Abstract
We study locally conformal calibrated \(G_2\)-structures whose underlying Riemannian metric is Einstein, showing that in the compact case the scalar curvature cannot be positive. As a consequence, a compact homogeneous 7-manifold cannot admit an invariant Einstein locally conformal calibrated \(G_2\)-structure unless the underlying metric is flat. In contrast to the compact case, we provide a non-compact example of homogeneous manifold endowed with a locally conformal calibrated \(G_2\)-structure whose associated Riemannian metric is Einstein and non Ricci-flat. The homogeneous Einstein metric is a rank-one extension of a Ricci soliton on the 3-dimensional complex Heisenberg group endowed with a left-invariant coupled \(\mathrm{SU}(3)\)-structure \((\omega , \Psi )\), i.e., such that \(d \omega = c \mathrm{Re}(\Psi )\), with \(c \in {\mathbb {R}}- \{ 0 \}\). Nilpotent Lie algebras admitting a coupled \(\mathrm{SU}(3)\)-structure are also classified.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
We recall that a seven-dimensional smooth manifold \(M\) admits a \(G_2\)-structure if the structure group of the frame bundle reduces to the exceptional Lie group \(G_2\). The existence of a \(G_2\)-structure is equivalent to the existence of a non-degenerate 3-form \(\varphi \) defined on the whole manifold (see for example [26]) and using this 3-form it is possible to define a Riemannian metric \(g_\varphi \) on \(M\).
If \(\varphi \) is parallel with respect to the Levi–Civita connection, i.e., \(\nabla ^{LC} \varphi =0\), then the holonomy group is contained in \(G_2\), the \(G_2\)-structure is called parallel and the corresponding manifolds are called \(G_2\)-manifolds. In this case, the induced metric \(g_\varphi \) is Ricci-flat. The first examples of complete metrics with holonomy \(G_2\) were constructed by Bryant and Salamon [6]. Compact examples of manifolds with holonomy \(G_2\) were obtained first by Joyce [24–26] and then by Kovalev [28] and by Corti, Haskins, Nordström, Pacini [12]. Incomplete Ricci-flat metrics of holonomy \(G_2\) with a 2-step nilpotent isometry group \(N\) acting on orbits of codimension 1 were obtained in [9, 20]. It turns out that these metrics are locally isometric (modulo a conformal change) to homogeneous metrics on solvable Lie groups, which are obtained as rank one extensions of a six-dimensional nilpotent Lie group endowed with an invariant \(\mathrm{SU}(3)\)-structure of a special kind, known in the literature as half-flat [10].
Examples of compact and non-compact manifolds endowed with non-parallel \(G_2\)-structures were given for instance in [1, 7–9, 14, 15, 17, 19, 27, 38]. In particular, in [9] conformally parallel \(G_2\)-structures on solvmanifolds, i.e., on simply connected solvable Lie groups, were studied. More in general, in [23] it was shown that a seven-dimensional compact Riemannian manifold \(M\) admits a locally conformal parallel \(G_2\)-structure if and only if it has as covering a Riemannian cone over a compact nearly Kähler 6-manifold such that the covering transformations are homotheties preserving the corresponding parallel \(G_2\)-structure.
By [5, 11, 18], it is evident that the Riemannian scalar curvature of a \(G_2\)-structure may be expressed in terms of the \(3\)-form \(\varphi \) and its derivatives. More precisely, in [5] an expression of the Ricci curvature and the scalar curvature in terms of the four intrinsic torsion forms \(\tau _i, i =0, \ldots , 3\), and their exterior derivatives was given. Moreover, using this it is possible to show that the scalar curvature has a definite sign for certain classes of \(G_2\)-structures.
If \(d\varphi = 0\), the \(G_2\)-structure is called calibrated or closed. The geometry of this family of \(G_2\)-structures was studied in [11]. Furthermore, Bryant proved in [5] that if the scalar curvature of a closed \(G_2\)-structure is non-negative then the \(G_2\)-structure is parallel.
We say that a \(G_2\)-structure \(\varphi \) is Einstein if the underlying Riemannian metric \(g_{\varphi }\) is Einstein. In [5, 11] it was proved, as an analogous of Goldberg conjecture for almost-Kähler manifolds, that on a compact manifold an Einstein (or, more in general, with divergence-free Weyl tensor [11]) calibrated \(G_2\)-structure has holonomy contained in \(G_2\). In the non-compact case, Cleyton and Ivanov showed that the same result is true with the additional assumption that the \(G_2\)-structure is \(*\)-Einstein, but it still an open problem to see if there exist (even incomplete) Einstein metrics underlying calibrated \(G_2\)-structures. Recently, some negative results were proved in the case of non-compact homogeneous spaces in [16]. In particular, the authors showed that a seven-dimensional solvmanifold cannot admit any left-invariant calibrated \(G_2\)-structure inducing an Einstein metric \(g_\varphi \) unless \(g_\varphi \) is flat.
In the present paper, we are mainly interested in the geometry of locally conformal calibrated \(G_2\)-structures, i.e., \(G_2\)-structures whose associated metric is conformally equivalent (at least locally) to the metric induced by a calibrated \(G_2\)-structure.
