Symplectic mapping class groups of K3 surfaces and Seiberg-Witten invariants (Q2127882)

Given a symplectic manifold \((X,\omega)\), the symplectic mapping class group of \(X\) is the \(\pi_0\) of the symplectomorphism group \(Symp(X)\). Let \(K(X,\omega)\) be the subgroup of the symplectic mapping class group consisting of elements that are smoothly isotopic to the identity. There are many symplectic manifolds with infinite \(K(X,\omega)\), but the only class of examples that we know to be infinitely generated is symplectic K3 surfaces. The proof of this, by \textit{N. Sheridan} and \textit{I. Smith} [J. Am. Math. Soc. 33, No. 3, 875--915 (2020; Zbl 1482.53111)], combines homological mirror symmetry, Bridgeland stability and makes essential use of the algebraic geometry of K3 surfaces. It is a beautiful and indirect proof. The paper under review gives a more direct proof, which covers more K3 surfaces including non-algebraic ones. The idea of the proof goes as follows. Let \((X,\omega)\) be a symplectic K3 surface. Let \(S_{[\omega]}\) be the space of symplectic forms on \(X\) that are isotopic to \(\omega\) via cohomologous symplectic forms. By the Moser argument and symplectic parallel transport, it defines a monodromy representation \(\pi_1(S_{[\omega]}) \to \pi_0(Symp(X,\omega))\). There are various methods to prove that certain non-trivial elements in \(\pi_1(S_{[\omega]})\) would map to non-trivial elements in the subgroup \(K(X,\omega)\) of \(\pi_0(Symp(X,\omega))\), but the difficult part is to show that there are no unwanted relations in \(\pi_0(Symp(X,\omega))\). For that, the author considers the fine moduli space \(B\) of \([\omega]\)-polarized K3 surfaces, and shows that the composition of the monodromy representation with the (modified) Kronheimer's homomorphism \(\pi_1(B) \to K(X,\omega) \to H_1(B,\mathbb{Z}_2)\) agrees with the natural map \(\pi_1(B) \to \pi_1(B)/[\pi_1(B),\pi_1(B)] \to H_1(B,\mathbb{Z}_2)\). After establishing this, it is not hard to see that \(K(X,\omega)\) is infinitely generated when \(H_1(B,\mathbb{Z}_2)\) has infinite rank. To prove that the two maps agree, the author analyzes the solutions of the family Seiberg-Witten equations associated to the (modified) Kronheimer's homomorphism. The argument turns out to be nice and short.