Genus zero Gopakumar-Vafa type invariants for Calabi-Yau 4-folds (Q1789483)

In [Commun. Math. Phys. 281, No. 3, 621--653 (2008; Zbl 1157.32022)], \textit{A. Klemm} and \textit{R. Pandharipande} extended 3D Gopakumar-Vafa invariants \(n_{g,\beta}\) (integers that count BPS states) to Calabi-Yau \(4\)-folds \(X\) by defining them in terms of the Gromov-Witten invariants, as usual. At the center of this paper is the conjecture that in genus zero \(n_{0,\beta}\) coincide with the 4D Donaldson-Thomas invariants introduced by Cao-Leung and Borisov-Joyce (for a suitable choice of deformation invariant orientation). A heuristic argument, imagining that all the moduli spaces of curves on \(X\) are smooth of expected dimension, is given to support the conjecture. The conjecture is proved for the following examples. {\parindent=0.7cm \begin{itemize}\item[{\(\bullet\)}] Elliptic fibrations \(\pi:X\to\mathbb{P}\), when \(\beta\) is a positive integer multiple of the general fiber class of \(\pi\). \item[{\(\bullet\)}] Direct products \(Y\times E\), where \(Y\) is a 3D Calabi-Yau, \(E\) is an elliptic curve, and \(\beta\) is a primitive curve class in \(H_2(Y)\). \item[{\(\bullet\)}] Total spaces of \(\mathcal{O}_S\oplus K_S\), where \(S\) is a smooth toric del Pezzo surface, and \(\beta\) is any curve class, or \(S\) is a rational elliptic surface, and \(\beta\) is some special primitive class. \end{itemize}} An equivariant version of the conjecture is also given for local smooth projective curves \(C\). It is proved for \(g(C)\geq1\), and, with some restrictions on the degrees of bundles over \(C\), for \(g(C)=0\). For genus \(1\) invariants \(n_{1,\beta}\) the authors give a speculative interpretation in terms of Pandharipande-Thomas stable pair invariants.