In Sect. 3, we prove that a compact manifold endowed with an Einstein locally conformal calibrated \(G_2\)-structure has non-positive scalar curvature (and then has either zero or negative curvature if it is also connected) and we show that a compact homogeneous 7-manifold cannot admit an invariant Einstein locally conformal calibrated \(G_2\)-structure unless the underlying metric is flat.
In the last section, we give a non-compact example of a homogeneous manifold endowed with an Einstein locally conformal calibrated \(G_2\)-structure. The homogeneous manifold is a solvmanifold, thus this example and the aforementioned result of [16] highlight a different behaviour of calibrated and locally conformal calibrated \(G_2\)-structures. Moreover, the homogeneous Einstein metric is a rank-one extension of a Ricci soliton on the complex Heisenberg group induced by a coupled \(\mathrm{SU}(3)\)-structure \((\omega , \Psi )\) such that \(d \omega = - \mathrm{Re} (\Psi )\). Recall that a half-flat \(\mathrm{SU}(3)\)-structure is said to be coupled if \(d\omega \) is proportional to \(\mathrm{Re} (\Psi )\) at each point (see [37]). Finally, we classify nilpotent Lie groups admitting a left-invariant coupled \(\mathrm{SU}(3)\)-structure, showing that the complex Heisenberg group is, up to isomorphisms, the only nilpotent Lie group admitting a coupled \(\mathrm{SU}(3)\)-structure \((\omega , \Psi )\) whose associated metric is a Ricci soliton.
2 Preliminaries on \(G_2\) and \(\mathrm{SU}(3)\)-structures
Let \(\left( e_1, \ldots , e_7\right) \) be the standard basis of \({\mathbb {R}}^7\) and \(\left( e^1, \ldots , e^7\right) \) be the corresponding dual basis. We set
where for simplicity \(e^{ijk}\) stands for the wedge product \(e^i \wedge e^j \wedge e^k\) in \(\Lambda ^3 ({({\mathbb {R}}^7)}^*)\). The subgroup of \(\mathrm{GL}(7, {\mathbb {R}})\) fixing \(\varphi \) is \(G_2\). The basis \((e^1,\ldots , e^7)\) is an oriented orthonormal basis for the underlying metric and the orientation is determined by the inclusion \(G_2\subset \mathrm{SO}(7)\). The group \(G_2\) also fixes the 4-form
where \(*\) denotes the Hodge star operator determined by the associated metric and orientation.
We recall that a \(G_2\)-structure on a \(7\)-manifold \(M\) is characterized by a positive 3-form \(\varphi \). Indeed, it turns out that there is a \(1-1\) correspondence between \(G_2\)-structures on a 7-manifold and 3-forms for which the bilinear form \(B_{\varphi }\) defined by
is positive definite, where \(i_{X}\) denotes the contraction by \(X\). A 3-form \(\varphi \) for which \(B_{\varphi }\) is positive definite defines a unique Riemannian metric \(g_{\varphi }\) and volume form \(dV_\varphi \) such that for any couple of vectors \(X\) and \(Y\) on \(M\) the following relation holds
As in [11], we let
and define the \(*\)-Ricci tensor of the \(G_2\)-structure as
A \(G_2\)-structure is said to be \(*\)-Einstein if the traceless part of the \(*\)-Ricci tensor vanishes, i.e., if \(\rho ^* = \frac{s^*}{7} g\), where \(s^*\) is the trace of \(\rho ^*\).
On a \(7\)-manifold endowed with a \(G_2\)-structure, the action of \(G_2\) on the tangent spaces induces an action of \(G_2\) on the exterior algebra \({\Lambda }^p(M)\), for any \(p \ge 2\). In [4], it was shown that there are irreducible \(G_2\)-module decompositions
where \(\Lambda ^p_k ({({\mathbb {R}}^7)}^*)\) denotes an irreducible \(G_2\)-module of dimension \(k\). Using the previous decomposition of \(p\)-forms, in [5] a simple expression of \(d\varphi \) and \(d*\varphi \) was obtained, where \(*\) denotes the Hodge operator defined by the metric \(g_\varphi \) and the volume form \(dV_\varphi \). More precisely, for any \(G_2\)-structure \(\varphi \) there exist unique differential forms \(\tau _0 \in \Lambda ^0 (M),\) \(\tau _1 \in \Lambda ^1 (M),\) \(\tau _2 \in \Lambda ^2_{14} (M),\) \(\tau _3 \in \Lambda ^3_{27} (M),\) such that
where \(\Lambda ^p_k (M)\) denotes the space of sections of the bundle \(\Lambda ^p_k (T^* M)\).
In the case of a closed \(G_2\) structure we have
By the results of [5], the scalar curvature is given by
and from this it is clear that it cannot be positive.
For a locally conformal calibrated \(G_2\)-structure \(\varphi \) one has \(\tau _0\equiv 0\) and \(\tau _3\equiv 0\), so
and taking the exterior derivative of the former it is easy to show that \(\tau _1\) is a closed 1-form. Moreover, in this case the scalar curvature has not a definite sign as one can check from its expression
where \(\delta \) denotes the adjoint of the exterior derivative \(d\) with respect to the metric \(g_\varphi \).
If the only nonzero intrinsic torsion form is \(\tau _1\), we have the so called locally conformal parallel \(G_2\)-structures. They are named in this way since a conformal change of the metric \(g_\varphi \) associated to a \(G_2\)-structure of this kind gives (at least locally) the metric induced by a parallel \(G_2\)-structure. In this case
We will give an example of such a structure at the end of Sect. 4.
