The paper studies certain geometric spaces called character varieties and Higgs bundle moduli spaces. These spaces are built by taking a group G and looking at all the ways it can act on simpler mathematical objects, like free abelian groups or geometric shapes called abelian varieties. Because these spaces are constructed by quotienting out symmetries, they tend to have singularities, meaning they have points where the usual smooth geometric structure breaks down. The authors compute "stringy invariants" for these spaces, which are numerical quantities originally inspired by string theory that serve as a refined way to measure the shape and topology of singular spaces, essentially encoding information about their cohomology while carefully accounting for the singularities.
One major result is a direct proof of a phenomenon called topological mirror symmetry. In physics, mirror symmetry is a duality between two seemingly different geometric spaces that nonetheless share deep mathematical properties. Here, the relevant pairs of spaces come from swapping G for its Langlands dual group, a partner group determined by flipping the role of the group's root system. The authors show that the stringy invariants of these paired spaces are equal, confirming the mirror symmetry statement in a clean, self-contained way without relying on indirect arguments.
The paper also investigates when these singular spaces can be "resolved," meaning replaced by a smooth space that closely approximates the original. Such resolutions with especially nice properties, called symplectic resolutions, are rare and valuable. The authors show that symplectic resolutions can only exist when G belongs to certain families of groups, specifically those of Dynkin type A, B, or C. In the cases where resolutions do exist, the stringy invariants match the ordinary topological invariants of the smooth resolved space, providing a satisfying consistency check and deeper geometric meaning for the stringy quantities.