Physics Letters B 772 (2017) 300–305

Contents lists available at ScienceDirect

Physics Letters B
www.elsevier.com/locate/physletb

General relativity from three-forms in seven dimensions

Kirill Krasnov
School of Mathematical Sciences, University of Nottingham, UK

a r t i c l e i n f o a b s t r a c t

Article history: We consider a certain theory of 3-forms in 7 dimensions, and study its dimensional reduction to 4D,
Received 5 May 2017 compactifying the 7-dimensional manifold on the 3-sphere of a fixed radius. We show that the resulting
Received in revised form 8 June 2017 4D theory is (Riemannian) General Relativity (GR) in Plebanski formulation, modulo corrections that are
Accepted 9 June 2017
negligible for curvatures smaller than Planckian. Possibly the most interesting point of this construction
Available online 29 June 2017
Editor: M. Cvetič
is that the dimensionally reduced theory is GR with a non-zero cosmological constant, and the value of
the cosmological constant is directly related to the size of S 3 . Realistic values of  correspond to S 3 of
Planck size.
© 2017 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/by/4.0/). Funded by SCOAP3 .

It is almost universally agreed that General Relativity (GR) the 3-sphere (of a fixed size) gives a 4D gravity theory that at low
should be viewed as a low energy approximation to some other energies is indistinguishable from GR.
theory. The usual view is that this is some quantum gravity theory Mathematically, the idea of our construction is rather simple
that gives the ultra-violet (UV) completion of perturbatively quan- and natural and can be explained already here. It is well-known
tised GR. that gravity can be usefully described using the language of frame
It is also possible that GR arises as the low energy limit of some fields instead of the metric. A frame e I is a collection of one-
other classical theory, and it is only this distinct from GR classical forms that are declared orthonormal. This defines the metric via
theory that is to be UV completed quantum mechanically. A sce- ds2 = η I J e I e J , where η I J is the flat metric. When the frame field
nario of this sort is part of string theory, where 4D gravity arises is used to describe geometry, it is also very natural to allow the
by compactification from 11D supergravity. The latter is to be com- spin connection w I J (later used to construct the curvature) to be-
pleted quantum mechanically by M-theory. come an independent object. Its relation to derivatives of the frame
Scenarios embedding 4D GR into other classical theories are in- is then fixed by its field equation.
teresting for several reasons. First, what comes out from such an The (spin) connection w I J can locally be viewed as a one-form
embedding is usually more than just GR. There are typically more with values in the Lie algebra of Lorentz group, or orthogonal
degrees of freedom, and this is of interest both phenomenologi- group SO( D ) of appropriate dimension D if one wants to de-
cally as well as for the question of the UV completion. Indeed, UV scribe metrics of Riemannian signature. On the other hand, the
incomplete theories may arise as the low energy approximation of frame one-forms are not so ( D ) Lie algebra valued (apart from the
UV complete theories, the latter having more degrees of freedom case of 3D gravity where the frame index can be identified with
than the former. the Lie algebra index). However, one can construct two-forms that
In the extensively studied Kaluza–Klein scenarios 4D gravity are Lie algebra valued, taking the wedge product of frame with
arises from a higher dimensional theory that is again gravity (pos- itself B I J := e I ∧ e J . This gives rise to a formalism for gravity
sibly with extra fields). The purpose of this letter is to point out in which the basic fields are Lie algebra valued one-forms w I J
that (Riemannian signature) General Relativity in four dimensions and two-forms B I J . The two-forms must however be appropriately
may be obtained by compactification from a higher-dimensional constrained in order to guarantee that they come from the frame
theory of a very different nature. Thus, we show that a certain dy- field. This formalism is particularly simple in 4 dimensions, see
namical theory of 3-forms in 7 dimensions, when compactified on below, but can also be used to describe gravity in any dimension,
see [1].
The main idea behind the constructions of this letter is that
E-mail address: [email protected]. it is natural to combine the one- and two-forms of such a for-

http://dx.doi.org/10.1016/j.physletb.2017.06.025
0370-2693/© 2017 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). Funded by
SCOAP3 .
 K. Krasnov / Physics Letters B 772 (2017) 300–305 301