We recall that a six-dimensional smooth manifold admits an \(\mathrm{SU}(3)\)-structure if the structure group of the frame bundle can be reduced to \(\mathrm{SU}(3)\). It is possible to show that the existence of an \(\mathrm{SU}(3)\)-structure is equivalent to the existence of an almost Hermitian structure \((h, J, \omega )\) and a unit \((3,0)\)-form \(\Psi \).
Since \(\mathrm{SU}(3)\) is the stabilizer of the transitive action of \(G_2\) on the 6-sphere \(S^6,\) it follows that a \(G_2\)-structure on a 7-manifold induces an \(\mathrm{SU}(3)\)-structure on any oriented hypersurface. If the \(G_2\)-structure is parallel, then the \(\mathrm{SU}(3)\)-structure is half-flat [10]. In terms of the forms \((\omega , \Psi )\) this means \(d (\omega \wedge \omega ) =0, d (\mathrm{Re}(\Psi ))=0\).
In our computations we will use another characterization of \(\mathrm{SU}(3)\)-structures which follows from the results of [22, 36]. We describe it here. Consider a six-dimensional oriented real vector space \(V\), a \(k\)-form on \(V\) is said to be stable if its GL(V)-orbit is open. Let \(A:\Lambda ^5(V^*) \rightarrow V\otimes \Lambda ^6(V^*)\) denote the canonical isomorphism given by \(A(\gamma ) = w \otimes \Omega \), where \(i_w\Omega = \gamma \), and define for a fixed 3-form \(\sigma \in \Lambda ^3(V^*)\)
and
A 3-form \(\sigma \) is stable if and only if \(\lambda (\sigma )\ne 0\) and whenever this happens it is possible to define a volume form by \(\sqrt{|\lambda (\sigma )|} \in \Lambda ^6(V^*)\), where the positively oriented root is chosen, and an endomorphism
which is a complex structure when \(\lambda (\sigma )<0\).
A pair of stable forms \((\omega ,\sigma )\in \Lambda ^2(V^*)\times \Lambda ^3(V^*)\) is called compatible if \(\omega \wedge \sigma =0\) and normalized if \(J^*_\sigma \sigma \wedge \sigma = \frac{2}{3}\omega ^3\) (the latter identity is non-zero since a 2-form \(\omega \) is stable if and only if \(\omega ^3\ne 0\)). Such a pair defines a (pseudo) Euclidean metric \(h(\cdot ,\cdot ) = \omega (J_\sigma \cdot ,\cdot )\). As a consequence, on a six-dimensional smooth manifold \(N\) there is a one to one correspondence between \(\mathrm{SU}(3)\)-structures and pairs \((\omega ,\sigma )\in \Lambda ^2(N)\times \Lambda ^3(N)\) such that for each point \(p\in N\) the pair of forms defined on \(T_pN\) \((\omega _p,\sigma _p)\) is stable, compatible, normalized, has \(\lambda (\sigma _p)<0\) and induces a Riemannian metric \(h_p(\cdot ,\cdot ) = \omega _p(J_{\sigma _p}\cdot ,\cdot )\). In this case we have \(\Psi = \sigma + \mathrm{i}J^*_\sigma \sigma \) and, then, \(\sigma = \mathrm{Re}(\Psi )\). We refer to \(h\) as the associated Riemannian metric to the \(\mathrm{SU}(3)\)-structure \((\omega ,\sigma )\).
An \(\mathrm{SU}(3)\)-structure \((\omega ,\sigma )\) on a \(6\)-manifold \(N\) is called coupled if \(d \omega = c \sigma \), with \(c\) a non-zero real number. Note that in particular a coupled \(\mathrm{SU}(3)\)-structure is half-flat since \(d (\omega ^2) =0\) and \(d \sigma =0\) and its intrinsic torsion belongs to the space \({{\mathcal {W}}_1}^- \oplus {{\mathcal {W}}_2}^- \), where \({{\mathcal {W}}_1}^- \cong {\mathbb {R}}\) and \({{\mathcal {W}}_2}^- \cong \mathfrak {su} (3)\) (see [10]).
It is interesting to notice that the product manifold \(N \times \mathbb {R}\), where \(N\) is a \(6\)-manifold endowed with a coupled \(\mathrm{SU}(3)\)-structure \((\omega , \sigma )\), has a natural locally conformal calibrated \(G_2\)-structure defined by
Indeed,
since in local coordinates the components of \(\sigma \) are functions defined on \(N\) and thus they do not depend on \(t\). Then, \(\tau _0\equiv 0, \tau _3\equiv 0\) and \(\tau _1= \left( -\frac{1}{3}c\right) dt\).
3 Einstein locally conformal calibrated \(G_2\)-structures on compact manifolds
We will show now that a seven-dimensional, compact, smooth manifold \(M\) endowed with an Einstein locally conformal calibrated \(G_2\)-structure \(\varphi \) has \(\mathrm{Scal} (g_{\varphi }) \le 0\). It is worth observing here that, up to now, there are no known examples of smooth manifolds endowed with a locally conformal calibrated \(G_2\)-structure whose associated metric is Ricci-flat (and then has zero scalar curvature).
