Abstract
We consider balanced metrics on complex manifolds with holomorphically trivial canonical bundle, most commonly known as balanced SU(n)structures. Such structures are of interest for both Hermitian geometry and string theory, since they provide the ideal setting for the Hull–Strominger system. In this paper, we provide a nonexistence result for balanced nonKähler \(\text {SU}(3)\)structures which are invariant under a cohomogeneity one action on simply connected sixmanifolds.
Introduction
A \(\text {U}(n)\)structure on a 2ndimensional smooth manifold M is the data of a Riemannian metric g and a gorthogonal almost complex structure J. The pair \(\left( g,J\right) \) is also known as an almost Hermitian structure on M. When J is integrable, i.e., \(\left( M,J\right) \) is a complex manifold, the pair \(\left( g,J\right) \) defines a Hermitian structure on M. In this case, the metric g is called balanced when \(\text {d}\omega ^{n1}=0\), \(\omega {:}{=}g\left( J\cdot ,\cdot \right) \) denoting the associated fundamental form, and we shall refer to \(\left( g,J\right) \) as a balanced \(\text {U}(n)\)structure on M. Balanced metrics have been extensively studied in [4, 10,11,12,13, 23, 25] (see also the references therein).
Balanced metrics are also interesting in the context of \(\text {SU}(n)\)structures, especially in the sixdimensional case, thanks to their applications in physics. An \(\text {SU}(n)\)structure \(\left( g,J,\Psi \right) \) on a 2ndimensional smooth manifold M, is a \(\text {U}(n)\)structure \(\left( g,J\right) \) on M together with a \(\left( n,0\right) \)form of nonzero constant norm \(\Psi =\psi _+ + i\psi _\) satisfying the normalization condition \(\Psi \wedge {\overline{\Psi }}=(1)^{\frac{n(n+1)}{2}}(2i)^n \frac{\omega ^n}{n!}\). An \(\text {SU}(n)\)structure \(\left( g,J,\Psi \right) \) on M with underlying balanced \(\text {U}(n)\)structure \(\left( g,J\right) \) for which \(\text{ d }\omega \ne 0\) and \(\text {d}\Psi =0\) will be referred to as a balanced nonKähler \(\text {SU}(n)\)structure.
In 1986, Hull and Strominger [22, 30], independently, introduced a system of pdes, now known as the Hull–Strominger system, to formalize certain properties of the inner space model used in string theory. Let M be a 2ndimensional complex manifold equipped with a nowherevanishing holomorphic \(\left( n,0\right) \)form \(\Psi \) and let E be a holomorphic vector bundle on M endowed with the Chern connection. The Hull–Strominger system consists of a set of pdes involving a pair of Hermitian metrics \(\left( g,h\right) \) on \(\left( M,E\right) \). One of these equations dictates the metric g on M to be conformally balanced, more precisely \(\text {d}\left( \Vert \Psi \Vert _{\omega }\omega ^{n1}\right) =0\), where \(\Vert \Psi \Vert _{\omega }\) is the norm of \(\Psi \) given explicitly by \(\Psi \wedge {\overline{\Psi }}=(1)^{\frac{n(n+1)}{2}}\frac{i^n}{n!} \Vert \Psi \Vert _{\omega }^2 \omega ^n\). When one assumes all structures to be invariant under the smooth action of a certain Lie group G, the aforementioned condition reduces to the balanced equation \(\text {d}\omega ^{n1}=0\), since the norm of \(\Psi \) is constant. Notice that in these cases, \((g,J,\Psi )\) is a balanced \(\text {SU}(n)\)structure on M, up to a suitable uniform scaling of \(\Psi \).
The issue of the existence and uniqueness of a general solution to the Hull–Strominger system is still an open problem. Nonetheless, solutions have been found under more restrictive hypotheses; for the nonKählerian case, we refer the reader, for instance, to [6, 7, 14,15,16, 18, 24]. Other interesting solutions are given in [8], where a class of invariant solutions to the Hull–Strominger system on complex Lie groups was provided; these solutions extend to solutions on all compact complex parallelizable manifolds, by Wang’s classification theorem [33]. Moreover, in [10], it was shown that a compact complex homogeneous space with invariant complex volume admitting a balanced metric is necessarily a complex parallelizable manifold. Then, the invariant solutions given in [8] exhaust the complex compact homogeneous case. If one allows the Lie group acting on the homogeneous space to be real, many other solutions to the Hull–Strominger system are known in the literature, see for instance [18, 27, 28, 31]. Then, one may wonder what happens in the cohomogeneity one case. A cohomogeneity one manifold M is a connected smooth manifold with an action of a compact Lie group G having an orbit of codimension one. Currently, there are no known examples of balanced nonKähler \(\text {SU}(n)\)structures invariant under a cohomogeneity one action. In this paper, we investigate their existence. In particular, we focus on the simply connected \(2n=6\)dimensional case. Recall that when a cohomogeneity one manifold M has finite fundamental group, then M/G is homeomorphic to an interval I, see [5]. If we denote by \(\pi :M\rightarrow M/G\) the canonical projection onto the orbit space, we shall call \(\pi ^{1}(t)\), for every \( t\in \overset{\circ }{I}\), principal orbits and the inverse images of the boundary points singular orbits. Denoting by \(M^{\text {princ}}\) the union of all principal orbits, which is a dense open subset of M, and by K the isotropy group of a principal point, which is unique up to conjugation along \(M^{\text {princ}}\), the pair \(\left( G,K\right) \) completely determines the principal part \(M^{\text {princ}}\) of the cohomogeneity one manifold, up to Gequivariant diffeomorphisms. Given a Lie group H, we denote its Lie algebra \(\text {Lie}(H)\) by the corresponding gothic letter \(\mathfrak {h}\).
We first give a local result for the existence of balanced nonKähler \(\text {SU}(3)\)structures by working on \(M^{\text {princ}}\).
Theorem A
Let M be a 6dimensional simply connected cohomogeneity one manifold under the almost effective action of a connected Lie group G, and let K be the principal isotropy group. Then, the principal part \(M^{\text {princ}}\) admits a Ginvariant balanced nonKähler \(\text {SU}(3)\)structure \(\left( g,J,\Psi \right) \) if and only if M is compact and \(\left( \mathfrak {g}, \mathfrak {k} \right) =\left( \mathfrak {su}(2)\oplus 2\mathbb {R},\{0\} \right) \).
We then prove that none of these local solutions can be extended to a global one. This leads us to state our main theorem:
Theorem B
Let M be a sixdimensional simply connected cohomogeneity one manifold under the almost effective action of a connected Lie group G. Then, M admits no Ginvariant balanced nonKähler \(\text {SU}(3)\)structures.
In [13], balanced metrics were constructed on the connected sum of \(k \ge 2\) copies of \(S^3 \times S^3\). However, it is not known whether \(S^3 \times S^3\) admits balanced structures. In [23, Example 1.8], Michelsohn proved that \(S^3\times S^3\) endowed with the Calabi–Eckmann complex structure does not admit any compatible balanced metric. By [2, Remark 1], in a manifold with six real dimensions, there is no nonKähler Hermitian metric which is simultaneously balanced and strong Kählerwithtorsion (a.k.a SKT). In [11], Fino and Vezzoni conjectured that on nonKähler compact complex manifolds it is never possible to find an SKT metric and also a balanced metric. In [17], an example of a SKT structure on \(S^3 \times S^3\) is provided. The key case that needs to be tackled in Theorem B is precisely \(S^3 \times S^3\).
The paper is organized as follows. In Sect. 2, we review some basic facts about cohomogeneity one manifolds and \(\text {SU}(3)\)structures which will be useful for our discussion. In Sect. 3, we present our problem, write a classification of the pairs \((\mathfrak {g},\mathfrak {k})\) that can occur, and use the hypothesis of simply connectedness to reduce the list to only three possibilities. At the end of Sect. 3, we state Theorem A, which we prove in Sect. 4 via a casebycase analysis. Finally, in Sect. 5, we prove Theorem B.
Preliminary notions
Cohomogeneity one manifolds
Here, we recall the basic structure of cohomogeneity one manifolds. For further details, see for instance [1, 5, 20, 21, 34].
Definition 2.1
A cohomogeneity one manifold is a connected differentiable manifold M with an action \(\alpha :G\times M\rightarrow M\) of a compact Lie group G having an orbit of codimension one. We denote by \({\tilde{\alpha }}:G \rightarrow \text {Diff}\left( M\right) \) the Lie group homomorphism induced by the action.
From now on, let us assume that M is a simply connected cohomogeneity one manifold, and G is connected. By the compactness of G, the action \(\alpha \) is proper and there exists a Ginvariant Riemannian metric g on M; this is equivalent to saying that G acts on the Riemannian manifold \(\left( M,g\right) \) by isometries. Moreover, we assume that the action \(\alpha \) is almost effective, namely \(\text {ker}\, {\tilde{\alpha }}\) is discrete. As usual, we denote by \(\pi :M\rightarrow M/G\) the canonical projection and we equip M/G with the quotient topology relative to \(\pi \). By a result of Bérard Bergery [5], the quotient space M/G is homeomorphic to a circle or an interval. As we are assuming that M is simply connected, we have that M/G is homeomorphic to an interval I. The inverse images of the interior points of the orbit space M/G are known as principal orbits, while the inverse images of the boundary points are called singular orbits. We denote by \(M^{\text {princ}}\) the union of all principal orbits, which is an open dense subset of M, and by \(G_p\) the isotropy group at \(p\in M\).
First, we will suppose M is compact. It follows that M/G is homeomorphic to the closed interval \(I=[1,1]\). Denote by \(\mathcal {O}_1\) and \(\mathcal {O}_2\) the two singular orbits \(\pi ^{1}\left( 1\right) \) and \(\pi ^{1}\left( 1\right) \), respectively, and fix \(q_1 \in \mathcal {O}_1\). By compactness of the Gorbits, there exists a minimizing geodesic \(\gamma _{q_1}:[1,1]\rightarrow M\) from \(q_1\) to \(\mathcal {O}_2\) which is orthogonal to every principal orbit. We call a normal geodesic a geodesic orthogonal to every principal orbit. Let \(\gamma :[1,1]\rightarrow M\) be a normal geodesic between \(\pi ^{1}\left( 1\right) \) and \(\pi ^{1}\left( 1\right) \); up to rescaling, we can always suppose that the orbit space M/G is such that \(\pi \circ \gamma =\text {Id}_{[1,1]}\). Then, by Kleiner’s Lemma, there exists a subgroup K of G such that \(G_{\gamma \left( t\right) }=K\) for all \(t\in (1,1)\) and K is subgroup of \(G_{\gamma \left( 1\right) }\) and \(G_{\gamma \left( 1\right) }\).
For M noncompact, M/G is homeomorphic either to an open interval or to an interval with a closed end. In the former case, M is a product manifold \(M \cong I \times G/K\). In the latter case, there exists exactly one singular orbit, and \(M/G \cong I\) where \(I=[0, L)\) and L is either a positive number or \(+\infty \). Analogously to the compact case, there exists a normal geodesic \(\gamma : [0, L) \rightarrow M\) such that \(\gamma (0) \in \pi ^{1}(0)\) and we can suppose \(\pi \circ \gamma =\text {Id}_{[0,L)}\). In addition, there exists a subgroup K of G such that \(G_{\gamma (t)}=K\) for all \(t \in (0,L)\) and if \(H {:}{=}G_{\gamma (0)}\), K is a subgroup of H.
So we have that:

\(\pi ^{1}\left( t\right) \cong G/K\) for all \(t\in \overset{\circ }{I}\),

\(M^{\text {princ}} = \bigcup _{t\in \overset{\circ }{I}}\pi ^{1}\left( t\right) =\bigcup _{t\in \overset{\circ }{I}} G\cdot \gamma \left( t\right) \),

for every \(p_1, p_2\in M^{\text {princ}}\), \(G\cdot p_1\) and \(G\cdot p_2\) are diffeomorphic.
Therefore, up to conjugation along the orbits, when M is compact we have three possible isotropy groups \(H_1{:}{=}G_{\gamma \left( 1\right) }\), \(H_2 {:}{=}G_{\gamma \left( 1\right) }\) and \(K{:}{=}G_{\gamma \left( t\right) }\), \(t\in \left( 1,1\right) \). When M is noncompact and has one singular orbit, instead, we have two possible isotropy groups \(H {:}{=}G_{\gamma \left( 0\right) }\) and \(K{:}{=}G_{\gamma \left( t\right) }\), \(t\in \left( 0,L\right) \). From all of the above, we have that
and so, by fixing a suitable global coordinate system, we can decompose the Ginvariant metric g as
where \({\text {d}}t^2\) is the (0, 2)tensor corresponding to the vector field \(\xi {:}{=}\gamma '\left( t\right) \) evaluated at the point \(\gamma \left( t\right) \), and \(g_t\) is a Ginvariant metric on the homogeneous orbit \(G\cdot \gamma \left( t\right) \) through the point \(\gamma \left( t\right) \in M\).
Now, we will assume M is compact. By the density of \(M^{\text {princ}}\) in M and the Tube Theorem, M is homotopically equivalent to
where the geodesic balls \( S_{\gamma \left( \pm 1\right) }{:}{=}\text {exp}\left( B_{\varepsilon ^{\pm }}\left( 0\right) \right) \), \(B_{\varepsilon ^{\pm }}\left( 0\right) \subset T_{\gamma \left( \pm 1\right) }\left( G\cdot \gamma \left( \pm 1\right) \right) ^{\perp }\), are normal slices to the singular orbits in \(\gamma \left( \pm 1\right) \). Here, \( G \times _{H_i} S_{\gamma \left( \pm 1\right) }\) is the associated fiber bundle to the principal bundle \(G \rightarrow G/H_i\) with type fiber \(S_{\gamma \left( \pm 1\right) }\). By Bochner’s linearization theorem, M is also homotopically equivalent to
The isotropy groups \(H_i\) act on \( B_{\varepsilon ^{\pm }}\left( 0\right) \) via the slice representation and, since the boundary of the tubular neighborhood \(\text {Tub}(\mathcal {O}_i) {:}{=}G \times _{H_i} B_{\varepsilon ^{\pm }}\left( 0\right) \), \(i=1,2\), is identified with the principal orbit G/K, and the Gaction on \(\text {Tub}(\mathcal {O}_i) \) is identified with the \(H_i\)action on \( B_{\varepsilon ^{\pm }} \left( 0 \right) \), then \(H_i\) acts transitively on the sphere \(S^{l_i} {:}{=}\partial B_{\varepsilon ^{\pm }}\), \(l_i>0\) still having isotropy K. The normal spheres \(S^{l_i}\) are thus the homogeneous spaces \(H_i/K\), \(i=1,2\). The \(H_i\)action on \(S^{l_i}\), \(i=1,2\), may be ineffective, but it is sufficient to quotient \(H_i\) by the ineffective kernel to obtain an effective action: transitive effective actions of compact Lie groups on spheres were classified by Borel and are summarized in Table 1.
The collection of G with its isotropy groups \(G\supset H_1, H_2 \supset K \) is called the group diagram of the cohomogeneity one manifold M. Viceversa, let \(G\supset H_1, H_2 \supset K \) be compact groups with \(H_i/K=S^{l_i}\), \(i=1,2\). By the classification of transitive actions on spheres one has that the \(H_i\)action on \(S^{l_i}\) is linear and hence it can be extended to an action on \( B_{\varepsilon ^{\pm }}\) bounded by \(S^{l_i}\), \(i=1,2\). Therefore, 2.3 defines a cohomogeneity one manifold M. Analogously, if M is a noncompact cohomogeneity one manifold with one singular orbit, we define the group diagram of M to be the collection of G and the isotropy groups \(G \supset H \supset K\), where the homogeneous space H/K will be a sphere. The converse is also true: the group diagram defines a noncompact cohomogeneity one manifold M. In these cases, M is homotopically equivalent to \(G\times _{H}B_{\epsilon }(0)\), where \(B_{\epsilon }(0)\subseteq T_{\gamma (0)} (G\cdot \gamma (0))^{\perp }\) as before.
Let \(M_i\) be cohomogeneity one manifolds with respect to the action of Lie groups \(G_i\), \(i=1,2\). We say that the action of \(G_1 \) on \(M_1\) is equivalent to the action of \(G_2\) on \(M_2\) if there exists a Lie group isomorphism \(\varphi :G_1 \rightarrow G_2\) and an equivariant diffeomorphism \(f:M_1 \rightarrow M_2\) with respect to the isomorphism \(\varphi \). We shall study cohomogeneity one manifolds up to this type of equivalence.
Moreover, if a cohomogeneity one manifold M has group diagram \(G\supset H_1, H_2 \supset K\) or \(G\supset H \supset K\), one can show that any of the following operations results in a Gequivariantly diffeomorphic manifold:

(1)
switching \(H_1\) and \(H_2\),

(2)
conjugating each group in the diagram by the same element of G,

(3)
replacing \(H_i\) (respectively H) with \(aH_ia^{1}\) (respectively \(aHa^{1}\)) for \(a\in N(K)_0\), where \(N(K)_0\) is the identity component of the normalizer of K.
SU(3)structures
An \(\text {SU}(3)\)structure on a sixdimensional differentiable manifold M is the data of a Riemannian metric g, a gorthogonal almost complex structure J, and a (3, 0)form of nonzero constant norm \(\Psi =\psi _++i\psi _\) satisfying the normalization condition \(\psi _+ \wedge \psi _=\frac{2}{3}\omega ^3\).
Following a result obtained in [29] and later reformulated in [19, Section 2], one can show that giving an \(\text {SU}(3)\)structure is equivalent to giving a pair of differential forms \(\left( \omega , \psi _+ \right) \in \Lambda ^2\left( M\right) \times \Lambda ^3\left( M\right) \) satisfying suitable conditions. Here, \(\Lambda ^k\left( M\right) \) denotes the space of differential forms of degree k on M. To see this, let us briefly recall the concept of stability in the context of vector spaces.
Let V be a real sixdimensional vector space and let \(\alpha \) be a kform on V. We say that \(\alpha \) is stable if its orbit under the action of \(\text {GL}(V)\) is open in \(\Lambda ^k\left( V^*\right) \). Fix a volume form \(\Omega \in \Lambda ^6(V^*)\) on V and consider the isomorphism \(A:\Lambda ^5(V^*)\rightarrow V\otimes \Lambda ^6(V^*)\) defined for any \(\alpha \in \Lambda ^5(V^*)\) by \(A(\alpha )=v\otimes \Omega \), where \(v\in V\) is the unique vector such that \(\iota _v \Omega =\alpha \); here, \(\iota _v \Omega \) is the contraction of \(\Omega \) by the vector v. Fix a 3form \(\psi \in \Lambda ^3(V^*)\) and define
and
Finally, we define the function \(\lambda :\Lambda ^3(V^*)\rightarrow \mathbb {R}\), \(\lambda (\psi )=\iota _{\Omega \otimes \Omega }P(\psi )\).
Proposition 2.2
[19, 29] Let V be an oriented, sixdimensional real vector space. Then,

(1)
a 2form \(\omega \in \Lambda ^2(V^*)\) is stable if and only if it is nondegenerate, i.e., \(\omega ^3\ne 0\),

(2)
a 3form \(\psi \in \Lambda ^3(V^*)\) is stable if and only if \(\lambda (\psi )\ne 0\).
We denote by \(\Lambda ^3_+(V^*)\) the open orbit of stable 3forms satisfying \(\lambda (\psi )<0\). The \(\text {GL}_+(V)\)stabilizer of a 3form lying in this orbit is isomorphic to \(\text {SL}(3,\mathbb {C})\). As a consequence, every \(\psi \in \Lambda ^3_{+}(V^*)\) gives rise to a complex structure
which depends only on \(\psi \) and on the volume form \(\Omega \). Moreover, the complex form \(\psi + i J_{\psi } \psi \) is of type (3, 0) with respect to \(J_{\psi }\), and the real 3form \(J_{\psi } \psi \) is stable, too.
We say that a kform \(\alpha \in \Lambda ^k(M)\) is stable if \(\alpha _p\) is a stable form on the vector space \(T_pM\), for all \(p \in M\). Let \(\left( \omega ,\psi _+\right) \in \Lambda ^2(M)\times \Lambda ^3_+(M)\) be a pair of stable forms on M satisfying the compatibility condition \(\omega \wedge \psi _+=0\) and \(\lambda (\psi _+)<0\). Consider the almost complex structure \(J=J_{\psi _+}\) determined by \(\psi _+\) and the volume form \(\frac{\omega ^3}{6}\). Then, the 3form \(\psi _+\) is the real part of a nowherevanishing (3, 0)form \(\Psi {:}{=}\psi _+ +i \psi _\) with \(\psi _ {:}{=}J\psi _+ = \psi _+(J\cdot ,J\cdot ,J\cdot ) = \psi _+(J\cdot ,\cdot ,\cdot )\), where the last identity holds since \(\psi _+\) is of type \((3,0)+(0,3)\) with respect to J. Moreover, \(\omega \) is of type (1, 1) and, as a consequence, the (0, 2)tensor \(g {:}{=}\omega (\cdot ,J\cdot )\) is symmetric. Under these assumptions, the pair \((\omega ,\psi _+)\) defines an \(\text {SU}(3)\)structure \((g,J,\Psi )\) on M provided that g is a Riemannian metric and the normalization condition \( \psi _+ \wedge \psi _=\frac{2}{3} \omega ^3=4 \, dV_g \) is satisfied, \(dV_g\) being the Riemannian volume form. Conversely, given an \(\text {SU}(3)\)structure \((g,J,\Psi )\) on M, the pair \((\omega ,\psi _+)\) given by
satisfies the compatibility condition \(\omega \wedge \psi _+=0\) and the stability condition \(\lambda (\psi _+)<0\).
Balanced \(\text {SU}(3)\)structures
Following [9], we call an \(\text {SU}(3)\)structure \(\left( g,J,\Psi \right) \) on a 6manifold M balanced if:

J is integrable, i.e., \(\left( M,J\right) \) is a complex manifold. We recall that for \(\text {SU}(3)\)structures, the integrability of J is equivalent to requiring \((\text {d}\Psi )^{2,2}=0\),

\(\Psi \) is a holomorphic (3, 0)form,

\(\text {d}\omega ^{2}=0\), \(\omega \) being the fundamental form.
In particular, we are interested in the nonKählerian case, i.e., \(d\omega \ne 0\).
Remark 2.3
We can equivalently say that an \(\text {SU}(3)\)structure \(\left( g,J,\Psi \right) \) on M is balanced if and only if
since \(\text {d}\Psi =0\) if and only if \(\Psi =\psi _+ + i \psi _ \) is holomorphic and the induced almost complex structure \(J= J_{\psi _+}\) is integrable.
Remark 2.4
From the formulas in [3], we have that if \(\left( g,J,\Psi \right) \) is a balanced nonKähler \(\text {SU}(3)\)structure on a sixdimensional differentiable manifold M, \(\text {Scal}(g)< 0\), \(\text {Scal}(g)\) being the scalar curvature associated with the metric g.
Balanced \(\text {SU}(3)\)structures on sixdimensional cohomogeneity one manifolds
Let \(\left( g,J,\Psi \right) \) be an \(\text {SU}(3)\)structure on a simply connected cohomogeneity one manifold M of complex dimension 3 for the almost effective action of a compact connected Lie group G. We are thus requiring G to preserve the \(\text {SU}(3)\)structure on M. For the convenience of the reader, recall that

G preserves the metric g if and only if \({\tilde{\alpha }}\left( h\right) \) is an isometry for each \(h\in G\),

G preserves the almost complex structure J if and only if J commutes with the differential \( \text {d}{\tilde{\alpha }}\left( h\right) \) for each \(h \in G\),

G preserves the 3form \(\Psi \) if and only if \({\tilde{\alpha }}\left( h\right) ^*\Psi =\Psi \), for each \(h\in G\).
This in particular implies that the principal isotropy K acts on \(T_p M\) preserving \(\left( g_p,J_p,\Psi _p\right) \) for any \(p\in M\), which means that K is a subgroup of \(\text {SU}(3)\). Now, since the Jinvariant Kaction fixes the subspace \(\langle \xi _p\rangle \) of \(T_pM\), then it fixes \(\langle {J\xi }_p\rangle \) as well. Let us write \(T_p M \) as
where V is the fourdimensional \(g_p\)orthogonal complement of \(\langle \xi _p,J\xi _p\rangle \) in \(T_p M\). Notice that V is \(J_p\) and Kinvariant. To see the Kinvariance, let \(h \in K\) and \(v \in V\). Then, if \(\text {d}{\tilde{\alpha }}(h)_p \left( v \right) = \lambda \left( J \xi _p \right) +w\), for some \(\lambda \in \mathbb {R}\), \(w\in V\), we would have \(J \left( \text {d} {\tilde{\alpha }}(h)_p \left( v \right) \right) =\text {d}{\tilde{\alpha }}(h)_p \left( J_pv \right) =\lambda \,\xi _p+J_p w\), which is a contradiction since the Kaction is closed along the Gorbits. Therefore, for each \(h\in K\), its action on \(T_p M\) is described by a \(6 \times 6\) block matrix
with respect to the decomposition of \(T_p M= \langle \xi _p\rangle \oplus \langle {J \xi }_p\rangle \oplus V\). Since the matrix above is in \(\text {SU}(3)\), we have \(A \in \text {SU}(2)\) and hence K can be identified with a subgroup of \(\text {SU}(2)\). Therefore, \( \mathfrak {k}{:}{=}\text {Lie}\left( K\right) \) is \(\{0\}, \, \mathbb {R}\), or \(\mathfrak {su}\left( 2\right) \). As observed in [26], all the possible candidate pairs \(\left( \mathfrak {g},\mathfrak {k}\right) \), with \(\mathfrak {g}\) compact, which may admit an \(\text {SU}(3)\)structure in cohomogeneity one are:

(a)
if \(\mathfrak {k}=\{0\}\), then

(1)
\(\mathfrak {g}=\mathfrak {su}\left( 2\right) \oplus \mathbb {R} \oplus \mathbb {R} \),

(2)
\(\mathfrak {g}= \underbrace{\mathbb {R} \oplus \ldots \oplus \mathbb {R}}_{\text {5 times}}\),

(1)

(b)
if \(\mathfrak {k}=\mathbb {R}\), then

(1)
\(\mathfrak {g}=\mathfrak {su}\left( 2\right) \oplus \mathfrak {su}\left( 2\right) \),

(2)
\( \mathfrak {g}= \mathfrak {su}\left( 2\right) \oplus \mathbb {R} \oplus \mathbb {R} \oplus \mathbb {R} \),

(3)
\(\mathfrak {g}= \underbrace{\mathbb {R} \oplus \ldots \oplus \mathbb {R}}_{\text {6 times}}\),

(1)

(c)
if \(\mathfrak {k}=\mathfrak {su}\left( 2\right) \), then

(1)
\(\mathfrak {g}=\mathfrak {su}\left( 2\right) \oplus \mathfrak {su}\left( 2\right) \oplus \mathbb {R} \oplus \mathbb {R}\),

(2)
\(\mathfrak {g}= \mathfrak {su}\left( 2\right) \oplus \underbrace{\mathbb {R} \oplus \ldots \oplus \mathbb {R}}_{\text {5 times}}\),

(3)
\(\mathfrak {g}= \mathfrak {su}\left( 3\right) \).