mulation of gravity into a single object — a 3-form in a higher it is usually much more effective to compute the volume volC from
dimensional space. The space in question is (locally) the product the determinant of g C .
of the spacetime and the SO( D ) group manifolds, i.e. it is the to- Let us now make C dynamical. Consider the following action
tal space of the principal SO( D ) bundle over space(time). Let m I J principle
be Lie algebra valued one-forms on the SO( D ) group manifold, i.e. 
1
Maurer–Cartan forms. To exhibit the fields w I J and B I J as com- S [C ] = C ∧ dC + 6λ volC . (4)
ponents of a 3-form in the total space of the bundle, we use the 2
M
matrix notation in which m I J , w I J are one-forms and B I J are two-
forms valued in the space of anti-symmetric D × D matrices. We The first term here (i.e. the case λ = 0) describes a topological field
then consider the following 3-form theory considered in [11]. The Euler–Lagrange equations following
from (4) are
C = Tr (m ∧ m ∧ m + m ∧ m ∧ w + m ∧ B ) . (1)
dC = λ ∗ C . (5)
IJ
Thus, the spin connection 1-forms w arise as two vertical one
horizontal components of the 3-form C above, and the fields B I J Here ∗ C is the Hodge dual of C computed using g C . The numer-
arise as the one vertical two horizontal components. This recovers ical coefficient on the right-hand-side here is simplest verified by
all fields of the 2-form formulation of gravity as components of noticing that volC = −(1/7)C ∧ ∗ C , and then using the homogene-
a 3-form in the total space of the SO( D ) group bundle over the ity to compute the variation of volC with respect to C . We note
that, because the two terms in (4) scale differently, by rescaling C
spacetime. In [2,3] it was explained how 3D gravity can be usefully
we can always achieve λ = 1 at the expense of introducing a pa-
described in this fashion. The purpose of this letter is to outline the
rameter in front of the action. We will do so from now on. Thus,
construction that works for 4D gravity.
there are no free parameters in the theory (4).
While the idea described is general enough to work in any
Real 3-forms C are of two possible types, see e.g. [7]. Forms
dimension, the 3-forms are most interesting objects in 6 and 7 di-
of one type give g C of signature (4, 3). Forms of the other type
mensions, as we review below. On the other hand, the total space
give the Riemannian signature metrics g C . Such 3-forms satisfying
of the SO(4) group principal bundle over a 4-dimensional mani-
(5) describe what [8] call nearly parallel G 2 structures. Note that
fold is 10 dimensional. But instead of working with the full 4D
(5) implies that ∗ C is closed. However, this equation also says that
rotation group one can take one of its chiral halves, i.e. either its
dC = 0. Thus, the critical points of (4) are not the possibly more fa-
self-dual or anti-self-dual part, both of which are 3-dimensional.
miliar in this context torsion-free G 2 structures satisfying dC = 0,
There is a formalism for 4D gravity [4,5] that works with one- and
d∗ C = 0 and describing G 2 holonomy manifolds. A related observa-
two-forms with values in the Lie algebra of the chiral half of the
tion is that the equation (5) implies that the metric g C is Einstein
rotation group. We will base our construction on this formalism.
with non-zero scalar curvature, see proposition 3.10 from [8]. In
In this case the total space of the bundle is 7-dimensional, and we
contrast, G 2 -holonomy manifolds are Ricci flat, see e.g. [6].
are in the setting where 3-forms are most natural and interest-
Equations (5) have been studied in the literature [8]. The vari-
ing.
ational principle (4) is a subcase of a more general action (29)
The fundamental fact about a generic (or using the terminology
in [12], with 1- and 5-forms set to zero and the auxiliary met-
of [6] stable) 3-form C on a 7-dimensional manifold M is that it
ric integrated out. It is also similar to prepotentials appearing in
defines a metric g C . The metric is explicitly given by the following
the literature on G 2 compactifications of M-theory, see e.g. [13].
formula
Nevertheless, it seems that the theory of 3-forms (4) has not been
1 studied in the literature. Note that action (4) is different from the
g C (ξ, η)volC = − i ξ C ∧ i η C ∧ C . (2)
6 ones considered by Hitchin [6]. The simplest Hitchin action is the
last term in (4), restricted to 3-forms in a fixed cohomology class.
Here g C (ξ, η) is the result of the metric contraction of two vector
Our action is the sum of those in [11] and [6], with the only con-
fields ξ, η , volC is the volume form for g C , and i ξ,η is the opera-
straint on C being that it is generic (or stable), which is an open
tion of insertion of a vector field into a form. The metric (2) has
condition.
been known for more than a century, see e.g. [7] for a historical
Additional information about the theory (4) is given in a recent
perspective. It is ultimately related to the geometry of spinors in
paper [14]. In particular, this reference demonstrates that (4) is a
7 and 8 dimensions, see e.g. [8] for the discussion of the spinor
7D theory with 3 propagating degrees of freedom. Also, some so-
aspect, and to octonions, see e.g. [9].
lutions of the equations (5) are described, in particular those that
Generic 3-forms in 7 dimensions are related to the exceptional
can be obtained as S 3 bundles over S 4 .
group G 2 . This can be defined as the subgroup of GL(7) that sta-
We now describe a relation to 4D General Relativity (GR). We
bilises a generic 3-form, see [10] and more recently [6]. The space
claim that (4) dimensionally reduced on S 3 (of a fixed radius) is a
of generic 3-forms (at a point) can then be identified with the
4D theory of gravity that is for all practical purposes indistinguish-
coset GL(7)/G 2 . It is easy to see that the spaces GL(7)/G 2 and
able from (Riemannian signature) GR. This means that while the
3 R7 have the same dimension 35. The fact that C defines g C ex-
reduced theory is, strictly speaking, not GR, it coincides with GR
plains why G 2 is a subgroup of SO(7).
for Weyl curvatures smaller than Planckian, which is anyway the
The volume form volC , playing an important role below, can
regime where we can trust GR as a classical theory. All this is to
also be described explicitly as a homogeneity degree 7/3 object
be explained in more details below.
built from C . Thus, let ˜ a1 ...a7 be the densitiesed completely anti-
To explain why the outlined embedding of 4D GR into a the-
symmetric tensor taking values ±1 in any coordinate system. Here
ory of 3-forms in 7D may be interesting, let us remind the reader
a = 1, . . . , 7. We can then construct the following degree 7 and
the basics of Kaluza–Klein (KK) theory. Here one starts with GR
weight 3 scalar:
in higher dimensions and dimensionally reduces to 4 dimensions.
In the simplest and also historically the first setup one starts with
˜ a1 ...a7 ˜ b1 ...b7 ˜ c1 ...c7 C a1 b1 c1 . . . C a7 b7 c7 . (3)
GR (with zero cosmological constant) in 5 dimensions and dimen-
The cube root of this expression is a multiple of volC . This is not sionally reduces on S 1 . If one fixes the size of S 1 , as was done
the most useful in practise way of computing the volume form — in the first treatments, the dimensionally reduced 4D theory is GR
302 K. Krasnov / Physics Letters B 772 (2017) 300–305