First of all recall that given a Riemannian manifold \((M,g)\) of dimension \(n \ge 3\) it is possible to define the so called conformal Yamabe constant \(Q(M,g)\) in the following way: set \(a_n {:}= \frac{4(n-1)}{n-2}, p_n {:}= \frac{2n}{n-2}\) and let \(C^\infty _c(M)\) denote the set of compactly supported smooth real valued functions on \(M\). Then
The sign of \(Q(M,g)\) is a conformal invariant, in particular the following characterization holds:
Proposition 3.1
If \((M,g)\) is a compact Riemannian manifold of dimension \(n \ge 3\), then \(Q(M,g)\) is negative/zero/positive if and only if \(g\) is conformal to a Riemannian metric of negative/zero/positive scalar curvature.
Using the conformal Yamabe constant it is possible to prove the following
Theorem 3.2
Let \(M\) be a seven-dimensional, compact, smooth manifold endowed with an Einstein locally conformal calibrated \(G_2\)-structure \(\varphi \). Then \(\mathrm{Scal} (g_{\varphi })\le 0\). Moreover, if \(M\) is connected, \(\mathrm{Scal}(g_\varphi )\) is either zero or negative.
Proof
Suppose that \(\mathrm{Scal}(g_\varphi ) >0\), then the 1-form \(\tau _1\) is exact. Indeed, since \(d\tau _1=0\), we can consider the de Rham class \([\tau _1] \in H^1_{\mathrm{dR}} (M)\) and take the harmonic 1-form \(\xi \) representing \([\tau _1]\), that is, \(\tau _1= \xi + df\), where \(\Delta \xi = 0\) and \(f\in C^\infty (M)\). \(\xi \) has to vanish everywhere on \(M\) since it is compact, oriented and has positive Ricci curvature. Then \(\tau _1=df\). Let us consider \(\tilde{\varphi }{:}= e^{-3f}\varphi \), it is clear that \(\tilde{\varphi }\) is a \(G_2\)-structure defined on \(M\). Moreover
so \(\tilde{\varphi }\) is a closed \(G_2\)-structure and \(\mathrm{Scal}(g_{\tilde{\varphi }}) \le 0\) by [5]. We have \(g_{\tilde{\varphi }} = e^{-2f}g_\varphi \), that is, \(g_{\tilde{\varphi }}\) is conformal to the Riemannian metric \(g_\varphi \) of positive scalar curvature, then the conformal Yamabe constant \(Q(M, g_{\tilde{\varphi }})\) is positive by the previous characterization.
Since \(M\) is compact, it has finite volume and is complete as a consequence of the well known Hopf–Rinow Theorem. Then, by [34], Corollary 2.2] we have that \(Q(M, g_{\tilde{\varphi }}) \le 0\), which is in contrast with the previous result. \(\square \)
As a consequence of the previous proposition we have the
Corollary 3.3
A seven-dimensional, compact, homogeneous, smooth manifold \(M\) cannot admit an invariant locally conformal calibrated Einstein \(G_2\)-structure \(\varphi \), unless the underlying metric \(g_{\varphi }\) is flat.
Proof
Recall that a homogeneous Einstein manifold with negative scalar curvature is not compact [3]. Thus, every seven-dimensional, compact, homogeneous, smooth manifold \(M\) with an invariant \(G_2\)-structure \(\varphi \) whose associated metric is Einstein has \(\mathrm{Scal}(g_{\varphi }) \ge 0\). Combining this result with the previous proposition we have \(\mathrm{Scal} (g_{\varphi }) = 0\) and, in particular, \(g_\varphi \) is Ricci-flat. The statement then follows recalling that in the homogeneous case Ricci flatness implies flatness [2]. \(\square \)
4 Noncompact homogeneous examples and coupled \(\mathrm{SU}(3)\)-structures
In this section, after recalling some facts about noncompact homogeneous Einstein manifolds, we first study the classification of coupled \(\mathrm{SU}(3)\)-structures on nilmanifolds and then we construct an example of a locally conformal calibrated \(G_2\)-structure \(\varphi \) inducing an Einstein (non Ricci-flat) metric on a noncompact homogeneous manifold.
All the known examples of noncompact homogeneous Einstein manifolds are solvmanifolds, i.e., simply connected solvable Lie groups \(S\) endowed with a left-invariant metric (see for instance the recent survey [32]). D. Alekseevskii conjectured that these might exhaust the class of non-compact homogeneous Einstein manifolds (see [3], 7.57]).
Lauret in [33] showed that every Einstein solvmanifold is standard, i.e., it is a solvable Lie group \(S\) endowed with a left-invariant metric such that the orthogonal complement \({\mathfrak {a}}=[{\mathfrak {s}},{\mathfrak {s}}]^{{\perp }}\), where \({\mathfrak {s}}\) is the Lie algebra of \(S\), is abelian. We recall that given a metric nilpotent Lie algebra \({\mathfrak {n}}\) with an inner product \(\langle \cdot , \cdot \rangle _\mathfrak {n}\), a metric solvable Lie algebra \(({\mathfrak {s}} = {\mathfrak {n}} \oplus {\mathfrak {a}}, \langle \cdot , \cdot \rangle _\mathfrak {s})\) is called a metric solvable extension of \(({\mathfrak {n}}, \langle \cdot , \cdot \rangle _\mathfrak {n})\) if \([\mathfrak {s},\mathfrak {s}] = \mathfrak {n}\) and the restrictions to \({\mathfrak {n}}\) of the Lie bracket of \({\mathfrak {s}}\) and of the inner product \(\langle \cdot , \cdot \rangle _\mathfrak {s}\) coincide with the Lie bracket of \({\mathfrak {n}}\) and with \(\langle \cdot , \cdot \rangle _\mathfrak {n}\), respectively. The dimension of \({\mathfrak {a}}\) is called the algebraic rank of \({\mathfrak {s}}\).
