The classical Lefschetz fixed point theorem is a tool in topology that counts the fixed points of a map from a space to itself. It does this by combining local information near each fixed point into a global numerical invariant, the Lefschetz number, which can be computed from the map's action on the homology of the space. The paper extends this theory to a more refined geometric setting: compact complex manifolds that have a boundary made up of components crossing each other in the simplest possible way, called simple normal crossings. The word "logarithmic" refers to a style of calculus adapted to this kind of boundary, where differential forms are allowed to have mild singularities along it.
The main achievement is a precise formula for how to count fixed points when the map sends the manifold to itself in a way that respects the boundary structure. Near a boundary fixed point, the geometry is more subtle than in the interior because the boundary itself has structure, and the paper introduces a careful procedure using rescaling of the normal directions and a duality theory relative to the boundary to assign a well-defined numerical contribution to each fixed point. This takes some work because the usual approaches break down when fixed points sit on or near the boundary.
A key finding concerns when boundary fixed points actually contribute to the formula. The paper distinguishes between "resonant" and "non-resonant" boundary fixed points, depending on a condition involving how the map behaves in the normal directions to the boundary. Non-resonant fixed points turn out to contribute nothing to the Lefschetz count, while resonant ones contribute amounts that capture two geometric features: how the map contacts the boundary in the normal direction, and how it behaves tangentially along the boundary. This gives a clean and computable version of the Lefschetz formula that accounts for all the subtleties introduced by the boundary.