coupled to Maxwell. Allowing the size of the circle to become dy- components of C has been lost here, as a simple count of com-
namical gives rise to an additional massless scalar field in 4D, and ponents in φ, A , B , c shows. The above parametrisation of C is
to avoid conflict with observations this must be given a mass, or however not the one most suited for computations. We note that
stabilised in some other way. the term quadratic in m can always be eliminated by shifting m by
The Kaluza–Klein mechanism gives a geometrically compelling a Lie algebra valued 1-form. This also redefines B , c. This suggests
unification of gravity with electromagnetism. Also, as pointed out we parametrise
by Kaluza, it relates the quantum of electric charge to the size of  
1
the compact extra dimension. The Kaluza–Klein mechanism can be C = −2 Tr φ 3 W 3 + φ W B + c. (6)
generalised to describe non-Abelian gauge fields. A comprehensive 3
review on KK is e.g. [15]. Here W = m + A is a connection in the total space of the bun-
Let us return to our story. We claimed that 4D GR arises as the dle. Geometrically, an SU(2) invariant 3-form C in the total space
dimensional reduction of the theory (4) of 3-forms in 7D. Unlike of an SU(2) bundle over M defines a connection by declaring the
the KK case, no unification is achieved here, the reduced theory is horizontal vector fields to be those that are in the kernel of the
pure gravity. Also unlike KK, gravity in 4D arises from a 7D theory 1-form i ξ V i η V C for arbitrary vertical vector fields ξ V , η V . In (6) we
of a very different sort — the theory (4) is a dynamical theory of simply chose to parametrise the 3-form C by this connection. The
3-forms, not metrics. While it may be amusing that 4D GR admits parametrisation (6) is most suited for practical computations. Nu-
a lift to a theory of such a different nature as (4), is this a useful merical prefactors are for future convenience.
perspective on 4D gravity? A simple computation then gives
Now comes what we believe is the most interesting physics 
point about our construction. As we will show, the dimensionally dC = −2 Tr φ 2 dφ W 3 + (φ 3 F + φ B ) W 2 (7)
reduced theory is GR with non-zero cosmological constant, and the 
value of the cosmological constant is directly related to the size of + (dφ B + φ d A B ) W + φ F B + dc .
the S 3 . As the 5D Kaluza–Klein story makes the electric charge a
dynamically determinable quantity, at least in principle, via some Here d A B = g −1 (dB + AB − BA) g is the lift to the bundle of the
“spontaneous compactification” mechanism, in our setup the 4D covariant derivative of Lie algebra-valued 2-form B with respect to
cosmological constant becomes in principle determinable by the the connection A. Another simple computation using some trace
dynamics of the extra dimensions. identities gives
Thus, our 7D lift of 4D GR makes the 4D cosmological con-
 