(1)
Under the assumption of simply connectedness of M, we can discard some pairs of this list. If M is compact, by Hoelcher’s classification [20, Proposition 3.1], we can readily discard cases (a.2), (b.2), (b.3), (c.1), and (c.2). For the case when M is noncompact and has one singular orbit, we can suitably adapt [20, Proposition 1.8] which deals with the compact case, to obtain:
Proposition 3.1
Let M be the noncompact cohomogeneity one manifold given by the group diagram \(G \supset H \supset K\) with \(H/K = S^l\). Then, \(\pi _1(M) \cong \pi _1(G/K)/N\) where
In particular, M is simply connected if and only if the image of \(\pi _1 (S^l)\) generates \(\pi _1(G/K)\) under the natural inclusions.
We know that \(\pi _1 (S^l)\) is either \(\{ 0 \}\) or \(\mathbb {Z}\). Now, we observe that for cases (a.1) and (c.1), \(\pi _1 (G/K)= \mathbb {Z}^2\), for cases (a.2), (b.3) and (c.2), \(\pi _1 (G/K)= \mathbb {Z}^5\), and for case (b.2), \(\pi _1 (G/K)\) is either \(\mathbb {Z}^2\) or \(\mathbb {Z}^3\). If M is noncompact and has no singular orbits, \(\pi _1 (M)=\pi _1 (G/K)\). Hence, when M is noncompact, we can discard the pairs (a.1), (a.2), (b.2), (b.3), (c.1) and (c.2) as \(\pi _1(M)\) would be infinite. Therefore, the possible pairs which may admit a balanced \(\text {SU}(3)\)structure on a simply connected manifold of cohomogeneity one under the almost effective action of a compact connected Lie group G are (a.1) (only when M is compact), (b.1), and (c.3).
Remark 3.2
In case (b.1), we shall need to divide the discussion depending on the embeddings of \(\mathfrak {k}=\mathbb {R}\) in \(\mathfrak {g}=\mathfrak {su}(2)\oplus \mathfrak {su}(2)\) which, up to isomorphism, are all generated by an element of the form
with fixed \(p, q \in \mathbb {N}\). Up to uniform rescalings, which do not change the immersion of \(\mathfrak {k}\), we can assume either \((p,q)=(1,0)\) or p, q to be coprime if neither is zero. Notice that when \((p,q)=(1,1)\) or \((p,q)=(1,0)\), \(\mathfrak {k}\) induces a decomposition of \(\mathfrak {g}\) in \(\text {Ad}(K)\)modules, some of which are equivalent. In the former case, we shall say that \(\mathfrak {k}\) is diagonally embedded in \(\mathfrak {g}\), while in the latter \(\mathfrak {k}\) is said to be trivially embedded in one of the two \(\mathfrak {su}(2)\)factors of \(\mathfrak {g}\). When instead p, q are different and nonzero, the \(\text {Ad}(K)\)modules are pairwise inequivalent.
From now on, for each \(p\in M^{\text {princ}}\), let \(\mathfrak {m}_p{=}{:}\mathfrak {m}\) be an \(\text {Ad}(K)\)invariant complement of \(\mathfrak {k}\) in \(\mathfrak {g}\). For each \(p\in M^{\text {princ}}\), we have that \(T_p M=\langle \xi _p\rangle \oplus \widehat{\mathfrak {m}}_p\), where for every \(X \in \mathfrak {g}\), we denote by \({\widehat{X}}\) the action field
It is known that since \(M^{\text {princ}}\cong \overset{\circ }{I} \times G/K \), every Ginvariant structure on \(M^{\text {princ}}\) can be expressed via a Kinvariant structure on \(\langle \xi \rangle \oplus \widehat{\mathfrak {m}} \), with \(C^{\infty }(\overset{\circ }{I})\)coefficients. Let \(\mathfrak {m}=\mathfrak {m}_1 \oplus \ldots \oplus \mathfrak {m}_r\) be the decomposition of \(\mathfrak {m}\) into irreducible \(\text {Ad}(K)\)modules. Recall that if the \(\mathfrak {m}_i\)’s are pairwise inequivalent, then they are orthogonal with respect the metric \(g_t\), for every t (see 2.1). Otherwise, the expression of the metric strongly depends on the specific equivalence of the modules. In all cases, we recover the whole \(\text {SU}(3)\)structure from a pair of Ginvariant stable forms \(\left( \omega ,\psi _+\right) \) of degree two and three, respectively.
To fix the notations, in what follows, we shall denote by

\(\mathcal {B}\) the opposite of the Killing–Cartan form on \(\mathfrak {g}\),

\(\left\{ {\tilde{e}}_i\right\} _{i=1,2,3}\) the standard basis for \(\mathfrak {su}(2)\) given by
$$\begin{aligned} {\tilde{e}}_1=\begin{pmatrix} i &{} 0 \\ 0 &{} i \end{pmatrix}, \qquad {\tilde{e}}_2=\begin{pmatrix} 0 &{} i \\ i &{} 0 \end{pmatrix}, \qquad {\tilde{e}}_3=\begin{pmatrix} 0 &{} 1 \\ 1 &{} 0 \end{pmatrix}, \end{aligned}$$ 
\(\left\{ f_i\right\} _{i=1,\ldots , m}\) the generic basis for \(\mathfrak {g}=\mathfrak {k}\oplus \mathfrak {m}\), \(\mathfrak {k}=\langle f_1,\ldots ,f_k\rangle \), \(\mathfrak {m}=\langle f_{k+1},\ldots ,f_m\rangle \), where \(k={\text {dim}}\mathfrak {k}\), \(m={\text {dim}}\mathfrak {g}\),

\(e_1 {:}{=}\xi \cong \frac{\partial }{\partial t}\),

\(e_i {:}{=}\widehat{f}_{j}\), \(j={\text {dim}}\mathfrak {k}1+i\), the Killing vector fields on \(M^{\text {princ}}\) induced by the Gaction, for \(i=2,\ldots ,6\),

\(e^i\) the dual 1forms to \(e_i\).
Therefore, in what follows \(\left\{ e_i \right\} _{i=1,\ldots ,6}\) will be vectors on \(M^{\text {princ}}\) which provide a basis for \(T_pM\) at each point \(p=\gamma (t)\in M^{\text {princ}}\), where \(\gamma :\overset{\circ }{I} \rightarrow M\) is a normal geodesic through the point p.
Moreover, we recall some basic facts about Gactions which will be useful for our discussion:

Since \(g \cdot \gamma _p=\gamma _{g\cdot p } \) for the uniqueness of the normal geodesic \(\gamma \) starting from the point \(g\cdot p \), we have that \(\Phi _1^{\hat{X}}\circ \Phi _t^{\xi }(p)=\Phi _t^{\xi }\circ \Phi _1^{\hat{X}}(p)\), where \(\Phi _t^v\) denotes the flow of the vector field v evaluated at time t. This is equivalent to \([\xi , \hat{X}]=0\), for each \(X\in \mathfrak {g}\);

A kform \(\alpha \) on \(M^\text {princ}\) of the form
$$\begin{aligned} \alpha =\sum _{i_1<\ldots <i_k=1}^{6} a_{i_1\ldots i_k} \,e^{i_1\ldots i_k}, \end{aligned}$$with \(a_{i_1\ldots i_k} \in C^{\infty }(\overset{\circ }{I})\) for all \(i_1<\ldots <i_k\), is Ginvariant if and only if \(\alpha _p\) is Kinvariant for all \(p\in M^{\text {princ}}\). Here, \(e^{i_1\ldots i_k}\) is a shorthand for the wedge product \(e^{i_1}\wedge \ldots \wedge e^{i_k}\) of 1forms. Analogously, we shall indicate with \(\beta ^k\) the wedge product of \(\beta \) with itself for ktimes \(\beta \wedge \ldots \wedge \beta \);

