On integral rigidity in Seiberg-Witten theory
Francesco Lin, Mike Miller Eismeier
TL;DR
This work develops an integral rigidity framework for Seiberg-Witten invariants on closed 4-manifolds containing a non-separating hypersurface by leveraging a self-gluing viewpoint and a detailed chain-level analysis of negative-definite cobordisms. A key contribution is a filtration-compatible description of the induced cobordism maps on monopole Floer homology, with the associated graded map matching cup-homology calculations, enabling explicit trace formulas. The authors introduce RSF-spaces to capture tractable Floer-theoretic conditions and prove a general rigidity theorem: sums of SW invariants tied to a fixed end data on Y are determined by cobordism data $c(W,Y,\mathfrak{s}_Y)$ when $b^+(W)=0$, and vanish otherwise; this yields concrete determinant-type results for Y = $T^3$ and, in particular, for homology 4-tori. The results connect to Meng–Taubes, Donaldson’s TQFT perspective, and Ruberman–Strle-type rigidity, and are illustrated through explicit calculations and examples involving non-separating tori and mapping tori, providing a coherent framework for rigidity phenomena in 4-manifold topology.
Abstract
We introduce a framework to prove integral rigidity results for the Seiberg-Witten invariants of a closed $4$-manifold $X$ containing a non-separating hypersurface $Y$ satisfying suitable (chain-level) Floer theoretic conditions. As a concrete application, we show that if $X$ has the homology of a four-torus, and it contains a non-separating three-torus, then the sum of all Seiberg-Witten invariants of $X$ is determined in purely cohomological terms. Our results can be interpreted as $(3+1)$-dimensional versions of Donaldson's TQFT approach to the formula of Meng-Taubes, and build upon a subtle interplay between irreducible solutions to the Seiberg-Witten equations on $X$ and reducible ones on $Y$ and its complement. Along the way, we provide a concrete description of the associated graded map (for a suitable filtration) of the map on $\overline{HM}_*$ induced by a negative cobordism between three-manifolds, which might be of independent interest.