1 2
stant a dynamical object. Having said this, we must also say that CdC = − Tr(m3 ) (8)
2 3
in this letter we limit ourselves to just demonstrating the relation M SU(2)
between the radius of S 3 and . No attempt at studying the dy- 
namics of the extra dimensions (and thus predicting ) will be × −2 Tr(φ BF + (φ /2)BB) + φ 3 dc .
4 2

made. Still, this should be kept in mind as one of the motivations

M
for our construction, in addition to those described in the begin-
ning of this text. We learn that the dimensional reduction of the first, topological
After these motivational remarks, we are ready to describe the term in (4), modulo the prefactor equal to the volume of SU(2),
dimensional reduction. We phrase the discussion that follows in is the so-called BF theory with a -term, coupled to the scalar
terms of real objects. In this case the dimensionally reduced theory and 3-form fields. We find this result interesting in its own right.
is the Riemannian signature GR. All objects can also be complexi- The dimensional reduction of the topological theory is topological.
fied, in this case one obtains complexified GR. The subtler issue of Thus, if there is no second term in (4), varying with respect to c
reality conditions relevant for the Lorentzian signature theory will gives φ = const, and we recover the usual Lagrangian of the topo-
be treated elsewhere. logical BF theory with the  term.
To carry out the dimensional reduction, we need to assume that Let us now understand the dimensional reduction of the second
the 3-form C in 7D is “independent” of 3 of the 7 coordinates. The term in (4). This is a no-derivative term, so it only changes the “po-
appropriate for our purposes way of doing this is to assume that tential” for the φ, B , c fields. In this paper we will set φ = const
we have the group SU(2) that acts on M freely. This gives M and c = 0. The complete dimensional reduction is carried out in
the structure of an SU(2) principal bundle over a 4-dimensional [14]. Setting the size of the extra dimensions to a constant corre-
base M. Our considerations here are local, over a region in M. sponds to φ = const. At the same time, it is clear from (8) that
We parametrise fibre points as g ∈ SU(2), with the group action the 3-form field c is “conjugate” to φ and so setting this field to
on the fibre being the right action of SU(2) on itself. Denote by constant justifies setting c to zero.
m = g −1 dg the Maurer–Cartan one-forms on SU(2). These forms To compute the volume form corresponding to (6) (with c = 0)
transform covariantly under the right action of SU(2) on itself. we need to write this 3-form in SO(3) notations. This is achieved
To establish further notations, let A be an SU(2) connection on by decomposing all matrix-valued fields in terms of the SU(2) gen-
the base M, i.e. a 2 × 2 anti-Hermitian matrix valued one-form erators τ i = −(i/2)σ i , where σ i are the usual Pauli matrices. So,
on M, and let A = g −1 Ag be its lift into the total space of the we write W = W i τ i etc. This gives
bundle. Then W = m + A is the connection one-form in the to-
φ3
tal space of the bundle. Simple standard computation shows that C=  i jk W i W j W k + φ W i B i . (9)
F := dW + W W is a 2-form that is purely horizontal F = g −1 Fg, 6
where F = dA + AA is the curvature of the connection one-form on The metric g C and thus the volume form volC are then easiest
the base. Here and in what follows, for brevity, we omit the wedge computed by putting this C into its canonical form. As the canon-
product symbol. ical form we take
A general SU(2) invariant 3-form on X can be written as C = 1
Tr(φ m3 + Am2 + Bm) + c. Here φ ∈ 0 ( M ), c ∈ 3 ( M ), while A , B C=  i jk e i e j ek + e i i
. (10)
6
are lifts to the total space of the bundle of Lie algebra valued 1-
i
and 2-forms on the base M respectively. Note that none of the 35 Here is the basis of anti-self-dual 2-forms
 K. Krasnov / Physics Letters B 772 (2017) 300–305 303