In [21], 4.18], it was proved that the study of standard Einstein metric solvable Lie algebras reduces to the rank-one metric solvable extension of a nilpotent Lie algebra (i.e., those for which \(\dim ({\mathfrak {a}}) = 1\)). Indeed, by [21] the metric Lie algebra of any \((n + 1)\)-dimensional rank-one solvmanifold can be modelled on \(({\mathfrak {s}} = \mathfrak {n}\oplus \mathbb {R}H, \langle \cdot , \cdot \rangle _\mathfrak {s})\) for some nilpotent Lie algebra \(\mathfrak {n}\), with the inner product \(\langle \cdot , \cdot \rangle _\mathfrak {s}\) such that \( \langle H, {\mathfrak {n}} \rangle _\mathfrak {s} = 0\), \(\langle H,H\rangle _\mathfrak {s} = 1\) and the Lie bracket on \(\mathfrak {s}\) given by
where \([\cdot , \cdot ]_{\mathfrak {n}}\) denotes the Lie bracket on \({\mathfrak {n}}\) and \(D\) is some derivation of \(\mathfrak {n}\). By [30], a left-invariant metric \(h\) on a nilpotent Lie group \(N\) is a Ricci soliton if and only if the Ricci operator satisfies \(\mathrm{Ric}(h) = \mu I + D\), for some \(\mu \in {\mathbb {R}}\) and some derivation \(D\) of \(\mathfrak {n}\), when \(h\) is identified with an inner product on \(\mathfrak {n}\) or, equivalently, if and only if \((N, h)\) admits a metric standard extension whose corresponding standard solvmanifold is Einstein. The inner product \(h\) is also called nilsoliton.
Using the results of [29, 31], in [39] all the seven-dimensional rank-one Einstein solvmanifolds were determined, proving that each one of the 34 nilpotent Lie algebras \(\mathfrak {n}\) of dimension 6 admits a rank-one solvable extension which can be endowed with an Einstein inner product.
Six-dimensional nilpotent Lie algebras admitting a half-flat \(\mathrm{SU}(3)\)-structure were classified in [13]. For coupled \(\mathrm{SU}(3)\)-structures we can show the following
Theorem 4.1
Let \(\mathfrak {n}\) be a six-dimensional, non-abelian, nilpotent Lie algebra admitting a coupled \(\mathrm{SU}(3)\)-structure. Then \(\mathfrak {n}\) is isomorphic to one of the following
where for instance \( \mathfrak {n}_9 = (0,0,0,e^{12},e^{14}-e^{23},e^{15}+e^{34})\) means that there exists a basis \((e^1, \ldots , e^6)\) of \(\mathfrak {n}_9^*\) such that
Moreover, the only nilpotent Lie algebra admitting a coupled \(\mathrm{SU}(3)\)-structure inducing a nilsoliton is \(\mathfrak {n}_{28}\).
Proof
By the results in [13], the generic nilpotent Lie algebra \({\mathfrak {n}}\) admitting a half-flat \(\mathrm{SU}(3)\)-structure is isomorphic to one of the 24 Lie algebras described in Table 1. Consider on \({\mathfrak {n}}\) a generic 2-form
where \(b_i\in \mathbb {R}, i=1,\ldots ,15\), and the 3-form
The expression of \(\lambda (\sigma )\) for each nilpotent Lie algebra considered is given in Table 1.
We observe that among the 24 nilpotent Lie algebras admitting a half-flat \(\mathrm{SU}(3)\)-structure we have:
-
1 case \(({\mathfrak {n}}_{28})\) for which \(\lambda (\sigma )<0\) if \(b_{15}\ne 0\),
-
2 cases \(({\mathfrak {n}}_4\) and \({\mathfrak {n}}_9)\) for which the sign of \(\lambda (\sigma )\) depends on \(\omega \),
-
21 cases for which \(\lambda (\sigma )\) cannot be negative.
Therefore, the 21 algebras having \(\lambda (\sigma ) \ge 0\) do not admit any coupled \(\mathrm{SU}(3)\)-structure.
Consider \(\mathfrak {n}_4\), it has structure equations
First of all, observe that if \(b_{15}=0\) then \(\lambda (\sigma )=0\). So if we want to find an \(\mathrm{SU}(3)\)-structure we have to look for \(\omega \) with \(b_{15}\ne 0\). Moreover, \(\sigma \) induces an almost complex structure if and only if \(\lambda (\sigma )\) is negative, then we have to suppose in addition that \(b_{15}(b_{12}+b_{13})>b^2_{14}\). Since we want \(\omega \) to be the 2-form associated to an \(\mathrm{SU}(3)\)-structure, it must be a form of type \((1,1)\) and this happens if and only if \(\omega (\cdot ,\cdot )=\omega (J\cdot ,J\cdot )\), where \(J = J_{\sigma }\). Computing the previous identity with respect to the considered frame, we have that the following equations have to be satisfied by the components of \(\omega \):
(observe that \(\omega _{12}=b_1\), \(\omega _{13}=b_2\) and so on). Using these equations it is possible to write four of the \(b_i\) in terms of the remaining and obtain a new expression for \(\omega \). We can now compute the matrix associated to \(h(\cdot ,\cdot ) = \omega (J\cdot ,\cdot )\) with respect to the basis \((e_1,\ldots ,e_6)\) and observe that for the nonzero vector \(v = e_4 -\frac{b_{14}}{b_{15}}e_5 +\frac{b_{13}}{b_{15}}e_6\) we have \(h(v,v)=0\). Therefore, \(h\) cannot be positive definite and, as a consequence, it is not possible to find a coupled \(\mathrm{SU}(3)\)-structure on \(\mathfrak {n}_4\).