If \(\alpha \) is a Ginvariant kform on M and \(v_1,\ldots , v_k\) are Ginvariant vector fields on M, then \(\alpha \left( v_1,\ldots ,v_n\right) _p\) is constant along the Gorbit through p, for each \(p\in M\).
We are now ready to state Theorem A.
Theorem A
Let M be a sixdimensional simply connected cohomogeneity one manifold under the almost effective action of a connected Lie group G and let K be the principal isotropy group. Then, the principal part \(M^{\text {princ}}\) admits a Ginvariant balanced nonKähler \(\text {SU}(3)\)structure \(\left( g,J,\Psi \right) \) if and only if M is compact and \(\left( \mathfrak {g}, \mathfrak {k} \right) =\left( \mathfrak {su}(2)\oplus 2\mathbb {R},\{0\} \right) \).
Proof of Theorem A
From all the above discussion and the previous lemmas, the only possible pairs allowing \(M^{\text {princ}}\) to support a balanced \(\text {SU}(3)\)structure are (a.1) with M compact, (c.3), and (b.1). We investigate these three cases separately.
For any of these cases, we shall consider the generic pair \((\omega ,\psi _+)\) of Ginvariant forms on \(M^{\text {princ}}\) of degree two and three, respectively, with \(C^{\infty }(\overset{\circ }{I})\)coefficients. In order for the pair \((\omega ,\psi _+)\) to define a Ginvariant balanced nonKähler \(\text {SU}(3)\)structure on \(M^{\text {princ}}\), we have to impose the following conditions:

(1)
the stability conditions:

\(\omega ^3 \ne 0\),

\(\lambda {:}{=}\lambda \left( \psi _+\right) <0\),


(2)
the compatibility conditions \(\psi _{\pm } \wedge \omega =0\),

(3)
the normalization conditions:

\(\psi _+ \wedge \psi _=\frac{2}{3} \omega ^3\),

\(\frac{1}{6} \omega ^3=\pm \sqrt{{\text {det}}(g)} \,e^{1\ldots 6}\) where the sign ± depends on the fixed orientation \(\pm e^{1\ldots 6}\),


(4)
\(\text {d} \psi _{\pm }=0\)

(5)
the balanced condition \(\text {d}\omega ^2=0\),

(6)
the nonKähler condition \(\text {d}\omega \ne 0\),

(7)
the positivedefiniteness of the induced symmetric bilinear form \(g {:}{=}\omega (\cdot , J \cdot )\) on \(M^{\text {princ}}\).
We start by case (b.1).
Case (b.1)
\(\left( \mathfrak {g},\mathfrak {k}\right) =\left( \mathfrak {su}(2)\oplus \mathfrak {su}(2), \mathbb {R}\right) \).
In the notation of Remark 3.2, let us suppose p, q nonzero and coprime with \((p,q) \ne (1,1)\), first. Consider the \(\mathcal {B}\)orthonormal basis of \(\mathfrak {g}\) given by
and take \(\mathfrak {k}=\langle f_1\rangle \). Notice that since rk\((\mathfrak {su}(2))=1\), this assumption is not restrictive. The decomposition of \(\mathfrak {g}\) into irreducible \(\text {Ad}\left( K\right) \)modules is given by
where \(\mathfrak {a}{:}{=}\langle f_2\rangle \) is \(\text {Ad}(K)\)fixed, \(\mathfrak {b}_1{:}{=}\langle f_3, f_5\rangle \) and \(\mathfrak {b}_2{:}{=}\langle f_4, f_6\rangle \), hence \(\mathfrak {m}= \mathfrak {a}\oplus \mathfrak {b}_1\oplus \mathfrak {b}_2\). Fix the orientation given by \(\Omega =e^{1\ldots 6}\) and consider the generic Ginvariant 3form \(\psi _+\) on \(M^{\text {princ}}\),
where \(p_j\in C^{\infty }(\overset{\circ }{I})\), \(j=1,\ldots , 4\). A simple calculation shows that the stability condition \(\lambda (\psi _+)<0\) never holds, since \(\lambda (\psi _+)=\left( p_1 p_4  p_2 p_3 \right) ^2\ge 0\).
Now let \((p,q)=(1,0)\) and consider the \(\mathcal {B}\)orthogonal basis of \(\mathfrak {g}\) given by 4.1 when \((p,q)=(1,0)\) and assume \(\mathfrak {k}=\langle f_1\rangle \) as before. Then, the decomposition of \(\mathfrak {g}\) into irreducible \(\text {Ad}(K)\)modules is given by
where \(\mathfrak {b}_1 {:}{=}\langle f_3, f_5\rangle \), \(\mathfrak {a}_1{:}{=}\langle f_2\rangle \), \(\mathfrak {a}_2{:}{=}\langle f_4\rangle \) and \(\mathfrak {a}_3{:}{=}\langle f_6\rangle \). Observe that the \(\mathfrak {a}_i\)’s are equivalent. Consider the generic Ginvariant 3form \(\psi _+\) on \(M^{\text {princ}}\), which is of the form
where \(p_j\in C^{\infty }(\overset{\circ }{I})\), \(j=1,\ldots ,8\). It is straightforward to show that \(\lambda (\psi _+)=\left( p_1 p_8 + p_2 p_7  p_3 p_6 +p_4 p_5\right) ^2\ge 0\).
Remark 4.1
By the previous discussion we have that when \(\left( \mathfrak {g},\mathfrak {k}\right) =\left( \mathfrak {su}(2)\oplus \mathfrak {su}(2), \mathbb {R}\right) \) with \(\mathfrak {k}\) not diagonally embedded in \(\mathfrak {g}\), M admits no Ginvariant \(\text {SL}(3,\mathbb {C})\)structures, i.e., Ginvariant stable 3forms inducing an almost complex structure on M.
Finally, let us consider the case when \(\mathfrak {k}\) is diagonally embedded in \(\mathfrak {g}\). Without loss of generality, we can assume \((p,q)=(1,1)\). We consider the \(\mathcal {B}\)orthonormal basis of \(\mathfrak {g}\) given by 4.1 when \((p,q)=(1,1)\). The decomposition of \(\mathfrak {g}\) into irreducible \(\text {Ad}\left( K\right) \)modules is given by
where \(\mathfrak {k}=\langle f_1\rangle \), \(\mathfrak {a}{:}{=}\langle f_2\rangle \) is \(\text {Ad}(K)\)fixed, \(\mathfrak {b}_1{:}{=}\langle f_3, f_5\rangle \) and \(\mathfrak {b}_2{:}{=}\langle f_4, f_6\rangle \). Then, \(\mathfrak {m}= \mathfrak {a}\oplus \mathfrak {b}_1\oplus \mathfrak {b}_2\). Unlike the case \(p\ne q\) both nonzero, here the equivalence of the \(\mathfrak {b}_i\)modules implies that the metric g on \(M^{\text {princ}}\) is not necessarily diagonal but of the form:
for some \(f, h_1, h_2 \in C^{\infty }(\overset{\circ }{I})\), where \(\mathcal {Q}\) denotes a symmetric quadratic form on the isotypic component \(\mathfrak {b}_1\oplus \mathfrak {b}_2\). In particular, the metric coefficients \(g_{ij}{:}{=}g(e_i,e_j)\) must satisfy
where \(e_i\), \(i=1,\ldots ,6\), are the vector fields defined in the usual way. Fix the orientation given by \(\Omega {:}{=}e^{1\ldots 6}\), and consider a pair of Ginvariant forms \(\left( \omega ,\psi _+\right) \) of degree two and three, given respectively by
where \(h_i, p_j\in C^{\infty }(\overset{\circ }{I})\), \(i=1,\ldots ,5\), \(j=1,\ldots ,8\). Moreover, the structure equations are given by
In order to find a Ginvariant balanced nonKähler \(\text {SU}(3)\)structure on \(M^{\text {princ}}\), we have to impose the conditions 1 to 7 listed at the beginning of this Section, together with 4.2. We shall show that this system of equations is incompatible. This implies there are no Ginvariant balanced nonKähler \(\text {SU}(3)\)structures on the corresponding M. In order to see this, we write all conditions in terms of the coefficients \(h_i, p_j\) of \(\left( \omega ,\psi _+\right) \), for \(i=1,\ldots ,5\), \(j=1,\ldots ,8\). One has that \(\text {d}\omega ^2=0\) if and only if
and, in particular, \(\text {d}\omega =0\) if and only if
Similarly, \(\text {d}\psi _+=0\) if and only if
Let us suppose that \(\psi _+\) is stable with \(\lambda <0\), and consider the induced almost complex structure J on \(M^{\text {princ}}\). Recall that by Ginvariance, \(\psi _=J \psi _+\) needs to be of the same general form of \(\psi _+\), namely
where the \(q_i\)’s are functions of \(\{p_j\}_{j=1,\ldots ,8}\) for \(i=1,\ldots ,8\). Therefore, \(\text {d}\psi _=0\) if and only if
Moreover, 4.2 is equivalent to
where we have already assumed \(p_5=p_6\) from 4.3. Since \(p_6'=0\) and all the conditions for the Ginvariant balanced nonKähler \(\text {SU}(3)\)structure involve only homogeneous polynomials, we can assume either \(p_6=0\) or \(p_6=1\), up to scalings. Some possibilities can be excluded using the following lemmas.
Lemma 4.2
Assume \(p_6=0\). If \(p_1=0\), or \(p_2=0\), or \(p_7=0\), then conditions (1)(7) are incompatible.
Proof
Let us assume \(p_1=0\). Then, \(\lambda (\psi _+)=2(p_3 p_8p_4 p_7)^2\le 0\) and
where the signs of \(q_3, q_7\) depend on that of \((p_3 p_8p_4 p_7)\). Then, \(\text {d}\psi _{\pm }=0\) implies \(p_3=p_4=0\), from which \(\lambda =0\) follows.
Assume instead \(p_2=0\). Then, we have \(\lambda =2(p_3 p_8p_4 p_7)^2 \le 0\), as in the previous case. Moreover, one can easily compute
by which we can draw the same conclusion.
Finally, let us assume \(p_7=0\). Then, 4.3 implies \(p_4=0\). In this case, \(\lambda (\psi _+)=2 p_8^2 (p_1 p_2p_3^2)\) can be strictly negative, and one can compute that
Therefore, assuming \(p_8\ne 0\) to ensure \(\lambda (\psi _+)\ne 0\), the requirement \(\text {d}\psi _{\pm }=0\) imposes
which implies \(\lambda \ge 0\). \(\square \)
Lemma 4.3
If \(h_5 \ne 0\), \(p_6=1\), \(p_8=0\), then conditions (1)(7) and 4.5 are incompatible.
Proof
From 4.5 and the closure of \(\psi _+\), one has
from which it follows that \(\lambda =4p_2^2(p_7^21)\), and
Thus \(q_5=q_6=0\), from 4.4, which would force \(\lambda \) to vanish. \(\square \)
We can divide the discussion into the following cases:

(1)
\(h_5\ne 0, p_6=0\),

(2)
\(h_5 \ne 0, p_6=1\),

(3)
\(h_5=0, p_6=0\),

(4)
\(h_5=0, p_6=1\).
We study each case separately.
Case (1). By 4.5, Lemma 4.2 and \(\text {d}\psi _{+}=0\), it follows that \(p_3=p_8=h_4=0.\) Then, \(\text {d}\psi _{}=0\) implies \(p_1=p_2\). Now, \(\lambda =2 p_7^2 (2 p_2^2p_4^2)\), and the compatibility condition \(\psi _+\wedge \omega =0\) holds if and only if \(p_2 (h_2 + h_3)=2 h_5 p_4\). Then, if \(h_2\ne h_3\), we can write \(p_2=\frac{2 p_4 h_5}{(h_2+h_3)}\). Therefore, 4.5 reduces to
all of whose solutions imply \(\lambda \ge 0 \). When \(h_2=h_3 \), the condition \(\psi _+\wedge \omega =0\) implies \(p_4=0\) from which \(\lambda \ge 0\).
Case (2). By Lemma 4.3, we can assume \(p_8\ne 0\). Then, by 4.5, we have that
Moreover, since in this case \(\lambda =2(p_8^22)(p_2 p_7p_3)^2\), 4.5 implies \(h_4=0\) as well. Then, 4.5 implies
from which it follows that \(h_2=h_3=2 \frac{h_5}{p_8}\), since \(\lambda \) must not vanish. Then, \(\psi _+ \wedge \omega =0\) if and only if \(p_8^22=0\), which would imply \(\lambda =0\).
Case (3). By 4.5 and Lemma 4.2, we have \(h_4=0\), which implies \({\text {det}}(g)=h_1^2 h_2^2 h_3^2\). Then, from 4.5, we also have that
If \(p_3,p_8\ne 0\), then 4.6 implies that \(p_8^2+2p_7^2=0\), which contradicts our hypothesis. If \(p_8=0\), the closure of \(\psi _+\) implies \(p_3=0\). Then, we only need to discuss the remaining case \(p_3=0\). Supposing this is the case, then 4.6, together with Lemma 4.2, implies \(p_4=0\). Under these hypotheses, one can easily compute that \(\lambda =2 p_1 p_2 (2p_7^2+p_8^2)\),
so that \(\lambda <0\) forces \(q_5 \ne q_6\), a contradiction.
Case (4). Here, the compatibility condition \(\psi _+\wedge \omega =0\), which holds if and only if
together with 4.5, implies that one of the following must hold:

(4.a)
\( h_4=0\),

(4.b)
\(p_4=0\),

(4.c)
\(2 p_7^2+p_8^2=2\).
Let us start with case (4.a). By 4.7, we have that \(h_2=h_3\). In particular, since \({\text {det}}(g)=h_1^2 h_2^2 h_3^2\), we must have \(h_3\ne 0 \). Then, a simple calculation show that d\(\omega ^2=0\) if and only if d\(\omega =0\). As for Case (4.b), by 4.7 and 4.5, we have that
from which it follows that \(\lambda =2 (2 p_7^2 + p_8^2 2)(2 p_2 p_3 p_7 + p_2^2 + p_3^2)\). Moreover, one can show that \(q_5=q_6\) implies \(p_2=p_3 p_7\). Now, 4.5 implies \(h_4=0\), which was already ruled out in the previous case. In case (4.c), again by 4.7 and 4.5, we have that
which implies \(\lambda =0\). This concludes case (b.1).
Case (c.3)
\(\mathfrak {g}=\mathfrak {su}(3)\), \(\mathfrak {k}=\mathfrak {su}(2)\).
Consider the \(\mathcal {B}\)orthogonal basis of \(\mathfrak {g}\) given by
Then, \(\mathfrak {k}=\langle f_1, f_2, f_3\rangle \). Let \(\mathfrak {a}{:}{=}\langle f_8\rangle \) and \(\mathfrak {n}{:}{=}\langle f_4, f_5, f_6, f_7\rangle \); hence, \(\mathfrak {m}=\) \(\mathfrak {a}\oplus \mathfrak {n}\). Since the \(\text {Ad}(K)\)invariant irreducible modules in the decomposition of \(\mathfrak {g}\) are pairwise inequivalent, the metric g on \(M^{\text {princ}}\) is diagonal and, in particular, it is of the form
for some positive \(h, f \in C^{\infty }(\overset{\circ }{I})\). Moreover, with respect to the frame \(\{e_i\}_{i=1,\ldots ,6}\) of \(M^{\text {princ}}\), the structure equations are given by
Fix the volume form \(\Omega =e^{1\ldots 6}\). One can easily show that a pair of generic Ginvariant forms \(\left( \omega ,\psi _+\right) \) on \(M^\text {princ}\) of degree two and three is given, respectively, by
where \(h_i, p_j\in C^{\infty }(\overset{\circ }{I})\), \(i=1,\ldots ,4\), \(j=1,\ldots ,6\). As we did for case (b.1), we are going to show that the system of equations resulting from imposing conditions (1)(7) is incompatible.
A simple computation shows that d\(\psi _+=0\) if and only if
From the Ginvariance,
where the \(q_i\)’s are functions of \(\{p_j\}_{j=1,\ldots ,6}\) for \(i=1,\ldots ,6\). Therefore, d\(\psi _=0\) if and only if
In particular, from \(p_4=0\), it follows
with \(\lambda =4(p_1^2 \, (p_3^2+p_6^2)+(p_2\,p_6p_3\,p_5)^2)\). We suppose that \(\psi _+\) is stable with \(\lambda <0\). Then, \(q_4=0\) if and only if \(p_1=0\). Since \(p_1\) has to be equal to zero, it can be shown that the compatibility condition \(\psi _+ \wedge \omega =0\) is equivalent to the following system of equations:
Moreover, the positivedefiniteness of g implies \(h_1>0\). Then, the normalization condition \(\psi _+ \wedge \psi _=\frac{2}{3}\omega ^3\) is equivalent to
The balanced condition d\(\omega ^2=0\) is satisfied if and only if
Finally, the Kähler condition d\(\omega =0\) holds if and only if
Multiplying 4.9 by \(h_4\) and using 4.8, we obtain \(h_4h_1(h_2^2+h_3^2+h_4^2)=0\) and, since \(h_1>0\) and \(h_2=h_3=h_4=0\) would imply \(\omega ^3=0\), we necessarily have \(h_4=0\). Then, 4.8 implies
from which it follows \(h_3=0\) since \(p_2=p_3=0\) would imply \(\lambda =0\). Then, 4.10 reads \(h_2(\sqrt{3}h_1+h_2')=0\) and, since \(h_2\ne 0 \), in order to have \(\omega ^3\ne 0\), we have \(\sqrt{3}h_1+h_2'=0\), namely \(d\omega =0\). Therefore, any Ginvariant balanced \(\text {SU}(3)\)structure on the corresponding M is necessarily Kähler. This concludes case (c.3).
Case (a.1)
\(\mathfrak {g}=\mathfrak {su}(2)\oplus 2\mathbb {R}\), \(\mathfrak {k}=\{0\}\).
Since \(\mathfrak {k}=\{0\}\), we can write \(T_pM\cong \langle e_1_p\rangle \oplus \hat{\mathfrak {g}}\big _p\), for each \(p\in M^{\text {princ}}\). Moreover, every kform \(\alpha \) on \(M^{\text {princ}}\) of the form
where \(\alpha _{i_1\ldots i_k} \in C^{\infty }(\overset{\circ }{I})\), is Ginvariant. Let
be a pair of generic Ginvariant forms on \(M^{\text {princ}}\) of degree two and three, respectively, with coefficients \(h_{ij}, p_{ijk} \in C^{\infty }(\overset{\circ }{I})\). If we choose a \(\mathcal {B}\)orthogonal basis of \(\mathfrak {su}(2)\) with vectors of constant norm, say
and extend it to a basis \(\{f_i\}_{i=1,\ldots ,5}\) of \(\mathfrak {g}\), the structure equations with respect to the frame \(\{e_i\}_{i=1,\ldots ,6}\) of \(M^{\text {princ}}\) are given by
Fix the volume form \(\Omega {:}{=}e^{1\ldots 6}\). We consider the forms given in 4.12 and set
for each \(t\in (1,1)\), and all other coefficients equal to zero. Then, by performing the change of variable
one can easily check that the resulting pair \(\left( \omega ,\psi _+\right) \) defines a Ginvariant balanced nonKähler \(\text {SU}(3)\)structure on the corresponding \(M^{\text {princ}}\). With respect to the t parameter, the metric on \(M^{\text {princ}}\) is represented by the matrix
However, using the results from [32], we can check that this example cannot be extended to the singular orbits to give a smooth metric on the whole manifold. This concludes the proof of Theorem A.
Proof of Theorem B
We will finally prove our main theorem.
Theorem B
Let M be a sixdimensional simply connected cohomogeneity one manifold under the almost effective action of a connected Lie group G. Then, M admits no Ginvariant balanced nonKähler \(\text {SU}(3)\)structures.
By Theorem A, we only need to discuss if there exist balanced nonKähler \(\text {SU}(3)\)structures of cohomogeneity one arising as the compactification of the principal part determined by the pair \(\left( \mathfrak {g}, \mathfrak {k} \right) =\left( \mathfrak {su}(2)\oplus 2\mathbb {R},\{0\} \right) \).
By [21], a sixdimensional compact simply connected cohomogeneity one manifold M whose corresponding principal part is given by the pair \(\left( \mathfrak {g},\mathfrak {k}\right) =\left( \mathfrak {su}(2)\oplus 2\mathbb {R},\{0\}\right) \) at the Lie algebra level is Gequivariantly diffeomorphic to the product of two threedimensional spheres, i.e., \(M\cong S^3\times S^3\). If we denote by \(H_i\) for \(i=1,2\), the singular isotropy groups for the Gaction on M and by \(\mathfrak {h}_i=\text {Lie}(H_i)\), for \(i=1,2\), their Lie algebras, we have that both \(\mathfrak {h}_1\) and \(\mathfrak {h}_2\) are isomorphic to \(\mathbb {R}\) so that both the singular orbits of M are fourdimensional compact submanifolds of M. Let \(b_i\) be the ith Betti number of M, then, up to Gequivariant diffeomorphisms, we may assume \(b_4=0\). By Michelsohn’s obstruction [23, Corollary 1.7], if M admitted any 4dimensional compact complex submanifold S, then M would not admit a balanced metric. Therefore, we can make a few considerations by focusing on one tubular neighborhood of singular orbit \(G\supset H\supset K\) at a time. In particular, we divide the discussion depending on the immersion of \(\mathfrak {h}\subset \mathfrak {g}\). Let S be the singular orbit given by the group diagram \(G\supset H\supset K\). We notice that if S is Jinvariant, a complex structure on M would give rise to a complex structure on S, so we can discard all these cases by Michelsohn’s obstruction. In particular, we have that \(T_qM=T_qS \oplus V\) where \(V=T_qS^{\perp }\) is the slice at \(q\in S\); since S is fourdimensional, V is always a 2plane. We recall that the Haction on \(T_qS\) is given by the adjoint representation while the Haction on V is given via slice representation, and since V is bidimensional, this action is just a rotation on V of a certain speed, say a. Let us start by considering the case when \(\mathfrak {h}\) is contained in the center of \(\mathfrak {g}\), \(\xi (\mathfrak {g})\). In this case, the Haction on \(T_qS\) is trivial. Therefore, \(T_qS \oplus V\) are inequivalent modules of the Haction on \(T_qM\) and, since J commutes with the Haction, J preserves \(T_qS\) for each \(q\in S\), i.e., S is an almost complex manifold and we may apply Michelsohn’s obstruction to discard this case. Therefore, we may suppose that \(\mathfrak {h}\) has a nontrivial component in the \(\mathfrak {su}(2)\)factor of \(\mathfrak {g}\). In particular, since \(\text {rank}(\mathfrak {su}(2))=1\) and the adjoint action ignores components in the center of \(\mathfrak {g}\), we may assume, without loss of generality and using the notation from Sect. 4.3, \(\mathfrak {h}=\langle f_1\rangle \). Moreover, if we denote by \(\mathfrak {m}\) the tangent space to S via the usual identification, the decomposition of \(\mathfrak {m}\) in irreducible Hmodules is given by
where H acts on \(l_0\) trivially and on \(l_1\) via rotation of speed d. Therefore, when the integer a is different from d, the modules \(l_0\), \(l_1\) and V are inequivalent for the Haction and again, since J commutes with the Haction, it cannot exchange two different modules. In particular, \(J(T_qS)\subseteq T_qS\) and we may apply Michelsohn’s obstruction as before. For the remaining case \(a=d\), we have that the two modules \(l_1\) and V are equivalent; hence, \(J(l_1\oplus V)\subseteq l_1\oplus V\) but not necessarily \(J(l_1)\subseteq l_1\). In particular, when this case occurs, the orbit S is not Jinvariant, and we do not have obstructions to the existence of balanced metrics. Therefore, from now on, we assume this is the case.
