Polyfolds: a first and second look (Q502137)

The goal of this paper is to give an exposition of the general polyfold theory, developed by Hofer, Wysocki and Zehnder since 2004, accessible to a reader familiar with the general ideas of elliptic differential equations, compactification of moduli, and Banach spaces. The applications include Gromov-Witten theory, Floer homology, symplectic field theory, and the point is to develop an abstract nonlinear Fredholm theory that uniformly resolves the issues of compactness, convergence to singular curves, and transversality. The end result is a theory with an abstract perturbation scheme coupled with the implicit function theorem, which removes the need for most of the ad hoc analytic constructions that currently fill the literature. The polyfold theory is expected to interpret any compactified moduli space as the zero set of a ``scale smooth'' Fredholm section \(\sigma\) in a ``polyfold bundle''. Once the compactness of \(\sigma^{-1}(0)\) is established the theory provides arbitrarily small perturbations \(p\) such that \(\sigma+p\) is transverse to the zero section, the zero set of the perturbed section is a compact finite dimensional manifold (or orbifold when discrete automorphisms are present) with boundary/corners, and the zero sets of different perturbations are cobordant. The paper is split into two parts. The first, ``philosophical'', part presents the key ideas informally, using the gradient flow trajectories of the finite dimensional Morse theory as toy examples. The second, ``technical'', part streamlines formal definitions and key theorems to get to the abstract transversality result quickly. For proofs the reader is mostly referred to the original sources. There are two main issues that are addressed by the polyfold construction. First, the reparametrization action of automorphisms on infinite dimensional spaces of maps is not differentiable in any familiar Banach topology. In applications this forces one to look for equivariant transverse perturbations before taking the reparametrization quotient, which can only be found under severe restrictions on allowed maps, like injectivity. The polyfold theory instead uses a notion of differentiability in a Banach scale, which imparts the scale smooth structure to the space of re\-para\-metri\-zation-equivalence classes of maps, and interprets the Cauchy-Riemann operator as a smooth section over this space. Second, the compactification of the moduli usually requires maps over singular domains, which precludes a description of the compactified moduli as a subset of a single Banach manifold of maps. The polyfold theory introduces the notion of sc-retract (formalizing the pregluing construction), which allows one to build ambient spaces with neighborhoods of singular maps containing both singular and regular maps. Thus, part of the gluing construction is formalized by a Fredholm condition on the Cauchy-Riemann operator at singular maps, and the other parts are replaced by an implicit function theorem for Fredholm sections over sc-retracts.