1
= e 45 − e 67 , 2
= e 46 − e 75 , 3
= e 47 − e 56 . where the meaning of the hat symbol over the equality sign is “on-
shell”. Indeed, varying the right-hand-side with respect to H i j we
The notation here is e = e e , with wedge product implied. It is
ab a b
get X i j = μ det( H )( H −1 )i j . The condition det( H ) = 1 imposed by
then easy to check that for C in its canonical form (10), the metric
7 the Lagrange multiplier μ then sets μ = (det( X ))1/3 . Substituting
defined by C via (2) is ds2C = a=1 (ea )2 .
the resulting solution H i j = (det( X ))1/3 ( X −1 )i j into the first term
To compute the metric for (9) we need to rewrite it in
we reproduce the left-hand-side.
the canonical form (10). This is done by choosing a convenient
Using the above way of writing the last term in (17) we can
parametrisation of B i fields. To establish this parametrisation, we
rewrite the effective 4D Lagrangian as follows
note that the triple of 2-forms B i defines a metric on the base
in which these forms are anti-self-dual (ASD). This is the Urban- 1
L 4D /φ 4 = B i F i − Mij Bi B j (18)
tke metric [16]. In fact, a simple calculation with the formula (2) 2
shows that the Urbantke formula − 2μ(det(I + φ 2 M ) − φ 3 )vol .
1
g (ξ, η)vol = −  i jk i ξ i
∧ iη j
∧ k
(11) Here we defined a new matrix M i j so that φ H = I + φ 2 M, and
6 redefined the Lagrange multiplier μ. The key point now is that the
arises as the metric on the base from (2), with C in its canonical constraint det(I + φ 2 M ) = const, when expanded in powers of M,
form (10). This clearly points towards a 7-dimensional origin of the approximates the constraint Tr( M ) = const, and this is known to
Urbantke formula. The 2-forms B i can then always be parametrised give General Relativity in its Plebanski formulation [4,5].
as For readers lost in the sequence of field redefinitions per-
√ ij
formed to achieve (18) we describe in words the origin of all
Bi = X j
, (12) the fields. The field φ is the scalar field parametrising the size
of the extra dimensional S 3 in the 3-form (6). The 2-form field
where one uses an arbitrary branch of the square root of a sym-
B i is also explicitly present in (6) as B = g −1 B i τ i g. The object
metric matrix X i j , and two-forms i are orthonormal i ∧ j ∼
F i := d A i + (1/2) i jk A j A k is the curvature of the SO(3) connection
δ i j . Here i is an orthonormal basis of ASD 2-forms for the metric A i defined by C , with W from (6) given by W = m + g −1 A i τ i g.
defined (via Urbantke formula) by B i . The matrix X i j is defined as The auxiliary fields M i j and μ where introduced in such a way
that of the wedge products of B i . We have B i ∧ B j = −2 X i j vol , that their elimination by their algebraic field equations produces
where vol is the volume form of the metric whose ASD 2-forms the dimensional reduction of the “potential term”, i.e. the last term
are i . Substituting the parametrisation (12) into (9) we see that in (4). The volume form vol is defined by B i via formulas (12)
the 3-form can be written in the following way and (11).
  We now claim that (18) describes Riemannian signature GR,
1
C =ρ  i jk e i e j ek + e i i
, (13) plus higher order corrections immaterial in the regime of not too
6
high Weyl curvatures. To see this, let us remind the reader the Ple-
with banski Lagrangian [4]
   
