gh
96 R
ae
cd
R
f
e
R
gh
bf
R
cd
gh
+ 384 R
ae
cd
R
g
e
R
ch
bf
R
df
gh
+ 192 R
af
cd
R
c
e
R
gh
bf
R
de
gh

384 R
af
cd
R
g
e
R
ch
bf
R
de
gh
192 R
af
cd
R
g
e
R
eh
bf
R
cd
gh
384 R
ag
cd
R
c
e
R
dh
bf
R
ef
gh
+ 384 R
ag
cd
R
c
e
R
eh
bf
R
df
gh
+ 384 R
ag
cd
R
f
e
R
ch
bf
R
de
gh
+
+ 192 R
ag
cd
R
f
e
R
eh
bf
R
cd
gh
192 R
ag
cd
R
h
e
R
cd
bf
R
ef
gh
+ 384 R
ag
cd
R
h
e
R
ce
bf
R
df
gh
192 R
ad
bc
R
cf
de
R
hi
fg
R
eg
hi
+ 96 R
ad
bc
R
fg
de
R
hi
fg
R
ce
hi

384 R
ad
bc
R
fh
de
R
ci
fg
R
eg
hi
24 R
ae
cd
R
cd
be
R
hi
fg
R
fg
hi
+ 192 R
ae
cd
R
cf
be
R
hi
fg
R
dg
hi
48 R
ae
cd
R
fg
be
R
hi
fg
R
cd
hi
+ 192 R
ae
cd
R
fh
be
R
ci
fg
R
dg
hi
+
+ 96 R
af
cd
R
cd
be
R
hi
fg
R
eg
hi
+ 192 R
af
cd
R
cg
be
R
hi
fg
R
de
hi
+ 192 R
af
cd
R
gh
be
R
ei
fg
R
cd
hi
+ 384 R
ah
cd
R
cf
be
R
di
fg
R
eg
hi

384 R
ah
cd
R
cf
be
R
ei
fg
R
dg
hi
+ 192 R
ah
cd
R
fg
be
R
ci
fg
R
de
hi
+ 96 R
ah
cd
R
fg
be
R
ei
fg
R
cd
hi
+ 384 R
ah
cd
R
fi
be
R
ce
fg
R
dg
hi
).
For a check, note that (1) G
(4)a
a
=
n 8
8
L
(4)
and (2) the magnitudes of the numerical coefficients of G
(4)a
b
add up to
9!
2
5
4
=
1
8
22,680 = 2835.
APPENDIX
The quartic Lovelock Lagrangian coefficient L
(4)
is given by the formula
L
(4)
=
8!
2
4
R
[i
1
i
2
i
1
i
2
R
i
3
i
4
i
3
i
4
R
i
5
i
6
i
5
i
6
R
i
7
i
8
]
i
7
i
8
, (8)
which comprises 40,320 unique covariant index permutations, of which but 25together with numerical coefficientssuffice for rendering a general expression
for L
(4)
, the final result (after substituting contractions and re-labeling indices) being given by
L
(4)
= R
4
24 R
2
R
b
a
R
a
b
+ 6 R
2
R
cd
ab
R
ab
cd
+ 64 R R
b
a
R
c
b
R
a
c
96 R R
c
a
R
d
b
R
ab
cd
96 R R
b
a
R
de
bc
R
ac
de
8 R R
cd
ab
R
ef
cd
R
ab
ef
+ (9)
+ 32 R R
ce
ab
R
af
cd
R
bd
ef
+ 48 R
b
a
R
a
b
R
d
c
R
c
d
96 R
b
a
R
c
b
R
d
c
R
a
d
+ 384 R
b
a
R
d
b
R
e
c
R
ac
de
24 R
b
a
R
a
b
R
ef
cd
R
cd
ef
+
+ 192 R
b
a
R
c
b
R
ef
cd
R
ad
ef
+ 96 R
c
a
R
d
b
R
ef
cd
R
ab
ef
192 R
c
a
R
e
b
R
af
cd
R
bd
ef
+ 192 R
c
a
R
e
b
R
bf
cd
R
ad
ef
192 R
b
a
R
ad
bc
R
fg
de
R
ce
fg
+
+ 96 R
b
a
R
de
bc
R
fg
de
R
ac
fg
384 R
b
a
R
df
bc
R
ag
de
R
ce
fg
+ 3 R
cd
ab
R
ab
cd
R
gh
ef
R
ef
gh
48 R
cd
ab
R
ae
cd
R
gh
ef
R
bf
gh
+
+ 6 R
cd
ab
R
ef
cd
R
gh
ef
R
ab
gh
96 R
cd
ab
R
eg
cd
R
ah
ef
R
bf
gh
+ 48 R
ce
ab
R
ag
cd
R
bh
ef
R
df
gh
96 R
ce
ab
R
ag
cd
R
dh
ef
R
bf
gh
.
For a check, note that the magnitudes of the numerical coefficients of L
(4)
add up to
8!
2
4
= 2520.
The Lovelock tensors G
(p)a
b
for 0 p 3 are given by
G
(0)a
b
=
1!
2
0

