Manifolds with holonomy \(\mathrm{U}^\ast (2m)\) (Q743619)

Kähler geometry can be seen as the geometry determined by a torsion-free affine connection with holonomy group a subgroup of \(\mathrm{U}(n)\). Here, the author rather directs his attention to the geometry determined by a torsion-free affine connection with holonomy group inside \(\mathrm{U}^*(2m)\), a group also appearing in Berger's list of irreducible holonomy groups [\textit{M. Berger}, Bull. Soc. Math. Fr. 83, 279--330 (1955; Zbl 0068.36002)]. This quite mysterious holonomy type is of particular interest for string theorists, via the so-called (generalised) c-map construction [\textit{A. Haydys}, J. Geom. Phys. 58, No. 3, 293--306 (2008; Zbl 1143.53043)]. The group \(\mathrm{U}^*(2m)\) can be defined as the group generated by scalar rotations, and matrices of \(\mathrm{SL}(2m,\mathbb{C})\) commuting with \((z_1,\ldots,z_m,w_1,\dots, w_m)\mapsto (-\bar{w_1},\dots, \bar{w_m},\bar{z_1},\ldots,\bar{z_m})\); this way, \(\mathrm{U}^*(2m)\subset\mathrm{GL}(2m,\mathbb{C})\), bringing us to the framework of \textit{complex geometry}. As mentioned by the author, \(\mathrm{U}^*(2m)\) also has a quaternionic interpretation hence the alternative notation \(\mathrm{SL}(m\mathbb{H})\cdot\mathrm{U}(1)\), and Joyce's terminology ``quaternionic complex manifolds'' [\textit{D. Joyce}, Q. J. Math., Oxf. II. Ser. 49, No. 194, 129--162 (1998; Zbl 0924.14002)]. Moreover, back to differential geometry, connections with \(\mathrm{U}^*(2m)\)-holonomy are unique (Section 2). The third and fourth parts of the paper are then devoted to a study of the \(m=1\) (complex surfaces) and \(m\geq 2\) cases of compact manifolds with holonomy \(\mathrm{U}^*(2m)\). Noticing that \(\mathrm{U}^*(2)=\mathrm{U}(2)\), a convenient convention for surfaces is made that consists in assuming that the arising Kähler metric has zero scalar curvature; examples are produced, as quotients of the model product (round sphere)\(\times\)(hyperbolic disc). In both the \(m=1\) and \(m\geq 2\) cases, the main results of these parts (Propositions 1 and 2) are the vanishing of all plurigenera, provided that the Ricci tensor (attached to the \(\mathrm{U}^*(2m)\) structure) is not identically zero; in the \(m\geq 2\) case. This requires the use of local quaternionic triples of complex structures. Further classes of examples are constructed in Parts 5 and 6; this is where the generalised c-map picture enters. Considering a quaternion Kähler (or, holonomy \(\mathrm{Sp}(m)\cdot\mathrm{Sp}(1)\)) manifold \(M\), with an \(\mathbb{S}^1\)-action/a Killing vector field, a (uniquely determined) torsion-free affine connection with holonomy \(\mathrm{U}^*(2m)\) is produced on the complement \(M_0\) of the moment-section zero set. The moment-section is that determined by the above action/Killing field. Next, observing that a hyper-Kähler manifold with an \(\mathbb{S}^1\)-action rotating its holomorphic symplectic form gives rise to a quaternion Kähler manifold with an \(\mathbb{S}^1\)-action [\textit{A. Haydys}, J. Geom. Phys. 58, No. 3, 293--306 (2008; Zbl 1143.53043)], the previous construction is applied to such (noncompact) hyper-Kähler manifolds. As an example, starting with \textit{E. Calabi}'s metric on \(T\mathbb{C}P^m\) [Ann. Sci. Éc. Norm. Supér. (4) 12, 269--294 (1978; Zbl 0431.53056)], this correspondence provides \(\mathrm{SO}^*(2m+2)/\mathrm{SO}^*(2m)\times\mathrm{SO}(2)\) as a \(\mathrm{U}^*(2m)\)-manifold. The final part (Section 7) is devoted to a study of the correspondence on \(\mathbb{H}^m\), from which many noncompact hyper-Kähler manifolds are constructed as quotients. One explicitly ends up, thanks to a twistor analysis in the \(m\geq 2\) case, with the quotient of the space \(\mathbb{C}P^{m-1}\times S\), with \(S\) the exterior of an \((m+1)\)-dimensional Siegel domain \(\{\|u_1\|^2>{\text{ Im}} u_2\}\), by the integer action \(u_2\mapsto u_2+2\pi n\).