A Fourier-Mukai transform for real torus bundles. (Q1781423)

Building on previous work by \textit{D. Arinkin} and \textit{A. Polishchuk} [Fukaya category and Fourier transform. AMS/IP Stud. Adv. Math. 23, 261--274 (2001; Zbl 1082.14522)] and \textit{U. Bruzzo, G. Marelli} and \textit{F. Pioli} [J. Geom. Phys. 41, No. 4, 312--329 (2002; Zbl 1071.14515)], [J. Geom. Phys. 39, No. 2, 174--182 (2001; Zbl 1058.14505)], in this paper the authors develop a theory of a transform of the Fourier-Mukai type for (partial) connections on vector bundles whose base manifold is the total space of a flat torus bundle \(M\to B\). Fourier-Mukai transforms of this kind are defined for \textit{foliated bundles} (i.e., families of flat vector bundles on the fibres of the torus bundle \(M\to B\)). The objects that one obtains by transforming such bundles are \textit{spectral covers} of the base manifold \(B\) contained in the mirror fibration \(\hat M \to B\). Applications are given to locally unitary systems, instantons and monopoles on 4-dimensional circle fibrations. Spectral covers are not smooth as submanifolds of the mirror fibration \(\hat M \to B\). The authors claim that this difficulty may be circumvented by using the theory of smooth spaces à la Frölicher [see e.g, \textit{A. Frölicher} and \textit{A. Kriegl}, Linear spaces and differentiation theory, Wiley Series in Pure and Applied Mathematics; Wiley-Interscience Publication. (Chichester (UK) etc.: Wiley) (1988; Zbl 0657.46034)]. However the proofs, as presented in the paper, never give any detail about the use of this notion. As a matter of fact some of the intermediate results mentioned in this paper appear to be incorrect. However a careful analysis of the proofs, where one fills in all the details concerning smooth spaces, shows that the main statements are correct.