Primary operations in differential cohomology (Q1669041)

The paper at hand analyses the analog of primary cohomology operators for differential cohomology. In classical homology theory, (unstable) operations correspond to homotopy classes of maps between the classifying spaces of cohomology groups. Differential cohomology associates in a functorial way the so-called differential cohomology groups to smooth manifolds. These functors do not satisfy the axioms for a (generalized) cohomology theory: they are not homotopy invariant. Instead, they combine homological information with differential cohomology information encoded in differential form data. In the approach of the article, they are classified by certain stacks, and the corresponding operations therefore correspond to morphisms of these classifying stacks. It is shown in the paper that, equivalently, a natural transformation of differential cohomology (always considered with \(\mathbb{Z}\) coefficients) is a compatible collection of \begin{itemize} \item an ordinary (unstable) operation on integral cohomology \(H^n(-;\mathbb{Z})\to H^m(-;\mathbb{Z})\), \item on de Rham cohomology (of the same degrees), \item and on closed differential forms (again of the same degrees). \end{itemize} The main result is a classification of such differential cohomology operators. They are either of the form \[x\mapsto n x^m\text{ for }n\in\mathbb{Z},\; m\ge 0\] or are obtained from an ordinary cohomology operator \(\phi\colon H^n(-;\mathbb{Z})\to H^{m-1}(-;\mathbb{R}/\mathbb{Z})\) by pre- and postcompoosition with the standard functors from differential cohomology to ordinary cohomology (``take the underlying cohomology class'') and from ordinary cohomology with \(\mathbb{R}/\mathbb{Z}\)-cofficients to differential cohomology (``flat inclusion''). The paper also formulates a Künneth theorem for differential cohomology, and uses an extension (to certain non-manifolds) for the discussion of differential cohomology refinements of Steenrod operators (lifted to integral cohomology). This Künneth formula computes \(\hat H^n(X\times Y;\mathbb{Z})\) in terms of \(\hat H^n(X)\) and ordinary cohomology of \(Y\) with integer coefficients as well as of \(X\) with \(\mathbb{R}/\mathbb{Z}\)-coefficients. One should be cautious here: in the generality formulated in the paper, the statement is not correct. Easy counterexamples are obtained with \(X=\{*\}\) and \(Y\) any higher dimensional manifold. In private communication to the reviewer, the authors explain that the assertion is correct if \(Y\) is geometrically discrete and of finite type. For the case of ordinary manifolds, this means that \(Y\) is a finite set. Therefore, Remark~7 doesn't make sense. Moreover, the generalization used in the paper to the simplicial set \(B\mathbb{Z}/2\) still does hold, as this is geometrically discrete and of finite type (the latter means: finitely many simplices in each degree). The Künneth formula is used to analyse differential Steenrod operations and their relation to taking powers in differential cohomology.