a
b
, (10)
G
(1)a
b
=
3!
2
2
1

[a
b
R
i
1
i
2
]
i
1
i
2
, (11)
G
(2)a
b
=
5!
2
3
2

[a
b
R
i
1
i
2
i
1
i
2
R
i
3
i
4
]
i
3
i
4
, (12)
and
G
(3)a
b
=
7!
2
4
3

[a
b
R
i
1
i
2
i
1
i
2
R
i
3
i
4
i
3
i
4
R
i
5
i
6
]
i
5
i
6
, (13)
whence
G
(0)a
b
=
a
b
, (14)
G
(1)a
b
=
1
2
(
a
b
R + 2 R
a
b
), (15)
G
(2)a
b
=
1
4
(
a
b
R
2
4
a
b
R
d
c
R
c
d
+
a
b
R
ef
cd
R
cd
ef
4 R
a
b
R + (16)
+ 8 R
a
c
R
c
b
8 R
ad
bc
R
c
d
4 R
ae
cd
R
cd
be
),
and
G
(3)a
b
=
1
6
(
a
b
R
3
+ 12
a
b
R R
d
c
R
c
d
3
a
b
R R
ef
cd
R
cd
ef
16
a
b
R
d
c
R
e
d
R
c
e
+ 24
a
b
R
e
c
R
f
d
R
cd
ef
+ 24
a
b
R
d
c
R
fg
de
R
ce
fg
+ 2
a
b
R
ef
cd
R
gh
ef
R
cd
gh
(17)
8
a
b
R
eg
cd
R
ch
ef
R
df
gh
+ 6 R
a
b
R
2
24 R
a
c
R R
c
b
24 R
a
b
R
d
c
R
c
d
+ 48 R
a
c
R
d
b
R
c
d
48 R
a
c
R
e
d
R
cd
be
+ 6 R
a
b
R
ef
cd
R
cd
ef

24 R
a
c
R
ef
bd
R
cd
ef
+ 24 R
ad
bc
R R
c
d
+ 12 R
ae
cd
R R
cd
be
48 R
ad
bc
R
e
d
R
c
e
48 R
ae
cd
R
c
b
R
d
e
+ 48 R
ae
bc
R
f
d
R
cd
ef
+ 24 R
ae
cd
R
f
b
R
cd
ef

24 R
ae
cd
R
f
e
R
cd
bf
+ 48 R
af
cd
R
c
e
R
de
bf
+ 24 R
ad
bc
R
fg
de
R
ce
fg
12 R
ae
cd
R
fg
be
R
cd
fg
+ 48 R
af
cd
R
cg
be
R
de
fg
).
For a check, notefor 1 p 3that (1) G
(p)a
a
=
n 2p
2p
L
(p)
and (2) the magnitudes of the numerical coefficients of G
(p)a
b
add up to
(2p + 1)!
2
p + 1
p
.
The Lovelock Lagrangian coefficients L
(p)
for 0 p 3 are given by
L
(0)
=
0!
2
0
= 1, (18)
L
(1)
=
2!
2
1
R
[i
1
i
2
]
i
1
i
2
, (19)
L
(2)
=
4!
2
2
R
[i
1
i
2
i
1
i
2
R
i
3
i
4
]
i
3
i
4
, (20)
and
L
(3)
=
6!
2
3
R
[i
1
i
2
i
1
i
2
R
i
3
i
4
i
3
i
4
R
i
5
i
6
]
i
5
i
6
, (21)
whence
L
(0)
= 1, (22)
L
(1)
= R, (23)
L
(2)
= R
2
4 R
b
a
R
a
b
+ R
cd
ab
R
ab
cd
, (24)
and
L
(3)
= R
3
+ 12 R R
b
a
R
a
b
3 R R
cd
ab
R
ab
cd
16 R
b
a
R
c
b
R
a
c
+ (25)
+ 24 R
c
a
R
d
b
R
ab
cd
+ 24 R
b
a
R
de
bc
R
ac
de
+
+ 2 R
cd
ab
R
ef
cd
R
ab
ef
8 R
ce
ab
R
af
cd
R
bd
ef
.
For a check, note that (1) the magnitudes of the numerical coefficients of L
(p)
add upfor any pto
(2p)!
2
p
and (2)

R
L
(p)
= p L
(p - 1)
for p 1.