Let \(\partial / \partial x\) be a vector field such that \(\left( \xi _q, \partial / \partial x_q \right) \) is an orthonormal basis for the slice V and \(T_q^*M = \langle e^1_q, {\text {d}}x_q, e^3_q, e^4_q, e^5_q, e^6_q \rangle \). Let \(\varphi : \mathfrak {h} \rightarrow \text {End}(T_qM)\) be the \(\mathfrak {h}\)action on \(T_qM\). Then in order to have \(l_1\) and V \(\mathfrak {h}\)equivalent, \(\varphi (f_1)^*\) acts on 1forms given in the previous basis as
Fix the volume form \(\Omega = e^{1\ldots 6}\) and consider the three form
where \(p_j \in C^\infty ((1, 1))\) for any \(j = 1, \dots , 20\).
The condition \(d \psi _+=0\) is equivalent to the following ODE system:
From now on, we will assume \(p_{16}=p_{19}=p_{20}=0\).
Let the slice be \(V \cong \mathbb {R}^2\) so that the singular point \(q \in \mathcal {O}_1\) is identified with \(0 \in \mathbb {R}^2\), and let \(r:V \rightarrow \mathbb {R}\) be the radial distance, such that for \(v=(v_1, v_2) \in V\), \(r(v) =  v = \sqrt{v_1^2 + v_2^2}\). Then \(r \not \in C^\infty (V)\), and neither are the odd powers of r. Via the exponential map, we can identify \(t+1\) with the radial distance r.
Let \(\alpha \) be a Ginvariant 1form on M. Then,
for \(t \in (1,1)\) and some smooth functions \(\alpha _i\), \(i=1, ..., 6\). This expression has to extend smoothly to \(t=1\). In particular, the Taylor expansion of \(\alpha _k(t)\) around \(t=1\) for \(k \ge 2 \) only has even powers of \(t+1\):
Now for \(2 \le i< j<k \le 6\) fixed, the \(e^{ijk}\)coefficients extend smoothly to \(t=1\). Hence,
as well as for the Taylor expansion of \(p_{13}(t), p_{14}(t)\) and \(p_{15}(t)\) around \(t=1\). Therefore, \(\lim _{t \rightarrow 1} p_{12}'(t)= \lim _{t \rightarrow 1} p_{13}'(t)= \lim _{t \rightarrow 1} p'_{14}(t)= \lim _{t \rightarrow 1} p'_{15}(t)=0\). From 5.1, we obtain that \(\lim _{t \rightarrow 1} p_6(t)= \lim _{t \rightarrow 1} p_7(t)= \lim _{t \rightarrow 1} p_8(t)= \lim _{t \rightarrow 1} p_9(t)=0\).
The three form \(\psi _+\) at \(t=0\) has to be Hinvariant, and hence can be written as
for some \(c_3, c_4, c_6, c_7, c_8, c_9, c_{17}, c_{18} \in \mathbb {R}\). But \(c_i = \lim _{t \rightarrow 1} p_i(t)=0\) for \(i=6, 7, 8, 9\). Therefore, one can easily compute that
This concludes case (a.1).
We note that it is possible to reach a contradiction by just studying the behavior around one of the singular orbits. However, if we do not use the information coming from Michelsohn’s obstruction, the computations get significantly more complicated. The main point is that from \(d \psi _=0\) and using the stability condition \(\lambda <0\), we get \(p_{10}=0\). If we assume this too, the three form \(\psi _+\) at \(t=1\) can be written as
for some \(c_i \in \mathbb {R}\), \(i=1,\ldots ,18, i\ne 10,16\). Then, once again we find that \(\lambda _{t=1}= (c_{18} c_3  c_{17} c_4)^2 \ge 0\) which finishes the case.
Remark 5.1
We also note that in case (a.1) and when \(\mathfrak {h}= \mathbb {R}\), we can remove the hypothesis of simply connectedness from the noncompact case and still get a non existence result. Let M be a sixdimensional noncompact cohomogeneity one manifold under the almost effective action of a connected Lie group G and let K, H be the principal and singular isotropy groups, respectively, with \(\left( \mathfrak {g}, \mathfrak {h}, \mathfrak {k} \right) =\left( \mathfrak {su}(2)\oplus 2\mathbb {R},\mathbb {R},\{0\} \right) \). Then, M admits no Ginvariant balanced nonKähler \(\text {SU}(3)\)structures.
From Theorem B, we get the following Corollary.
Corollary 5.2
There is no nonKähler balanced \( \text {SU}(3)\)structure on \(S^3 \times S^3\) which is invariant under a cohomogeneity one action.
Remark 5.3
In the nonsimplyconnected case, by Theorem A, we can discard cases (b.1) and (c.3), as these do not admit local solutions to conditions (1)–(7). Moreover, as observed in [26, Section 3], one can also rule out cases (b.3) and (c.2), as the Gaction would not be almost effective, as well as case (c.1) since it would give rise to a threedimensional Jinvariant subspace, a contradiction.
Data sharing is not applicable to this article as no datasets were generated or analyzed during the current study.
References
 1.
Alexandrino, M.M., Bettiol, R.G.: Lie Groups and Geometric Aspects of Isometric Actions. Springer, Berlin (2015)
 2.
Alexandrov, B., Ivanov, S.: Vanishing theorems on Hermitian manifolds. Differ. Geom. Appl. 14, 251–265 (2001)
 3.
Bedulli, L., Vezzoni, L.: The Ricci tensor of SU(3)manifolds. J. Geom. Phys. 57, 1125–1146 (2007)
 4.
Bedulli, L., Vezzoni, L.: A parabolic flow of balanced metrics. J. Reine Angew. Math. 723, 79–99 (2017)
 5.
Bérard Bergery, L.: Sur des nouvelles varietes Riemanniennes d’Einstein, Publ. de Inst. E. Cartan 6 (1982), 1–60
 6.
Fei, T.: A construction of nonKähler Calabi–Yau manifolds and new solutions to the Strominger system. Adv. Math. 302, 529–550 (2016)
 7.
Fei, T., Huang, Z., Picard, S.: A construction of infinitely many solutions to the Strominger system. J. Differ. Geom. 117, 23–39 (2021)
 8.
Fei, T., Yau, S.T.: Invariant solutions to the Strominger system on complex Lie groups and their quotients. Commun. Math. Phys. 338, 1183–1195 (2015)
 9.
Fernández, M., Tomassini, A., Ugarte, L., Villacampa, R.: Balanced Hermitian metrics from SU(2)structures. J. Math. Phys. 50, 033507 (2009)
 10.
Fino, A., Grantcharov, G., Vezzoni, L.: AsthenoKähler and balanced structures on fibrations. Int. Math. Res. Not. 22, 7093–7117 (2019)
 11.
Fino, A., Vezzoni, L.: Special Hermitian metrics on compact solvmanifolds. J. Geom. Phys. 91, 40–53 (2015)
 12.
Fino, A., Vezzoni, L.: On the existence of balanced and SKT metrics on nilmanifolds. Proc. Am. Math. Soc. 144, 2455–2459 (2016)
 13.
Fu, J., Li, J., Yau, S.T.: Balanced metrics on nonKähler Calabi–Yau threefolds. J. Differ. Geom. 90, 81–129 (2012)
 14.
Fu, J.X., Tseng, L.S., Yau, S.T.: Local heterotic torsional models. Commun. Math. Phys. 289, 1151–1169 (2009)
 15.
Fu, J.X., Yau, S.T.: The theory of superstring with flux on nonKähler manifolds and the complex Monge–Ampére equation. J. Differ. Geom. 78, 369–428 (2008)
 16.
GarciaFernandez, M.: Tdual solutions of the Hull–Strominger system on nonKähler threefolds, arXiv:1810.04740 (2018)
 17.
Grantcharov, D., Grantcharov, G., Poon, Y.S.: Calabi–Yau connections with torsion on toric bundles. J. Differ. Geom. 78, 13–32 (2008)
 18.
Grantcharov, G.: Geometry of compact complex homogeneous spaces with vanishing first Chern class. Adv. Math. 226, 3136–3159 (2011)
 19.
Hitchin, N.: The geometry of threeforms in six dimensions. J. Differ. Geom. 55, 547–576 (2000)
 20.
Hoelscher, C.A.: Classification of cohomogeneity one manifolds in low dimensions. Pac. J. Math. 246, 129–185 (2010)
 21.
Hoelscher, C.A.: Diffeomorphism type of sixdimensional cohomogeneity one manifolds. Ann. Global Anal. Geom. 38, 1–9 (2010)
 22.
Hull, C.: Superstring compactifications with torsion and spacetime supersymmetry. In: Turin 1985 Proceedings Superunification and Extra Dimensions, 347–375 (1986)
 23.
Michelsohn, M.L.: On the existence of special metrics in complex geometry. Acta Math. 149, 261–295 (1982)
 24.
Otal, A., Ugarte, L., Villacampa, R.: Invariant solutions to the Strominger system and the heterotic equations of motion. Nucl. Phys. B 920, 442–474 (2017)
 25.
Phong, D.H., Picard, S., Zhang, X.: A flow of conformally balanced metrics with Kähler fixed points. Math. Ann. 374, 2005–2040 (2019)
 26.
Podestà, F., Spiro, A.: 6dimensional nearly Kähler manifolds of cohomogeneity one. J. Geom. Phys. 60, 156–164 (2010)
 27.
Pujia, M.: The Hull–Strominger system and the Anomaly flow on a class of solvmanifolds, arXiv:2103.09854
 28.
Pujia, M., Ugarte, L.: The Anomaly flow on nilmanifolds, arXiv:2004.06744
 29.
Reichel, W.: Über die Trilinearen Alternierenden Formen in 6 und 7 Veränderlichen, Ph.D. thesis, Greifswald (1907)
 30.
Strominger, A.: Superstrings with torsion. Nucl. Phys. B 274, 253–284 (1986)
 31.
Ugarte, L., Villacampa, R.: Balanced Hermitian geometry on 6dimensional nilmanifolds. Forum Math. 27, 1025–1070 (2015)
 32.
Verdiani, L., Ziller, W.: Smoothness Conditions in Cohomogeneity one manifolds, Transform. Groups, arXiv:1804.04680 (to appear)
 33.
Wang, H.C.: Complex parallisable manifolds. Proc. Am. Math. Soc. 5, 771–776 (1954)
 34.
Ziller, W.: On the geometry of cohomogeneity one manifolds with positive curvature. In: Riemannian Topology and Geometric Structures on Manifolds, Progress in Mathematics 271, Birkhäuser (2009)
Acknowledgements
The first named author wants to thank Andrew Dancer and Jason Lotay for introducing the problem and their support and help with it. The second named author would like to thank Lucio Bedulli for introducing the problem and for useful conversations and comments and Anna Fino for her constant support, encouragement and patient guidance. The second named author would like to thank also Alberto Raffero and Fabio Podestà for helpful discussions and remarks.
Author information
Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
The first author was supported by the EPSRC. The second author was supported by GNSAGA of INdAM.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Alonso, I., Salvatore, F. On the existence of balanced metrics on sixmanifolds of cohomogeneity one. Ann Glob Anal Geom (2021). https://doi.org/10.1007/s1045502109807z
Received:
Accepted:
Published:
DOI: https://doi.org/10.1007/s1045502109807z
Keywords
 Cohomogeneity one actions
 Balanced metrics
 \(\text {SU}(3)\)structures
Mathematics Subject Classification
 53C25