Moonshine for \(M_{24}\) and Donaldson invariants of \(\mathbb{C} \text{P}^2\) (Q2392990)

This paper is a sequel to [Geom. Topol. 16, No. 3, 1767-1833 (2012; Zbl 1255.57028)], and connects the \(u\)-plane integrals of Moore and Witten for a certain mock modular form \(H(\tau)\) with \({\text{SO}}(3)\)-Donaldson invariants for \({\mathbb C}{\mathbb P}^2\). \(H(\tau)\) is the mock modular form associated with the \(K3\) elliptic genus \[ Z(z,\tau)=8\left[\left({{\theta_{10}(z,\tau)}\over{\theta_{10}(0,\tau)}}\right)^2+ \left({{\theta_{00}(z,\tau)}\over{\theta_{00}(0,\tau)}}\right)^2+ \left({{\theta_{01}(z,\tau)}\over{\theta_{01}(0,\tau)}}\right)^2\right] \] by the formula \[ Z(z,\tau)={{\theta_{11}(z,\tau)^2}\over{\eta(\tau)^3}}\big(24\mu(z,\tau)+H(\tau)\big), \] where \[ \mu(z,\tau)={{ie^{i\pi z}}\over{\theta_{11}(z,\tau)}}\sum_{n\in{\mathbb Z}} (-1)^n{{q^{n(n+1)/2}e^{2\pi i nz}}\over{1-q^n e^{2\pi iz}}}. \] \(H(\tau)\) is the holomorphic part of a weight-\({{1}\over{2}}\) harmonic Maass form, and has a \(q\)-expansion \[ H(\tau)=2q^{-{{1}\over{8}}}\left( -1+\sum_{n=1}^\infty A_nq^n\right). \] The coefficients \(A_n\) are integers, the first few being \(A_1=45\), \(A_2=231\), \(A_3=370\), \(A_4=2277\)\dots. It was observed by Eguchi, Ooguri, Taormina and others that these are simple linear combinations of dimensions of irreducible representations of the Mathieu group \(M_{24}\). A precise, but still conjectural, statement on these lines is known as the umbral moonshine conjecture, by analogy with the famous monstrous moonshine relating representations of the Monster and coefficients of the \(j\)-function. The main theorem here is that the \(u\)-plane integrals \({\mathbf D}_{m,2n}\big[H(\tau)\big]\) are related to the \({\text{SO}}(3)\)-Donaldson invariants \({\mathbf \Phi}_{m,2n}\) for \({\mathbb C}{\mathbb P}^2\), very simply by \[ 12{\mathbf \Phi}_{m,2n}={\mathbf D}_{m,2n}\big[H(\tau)\big]. \] Since the authors [loc.\ cit.] have already shown how to express \({\mathbf D}_{m,2n}\big[H(\tau)\big]\) in terms of the coefficients \(A_n\), this gives a conjectural expression for \({\text{SO}}(3)\)-Donaldson invariants for \({\mathbb C}{\mathbb P}^2\) in terms of dimensions of representations of \(M_{24}\). The proof is an application of the authors' earlier results. It is not particularly easy, but neither does it require new techniques. It is the result that is the real point of this paper.