For the Lie algebras \(\mathfrak {n}_9\) and \(\mathfrak {n}_{28}\) we can give an explicit example of coupled \(\mathrm{SU}(3)\)-structure. Consider on \(\mathfrak {n}_9\) the forms
We have
in particular \((\omega ,\sigma )\) is a compatible pair of stable forms. The associated almost complex structure \(J = J_{\sigma }\) has the following matrix expression with respect to the basis \((e_1,\ldots ,e_6)\):
and it is easy to check that \(J^*\sigma \wedge \sigma = \frac{2}{3}\omega ^3\), i.e., the pair \((\omega ,\sigma )\) is normalized.
The inner product \(h(\cdot ,\cdot ) = \omega (J\cdot ,\cdot )\) is given with respect to the basis \((e_1,\ldots ,e_6)\) by
and it is positive definite. Therefore, we can conclude that \((\omega ,\sigma )\) is a coupled \(\mathrm{SU}(3)\)-structure on \(\mathfrak {n}_9\).
For \(\mathfrak {n}_{28}\) consider the pair of compatible, normalized, stable forms
This pair defines a coupled \(\mathrm{SU}(3)\)-structure with \(d \omega = -\sigma \). Moreover, the associated inner product
is a nilsoliton with
Summarizing our results, we can conclude that \(\mathfrak {n}_9\) and \(\mathfrak {n}_{28}\) are, up to isomorphisms, the only six-dimensional nilpotent Lie algebras admitting a coupled \(\mathrm{SU}(3)\)-structure.
We have just provided a coupled \(\mathrm{SU}(3)\)-structure on \(\mathfrak {n}_{28}\) whose associated inner product is a nilsoliton, we claim that this is the unique case among all six-dimensional nilpotent Lie algebras. It is clear that to prove the previous assertion it suffices to show that \(\mathfrak {n}_9\) does not admit any coupled \(\mathrm{SU}(3)\)-structure inducing a nilsoliton inner product. In order to do this, we consider an orthonormal basis \((e_1,\ldots ,e_6)\) of \(\mathfrak {n}_9\) whose dual basis satisfies the structure equations
(by the results of [30] and [39], these are, up to isomorphisms, the structure equations for which the considered inner product on \(\mathfrak {n}_9\) is a nilsoliton). As we did before, consider a generic 2-form \(\omega \), the 3-form \(\sigma = c(d\omega )\), evaluate \(\lambda (\sigma )\) and impose that it is negative. Then compute \(J_\sigma \) and the matrix associated to \(h(\cdot ,\cdot ) = \omega (J_\sigma \cdot ,\cdot )\) with respect to the considered basis. Since \(h\) has to be the restriction to \(\mathfrak {n}_9\) of an Einstein inner product defined on \(\mathfrak {n}_9 \oplus \mathbb {R}e_7\) and since the latter is unique up to scaling, we have to impose that the symmetric matrix associated to \(h\) is a multiple of the identity. Solving the associated equations we find that \(\lambda (\sigma )\) has to be zero, which is a contradiction. \(\square \)
Starting from a six-dimensional nilpotent Lie algebra \(\mathfrak {n}\) endowed with a coupled \(\mathrm{SU}(3)\)-structure, it is possible to construct a locally conformal calibrated \(G_2\)-structure on the rank-one solvable extension \(\mathfrak {s} = \mathfrak {n} \oplus \mathbb {R}e_7\) under some extra hypothesis. Let \(\hat{d}\) denote the exterior derivative on \(\mathfrak {n}\) and \(d\) denote the exterior derivative on \(\mathfrak {s}\). Observe that given a \(k\)-form \(\theta \in \Lambda ^k(\mathfrak {n}^*)\) we have
for some \(\rho \in \Lambda ^k(\mathfrak {n}^*)\).
Proposition 4.2
Let \(\mathfrak {n}\) be a six-dimensional, nilpotent Lie algebra endowed with a coupled \(\mathrm{SU}(3)\)-structure \((\omega ,\sigma )\) with \(\hat{d} \omega = c \sigma \), \(c \in {\mathbb {R}}- \{ 0 \}\). Consider on its rank one solvable extension \(\mathfrak {s} = \mathfrak {n} \oplus \mathbb {R}e_7\) the \(G_2\)-structure defined by \(\varphi = \omega \wedge e^7 + \sigma \), where the closed 1-form \(e^7\) is the dual of \(e_7\). Then the \(G_2\)-structure is locally conformal calibrated with \(\tau _1=\frac{1}{3}ce^7\) if and only if \(d\sigma = -2c\sigma \wedge e^7\).