φ√ ij 1 
ρ = (det( X ))1/4 , e i = X W j. (14) L Pleb = M 2p Bi F i − ij
+ δi j B i B j . (19)
ρ 2 3
This puts C into a form that is a multiple of the canonical. The Here M 2p = 1/8π G is the Planck mass, G is the Newton’s, and 
metric g C is then ρ 2/3 time the metric for which the above ea is
is the cosmological constant. Here B i is a dimensionless field that
the frame. This gives
describes the metric (via Urbantke formula). We now absorb the
Xij Planck mass so as to make B i (and thus the metric) dimensionful.
ds2C = φ 2 W i W j + (det( X ))1/6 ds2 . (15) The dimensionful metric measures distances in units of the Planck
(det( X ))1/3 length. Thus, we redefine B i → B i / M 2p , → M 2p ,  → M 2p .
The volume form for this metric is  
 i i 1 ij  ij
φ 3 L Pleb = B F − + δ Bi B j. (20)
volC =  i jk W i W j W k (det( X ))1/3 vol . (16) 2 3
6
The new ,  are dimensionless. The object is the (anti-self-
We now put all pieces together and write the dimensionally dual part of) the Weyl curvature, measured in Planck units. So, it
reduced 4D Lagrangian, which is (4) on the ansatz (6) (with c = 0), satisfies 1 in all situations in which GR has been tested, or
divided by the volume of the fibre. We have can be trusted.
To exhibit close similarly to (18) we further rewrite the Pleban-
φ2
L 4D = φ 4 B i F i + B i B i + 3φ 3 (det( X ))1/3 vol . (17) ski Lagrangian by introducing an auxiliary field μ to impose the
2 constraint that i j is tracefree. Thus, we write
This is a Lagrangian of the type “BF theory plus a potential for
1
the B field”. From general considerations in [17] we know that L Pleb = B i F i − M i j B i B j − 2μ (Tr( M ) − ) vol . (21)
this is a 4D gravity theory, in the sense that the only degrees of 2
freedom it describes are, as in GR, the two polarisations of the The only difference between this Lagrangian and (18), apart from
graviton. We also know [18] that for sufficiently small Weyl cur- the overall factor, is that the constraints imposed on the auxiliary
vatures (i.e. sufficiently low energy) any such theory of gravity matrix M i j are different.
reduces to GR. To quantify the difference, we parametrise the matrix M i j via
We would now like to quantify the regime in which the above 
theory reduces to GR. To this end, let us rewrite the last term in Mij = ij
+ δi j . (22)
3
(17) by introducing two auxiliary fields. We have
Here i j is the tracefree part of M i j and ( ) is the function
6(det( X ))1/3 vol =
ˆ − H i j B i B j − 2μ(det( H ) − 1)vol , to be found by imposing the constraint obtained by varying the
304 K. Krasnov / Physics Letters B 772 (2017) 300–305

