SO(3)-Donaldson invariants of \(\mathbb CP^{2}\) and mock theta functions (Q441133)

In 1994 gauge theory saw a major revolution with the introduction of the Seiberg-Witten invariants. These were motivated by physical arguments, that predicted that a certain set of equations would lead to a set of invariants of \(4\)-manifolds that would be conjectually equivalent to the Donaldson invariants. Great progress was made quickly, including proofs of the Thom conjecture by Kronheimer and Mrowka, and the 10/8ths theorem of Furuta. It was possible to study the Seiberg-Witten equations without understanding the physical arguments that originally motivated the introduction of the Seiberg-Witten equations. Based on the physical arguments, Moore and Witten conjectured that the generating function for the \(\text{SO}(3)\)-Donaldson invariants of \(\mathbb{C}P^2\) was closely related to modular functions. In this substantial work, Malmendier and Ono prove the Moore-Witten conjecture and derive generating functions for the \(\text{SO}(3)\)-Donaldson invariants with \(2N_f\) massless monopoles. The \(\text{SO}(3)\)-Donaldson invariants count the solutions to the anti-self-dual equations with additional constraints coming from the homology of the manifold. The paper begins with a short summary of the definition of these invariants. Mathematically augmenting Donaldson theory with monopoles amounts to adding a number \(N_f\) of spinor (really \(\text{Spin}_c\)) fields, requiring that the spinor fields are in the kernel of the twisted Dirac operator, and requiring that the self-dual part of the curvature is a certain quadratic expression in these fields. For \(\mathbb{C}P^2\), a Weitzenböck argument implies that only trivial Spinor fields can arise in this way due to the positive curvature of \(\mathbb{C}P^2\). Even so, the monopole theory is slightly different because one may evaluate the fundamental class of the ASD moduli space on powers of the top Chern class of the determinant line bundle associated to the twisted line bundle in addition to the usual \(\mu\) classes from Donaldson theory. This is also explained near the start of this paper. Following this, Malmendier and Ono give an exceptionally clear description of the Seiberg-Witten curve and the \(u\)-plane. They then give a summary of the physical arguments that pure mathematicians will appreciate. The starting point for their proof of the Moore-Witten conjecture is Göttsche's expressions for the \(\text{SO}(3)\)-Donaldson invariants of \(\mathbb{C}P^2\). Then the heart of their argument relies on the theory of mock theta functions and Maass forms.