Proof
Suppose that \(d\sigma = -2c\sigma \wedge e^7\), we can write \(d\omega = \hat{d}\omega + \gamma \wedge e^7\) for some 2-form \(\gamma \in \Lambda ^2 (\mathfrak {n}^*)\). We obtain \(d \varphi = ce^7\wedge \varphi .\) Then, \(\varphi \) is locally conformal calibrated with \(\tau _1= \frac{1}{3}ce^7\).
Conversely, suppose that \(\varphi \) is locally conformal calibrated with \(\tau _1=\frac{1}{3}ce^7\). Then we have \(d\varphi = ce^7 \wedge \varphi \). Moreover, we know that \(d\sigma = \hat{d}\sigma + \beta \wedge e^7 = \beta \wedge e^7\) for some 3-form \(\beta \in \Lambda ^3 (\mathfrak {n}^*)\), since \(\sigma \) is \(\hat{d}\)-closed. We then have
and comparing this with the previous expression of \(d\varphi \) we obtain
from which follows \(\beta = -2c\sigma \). \(\square \)
Now we will construct an Einstein locally conformal calibrated \(G_2\)-structure on a rank-one extension of the Lie algebra \(\mathfrak {n}_{28}\) (Lie algebra of the \(3\)-dimensional complex Heisenberg group) endowed with the coupled \(\mathrm{SU}(3)\)-structure (1).
Example 4.3
Consider \(\mathfrak {n}_{28}\) and the metric rank-one solvable extension \(\mathfrak {s} = \mathfrak {n}_{28} \oplus \mathbb {R}e_7\) with structure equations
The associated solvable Lie group \(S\) is not unimodular and so it does not admit any compact quotient [35]. Consider on \(\mathfrak {n}_{28}\) the coupled \(\mathrm{SU}(3)\)-structure \((\omega , \sigma )\) given by (1) with the nilsoliton associated inner product
Then the inner product on \(\mathfrak {s}\)
is Einstein with Ricci tensor Ric\((g) = - 3 g\).
Since \(d\sigma = 2\sigma \wedge e^7\), by the previous proposition we have a locally conformal calibrated \(G_2\)-structure on \(\mathfrak {s}\) given by
and it is easy to show that \(g_{\varphi } = g\). Then the corresponding solvmanifold \((S, \varphi )\) is an example of non-compact homogeneous manifold endowed with an Einstein (non-flat) locally conformal calibrated \(G_2\)-structure.
Observe that the \(G_2\)-structure \(\varphi \) satisfies the conditions
Then
as we expected from Proposition 4.2, and
Moreover, the \(G_2\)-structure is not \(*\)-Einstein, since by direct computation with respect to the orthonormal basis \(\left( e_1, \ldots , e_7\right) \) one has
It is worth emphasizing here that, by [16], on seven-dimensional solvmanifolds there are no left-invariant calibrated \(G_2\)-structures inducing an Einstein non-flat metric. The previous example shows that the situation is different in the case of locally conformal calibrated \(G_2\)-structures.
We provide now a non-compact example of homogeneous manifold admitting an Einstein (non-flat) locally conformal parallel \(G_2\)-structure.
Example 4.4
The Einstein rank-one solvable extension of the six-dimensional abelian Lie algebra is the solvable Lie algebra with structure equations
where \(a\) is a nonzero real number. The Riemannian metric
is Einstein with Ricci tensor given by Ric\((g) = - 6 a^2 g\).
The 3-form
has stabilizer \(G_2\), is such that \(g_\varphi = g\) and satisfies the conditions
It is immediate to show that \(\tau _1= -ae^7 \) and \(\tau _0\equiv 0, \tau _2\equiv 0, \tau _3\equiv 0\), that is, the \(G_2\)-structure \(\varphi \) is locally conformal parallel.