Lagrangian with respect to μ. Thus, in the case of GR in Plebanski form of the product of Minkowski spacetime and (compact) inter-
formalism the constraint simply states that  is a constant — the nal manifolds. For this reason one usually extends the pure gravity
cosmological constant. In the case of the theory (18) one obtains  theory in 4 + D dimensions with extra fields, e.g. by considering
as a non-trivial function of i j instead. This is the parametrisation the Einstein–Yang–Mills system. The stress–energy tensor of these
in which this class of 4D gravity theories was discovered in [18]. extra fields then allows for solutions of the required product form,
For (18) the constraint gives see e.g. [20], Section 3. Probably the most famous compactifica-
 3   tion mechanism is that due to Freund and Rubin [21], where the
φ 2 3 φ 2 3-form field of the 11D supergravity is doing the job. In contrast,
1+ − 1+ φ 4 Tr( 2
) (23)
3 2 3 the theory (4) admits the solution that is the S 3 fibration over S 4 ,
see [14] for an explicit description. Thus, at least there is a solution
+ φ 6 det( ) = φ 3 .
of (4) of the desired type without having to introduce extra fields.
We can use this as an equation to solve for  in terms of the However, the cosmological constant for the S 3 fibration over S 4
tracefree part. Under assumption that 1 and to order 2 the solution is too large, see [14]. This is similar to the situation with
solution is the Freund–Rubin solution.
( ) φ−1 φ Thus, a compactification mechanism that would result in an
2 3
= + Tr( ) + O( ). (24) appropriately small cosmological constant is a very serious open
3 φ2 2
issue for our setup. It is possible that the only way forward is to
It is now clear that (18) is the theory of the same type as (20), add other fields. We then remark that there is a very natural ex-
the only difference being that  is not a constant, but a func- tension of the theory (4) that adds forms of all odd degrees. This is
tion ( ) given by (24). However, for 1 the dependence of the theory that appeared in [12], formula (29). It would be inter-
( ) on in (24) can be neglected and ( ) becomes a con- esting to study 4D compactifications of this more general theory.
stant. This shows that the theory (18) is indistinguishable from GR We hope to analyse this in the future.
in its form (21) in all situations where GR has been tested and/or Another open problem of the present approach is that of cou-
can be trusted. Note that the O ( 2 ) term in (24) is neglected as pling to matter. Again, a natural way to proceed is suggested by
compared to i j term in (22), not as compared to the constant, supergravity. One does not couple supergravity to extra fields, one
which can be small. Detailed study of effects of modification of simply studies what the modes already present become when
GR such as (24) on e.g. the spherically symmetric solution can be viewed from the 4D perspective. In particular, when compactifying
found in [19]. on a coset manifold all modes related to isometries of the inter-
The above argument can be rephrased as follows. The theory (4) nal space are known to be important. Indeed, recall that the gauge
or (18) is a classical theory with no scale in it. The metric that the group that arises in the KK compactification is the group of isome-
field B i determines measures distances that are dimensionless. As tries of the internal manifold, and its dimension may be larger
(24) shows, this theory behaves as GR with non-zero dimension- than the dimension of the internal space itself. In this paper we
less cosmological constant (φ − 1)/φ 2 for small 1 values of have considered a compactification on a group manifold, but only
the dimensionless Weyl curvature. To compare this with the usual retained half of the relevant isometries by considering the invari-
GR in which the distances are dimensionful, one needs to intro- ant dimensional reduction ansatz. It is clear that additional fields
duce a scale into theory (18). Alternatively one can redefine all GR will arise by enlarging the ansatz by taking into account all the
fields using some scale to make GR quantities dimensionless, as isometries. In this case, however, one must be careful about the is-
we have done in the passage from (19) to (21). This scale can be sue of consistent truncation, see e.g. [22] for a clear description of
chosen to be the Planck scale, and then the theory (18) will be all the issues arising. We leave a study of the dimensional reduc-
indistinguishable from GR for curvatures smaller than Planckian. tion on S 3 viewed as a coset S 3 = SO(4)/SO(3) to future research.
Thus, choosing the scale for identifying (18) with GR to be the Third, there is a question of how to describe Lorentzian sig-
Planck scale, the dimensionless  in (24) becomes the cosmolog- nature metrics using this formalism. To do this one must make
ical constant measured in Planck units, and is the extraordinary the 3-form C complex-valued, and then impose some appropriate
small number  ∼ 10−120 that embodies the cosmological con- reality conditions. Similar issues exist in all Plebanski-related for-
stant problem. Our theory (18) gives small cosmological constant mulations. We postpone their resolution to future work.
for values of radius of compactification φ close to unity. This must Finally, to avoid confusion, we would like to say that our
hold to extraordinary high accuracy present use of G 2 structures (3-forms in 7D) is different from what
 one can find in the literature on Kaluza–Klein compactifications of
φ−1= , (25) supergravity. In our approach a 3-form is not an object that ex-
3M 2p
ist in addition to the metric — it is the only object that exist. The
where we now reinstated the Planck mass so that this is the usual metric, and in particular the 4D metric, is defined by the 3-form
dimensionful , and omitted higher order terms. In (24) one can via (2). Also, in the supergravity context a 7D manifold with a G 2
also get small  for large φ , but this gives large GR modifications structure is used for compactifying the 11D supergravity down to
as in this case the Tr( 2 ) term in (24) can no longer be ignored. 4D. In contrast, we compactify from 7D to 4D.
There are several open questions that need to be addressed to
convert the model studied here into a realistic theory. First and Acknowledgements
foremost, one must find a dynamical mechanism for driving the
compactification radius φ to unity to produce a small cosmological The author was supported by ERC Starting Grant 277570-DIGT.
constant. Similar issue is present in the usual Kaluza–Klein sce- He acknowledges the hospitality of the Isaac Newton Institute,
narios where one needs to provide a mechanism for spontaneous Cambridge. These ideas were envisaged during the programme
compactification. “Twistors, gravity and amplitudes” hosted by INI. The author also
We note, however, that the situation in theory (4) is some- acknowledges support of the Albert Einstein Institute, Golm (Pots-
what better than in the usual KK setup. In the latter case, apart dam). This work has been completed during a research stay at AEI.
from the case of compactification on S 1 , the pure gravity the- Discussions with Yannick Herfay and Yuri Shtanov on the topics of
ory in 4 + D dimensions usually does not have solutions of the relation between 7D and 4D are gratefully acknowledged.
 K. Krasnov / Physics Letters B 772 (2017) 300–305 305

