Compact complex non-Kähler manifolds associated with totally real reciprocal units (Q2672688): Difference between revisions

In [\textit{M. Inoue}, Invent. Math. 24, 269--310 (1974; Zbl 0283.32019)], a new class of compact complex non-Kähler surfaces, now known as the Inoue surfaces, was introduced. The surfaces in this class come in three essential variants: \(S_M, S_N^{(+)}, S_N^{(-)}\) (where the last is a quotient of the second by an involution). For a long time, there was no generalization of their construction to higher dimensions. This changed when in [the first author and \textit{M. Toma}, Ann. Inst. Fourier 55, No. 1, 161--171 (2005; Zbl 1071.32017)] a construction in every dimension, generalizing that of \(S_M\) was given. These manifolds, now known as Oeljeklaus-Toma (OT) manifolds, are, like the Inoue surfaces, solvmanifolds: Quotients of a solvable Lie-group with left-invariant complex structure by a co-compact lattice (this was made explicit in [\textit{H. Kasuya}, Bull. Lond. Math. Soc. 45, No. 1, 15--26 (2013; Zbl 1262.53061)]). In the present paper, the authors give a solvmanifold construction in every dimension which generalizes that of \(S_N^{(+)}\). They also sketch how to generalize that of \(S_N^{(-)}\). Surveyed briefly, their construction is as follows: Identify \(\mathbb{C}\times \mathbb{H}=\{(a,b)\in\mathbb{C}^2\mid \operatorname{Im}(b)>0\}\) with a solvable matrix group \(G\) via \[ (z,w) \leftrightarrow \begin{pmatrix}1& \operatorname{Im}(z)& \operatorname{Re}(z)\\ 0& \operatorname{Im}(w)& \operatorname{Re}(w)\\ 0& 0& 1\end{pmatrix} \] For any totally real number field \(K\) of degree \(2d\) and a primitive totally positive unit \(u\in \mathcal{O}_K^{*,+}\) which is reciprocal (i.e., \(u\) and \(u^{-1}\) are conjugate), the authors construct an irreducible lattice on \(G^d\) and define their manifolds \(X\) to be the quotients. (Number fields admitting such units exist for every \(d\), as is shown in the paper.) The authors calculate the identity component of holomorphic automorphism group as \(\operatorname{Aut}^0(X)=(\mathbb{C}^\times)^d\). This acts freely, inducing a transversally holomorphic foliation, which is shown to have an Anosov-property for \(d=2\). Unlike in the case of Inoue surfaces, the whole automorphism group has infinitely many components if \(d\geq 2\). The authors also show that the Kodaira dimension is \(-\infty\), that the algebraic dimension is zero and that the manifolds have the de Rham cohomology ring of \((S^1\times S^3)^d\). Like the Inoue surfaces, the manifolds are never Kähler, which the authors show in two ways: As a consequence of their study of the automorphism group and by proving that there is no nonzero closed holomorphic one-form vanishes (contrasting the positive first Betti number). (We note that as a third way one could also invoke their calculation of the de Rham cohomology ring, since the Hard Lefschetz property fails.)