References
Agricola, I., Friedrich, T.: 3-Sasakian manifolds in dimension seven, their spinors and \(G_2\)-structures. J. Geom. Phys. 60(2), 326–332 (2010)
Alekseevsky, D.V., Kimelfeld, B.N.: Structure of homogeneous Riemannian spaces with zero Ricci curvature. Funkt. Anal. i Prolozěn 9, 511 (1975). Funct. Anal. Appl. 9, 97–102 (1975)
Besse, A.: Einstein Manifolds. Springer, Ergeb. Math. 10 (1987)
Bryant, R.: Metrics with exceptional holonomy. Ann. Math.(2) 126, 525–576 (1987)
Bryant, R.: Some remarks on \(G_2\)-structures. In: Proceedings of Gökova Geometry-Topology Conference 2005, Gökova Geometry/Topology Conference (GGT), Gökova, pp. 75–109 (2006)
Bryant, R., Salamon, S.: On the construction of some complete metrics with exceptional holonomy. Duke Math. J. 58, 829–850 (1989)
Cabrera, F.: On Riemannian manifolds with \(G_2\)-structures. Boll. UMI(7) 10A, 99–112 (1996)
Cabrera, F., Monar, M., Swann, A.: Classification of \(G_2\)-structures. J. Lond. Math. Soc. 53, 407–416 (1996)
Chiossi, S., Fino, A.: Conformally parallel \(G_2\) structures on a class of solvmanifolds. Math. Z. 252, 825–848 (2006)
Chiossi, S., Salamon, S.: The intrinsic torsion of SU(3) and \(G_2\) structures. In: Differential Geometry, Valencia 2001. World Sci. Publishing, River Edge, NJ, pp. 115–133 (2002)
Cleyton, R., Ivanov, S.: On the geometry of closed \(G_2\)-structures. Commun. Math. Phys. 270(1), 53–67 (2007)
Corti, A., Haskins, M., Nordström, J., Pacini, T.: \(G_2\)-manifolds and associative submanifolds via semi-Fano 3-folds. Duke Math. J. (to appear, preprint). arXiv:1207.4470
Conti, D.: Half-flat nilmanifolds. Math. Ann. 350(1), 155–168 (2011)
Fernández, M.: A family of compact solvable \(G_2\)-calibrated manifolds. Tohoku Math. J. 39, 287–289 (1987)
Fernández, M.: An example of compact calibrated manifold associated with the exceptional Lie group \(G_2\). J. Differ. Geom. 26, 367–370 (1987)
Fernández, M., Fino, A., Manero, V.: \(G_2\)-structures on Einstein solvmanifolds. Asian J. Math. 19, 321–342 (2015)
Fernández, M., Gray, A.: Riemannian manifolds with structure group \(G_2\). Ann. Mat. Pura Appl.(4) 32, 19–45 (1982)
Friedrich, Th, Ivanov, S.: Killing spinor equations in dimension 7 and geometry of integrable \(G_2\) manifolds. J. Geom. Phys. 48, 1–11 (2003)
Friedrich, Th, Ivanov, S.: Parallel spinors and connections with skew symmetric torsion in string theory. Asian J. Math. 6, 303–336 (2002)
Gibbons, G.W., Lü, H., Pope, C.N., Stelle, K.S.: Supersymmetric domain wall from metrics of special holonomy. Nuclear Phys. B 623(1–2), 3–46 (2002)
Heber, J.: Noncompact homogeneous Einstein spaces. Invent. Math. 133, 279–352 (1998)
Hitchin, N.: The geometry of three-forms in six dimensions. J. Differ. Geom. 55(3), 547–576 (2000)
Ivanov, S., Parton, M., Piccinni, P.: Locally conformal parallel \(G_2\) and \(Spin(7)\) manifolds. Math. Res. Lett. 13(2–3), 167–177 (2006)
Joyce, D.: Compact Riemannian 7-manifolds with holonomy \(G_2\). I. J. Differ. Geom. 43, 291–328 (1996)
Joyce, D.: Compact Riemannian 7-manifolds with holonomy \(G_2\). II. J. Differ. Geom. 43, 229–375 (1996)
Joyce, D.: Compact Manifolds with Special Holonomy. Oxford University Press, Oxford (2000)
Karigiannis, S.: Deformations of \(G_2\) and \(Spin(7)\) structures on manifolds. Can. J. Math. 57(5), 1012–1055 (2005)
Kovalev, A.: Twisted connected sums and special Riemannian holonomy. J. Reine Angew. Math. 565, 125–160 (2003)
Lauret, J.: Standard Einstein solvmanifolds as critical points. Q. J. Math. 52, 463–470 (2001)
Lauret, J.: Ricci soliton homogeneous nilmanifolds. Math. Ann. 319(4), 715–733 (2001)
Lauret, J.: Finding Einstein solvmanifolds by a variational method. Math. Z. 241, 83–99 (2002)
Lauret, J.: Einstein solvmanifolds and nilsolitons. In: New Developments in Lie Theory and Geometry, Contemp. Math., vol. 491, Am. Math. Soc., Providence, RI, pp. 1–35 (2009)
Lauret, J.: Einstein solvmanifolds are standard. Ann. Math. 172, 1859–1877 (2010)
Leung, M.C.: Conformal invariants of manifolds of non-positive scalar curvature. Geom. Dedic. 66, 233–243 (1997)
Milnor, J.: Curvatures of left invariant metrics on Lie groups. Adv. Math. 21(3), 293–329 (1976)
Reichel, W.: Über die Trilinearen Alternierenden Formen in 6 und 7 Veränderlichen. Dissertation, Greifswald (1907)
Salamon, S.: A tour of exceptional geometry. Milan J. Math. 71, 59–94 (2003)
Verbitsky, M.: An intrinsic volume functional on almost complex 6-manifolds and nearly Kähler geometry. Pac. J. Math. 235(2), 323–344 (2008)
Will, C.: Rank-one Einstein solvmanifolds of dimension 7. Differ. Geom. Appl. 19, 253–379 (2003)
Acknowledgments
The authors would like to thank Simon Salamon, Uwe Semmelmann and Luigi Vezzoni for useful comments on the paper. Aspects of the work were described at the Workshop “\(G_2\) days”(King’s and University College, London, 2012).
Author information
Authors and Affiliations
Corresponding author
Additional information
Research partially supported by the project PRIN Varietà reali e complesse: geometria, topologia e analisi armonica, the project FIRB Differential Geometry and Geometric functions theory and by GNSAGA (Indam) of Italy.
Rights and permissions
About this article
Cite this article
Fino, A., Raffero, A. Einstein locally conformal calibrated \(G_2\)-structures. Math. Z. 280, 1093–1106 (2015). https://doi.org/10.1007/s00209-015-1468-x
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00209-015-1468-x