References [12] N. Nekrasov, A la recherche de la M-theorie perdue Z theory: chasing M / f

theory, arXiv:hep-th/0412021.
[1] L. Freidel, K. Krasnov, R. Puzio, BF description of higher dimensional [13] C. Beasley, E. Witten, A note on fluxes and superpotentials in M theory com-
gravity theories, Adv. Theor. Math. Phys. 3 (1999) 1289, arXiv:hep-th/ pactifications on manifolds of G(2) holonomy, J. High Energy Phys. 0207 (2002)
9901069. 046, http://dx.doi.org/10.1088/1126-6708/2002/07/046, arXiv:hep-th/0203061.
[2] Y. Herfray, K. Krasnov, C. Scarinci, 6D interpretation of 3D gravity, Class. [14] K. Krasnov, Dynamics of 3-forms in seven dimensions, arXiv:1705.01741 [hep-
Quantum Gravity 34 (4) (2017) 045007, http://dx.doi.org/10.1088/1361-6382/ th].
aa5727, arXiv:1605.07510 [hep-th]. [15] M.J. Duff, B.E.W. Nilsson, C.N. Pope, Kaluza–Klein supergravity, Phys. Rep. 130
[3] Y. Herfray, K. Krasnov, Topological field theories of 2- and 3-forms in six di- (1986) 1, http://dx.doi.org/10.1016/0370-1573(86)90163-8.
mensions, arXiv:1705.04477 [hep-th]. [16] H. Urbantke, On integrability properties of SU(2) Yang–Mills fields. I. Infinitesi-
[4] J.F. Plebanski, On the separation of Einsteinian substructures, J. Math. Phys. 18 mal part, J. Math. Phys. 25 (1984) 2321.
(1977) 2511. [17] K. Krasnov, Plebanski gravity without the simplicity constraints, Class. Quan-
[5] K. Krasnov, Plebanski formulation of general relativity: a practical introduction, tum Gravity 26 (2009) 055002, http://dx.doi.org/10.1088/0264-9381/26/5/
Gen. Relativ. Gravit. 43 (2011) 1, http://dx.doi.org/10.1007/s10714-010-1061-x, 055002, arXiv:0811.3147 [gr-qc].
arXiv:0904.0423 [gr-qc]. [18] K. Krasnov, Renormalizable non-metric quantum gravity?, arXiv:hep-th/
[6] N.J. Hitchin, Stable forms and special metrics, arXiv:math/0107101 [math-dg]. 0611182.
[7] I. Agricola, Old and new on the exceptional group G 2 , Not. Am. Math. Soc. 55 [19] K. Krasnov, Y. Shtanov, Non-metric gravity. II. Spherically symmetric solu-
(2008) 922. tion, missing mass and redshifts of quasars, Class. Quantum Gravity 25 (2008)
[8] T. Friedrich, I. Kath, A. Moroianu, U. Semmelmann, On nearly parallel G(2) 025002, http://dx.doi.org/10.1088/0264-9381/25/2/025002, arXiv:0705.2047
structures, J. Geom. Phys. 23 (1997) 259–286. [gr-qc].
[9] D.A. Salamon, T. Walpuski, Notes on the octonions, arXiv:1005.2820 [math.RA]. [20] D. Bailin, A. Love, Kaluza–Klein theories, Rep. Prog. Phys. 50 (1987) 1087,
[10] R.L. Bryant, Metrics with exceptional holonomy, Ann. Math. 126 (1987) http://dx.doi.org/10.1088/0034-4885/50/9/001.
525–576. [21] P.G.O. Freund, M.A. Rubin, Dynamics of dimensional reduction, Phys. Lett. B 97
[11] A.A. Gerasimov, S.L. Shatashvili, Towards integrability of topological strings. (1980) 233, http://dx.doi.org/10.1016/0370-2693(80)90590-0.
I. Three-forms on Calabi–Yau manifolds, J. High Energy Phys. 0411 [22] M. Cvetic, G.W. Gibbons, H. Lu, C.N. Pope, Consistent group and coset reduc-
(2004) 074, http://dx.doi.org/10.1088/1126-6708/2004/11/074, arXiv:hep-th/ tions of the bosonic string, Class. Quantum Gravity 20 (2003) 5161, http://
0409238. dx.doi.org/10.1088/0264-9381/20/23/013, arXiv:hep-th/0306043.