The paper studies self-maps of projective space, which are geometric transformations that send an n-dimensional projective space back to itself with a certain degree of complexity measured by the number d. A central object of interest is the "critical locus," the set of points where such a map fails to be locally one-to-one, analogous to how a smooth function on the real line has critical points where its derivative vanishes. Understanding the geometry of this critical locus, in particular whether it forms a single irreducible piece rather than breaking into separate components, is an important question in arithmetic dynamics and algebraic geometry.
The main result is that for a "general" or typical such map, the critical locus is as nice as possible: it forms a single, irreducible hypersurface (a variety of one dimension less than the ambient space). This kind of irreducibility result was previously known only when the characteristic of the underlying field does not divide the degree d. The new contribution is handling all characteristics uniformly, including the so-called "wild" case where the characteristic does divide d, which is significantly harder because the usual Jacobian determinant can behave poorly in that setting.
To handle the wild case, the authors introduce a substitute polynomial that mimics the behavior of the Jacobian determinant even when the standard one breaks down. The proof strategy involves carefully constructing explicit examples of maps whose Jacobian is absolutely irreducible, meaning it stays irreducible even after extending to larger fields, and then using the fact that irreducibility is an open condition to conclude the result holds for general maps. The tools are relatively elementary, relying on basic facts about polynomial rings rather than heavy machinery, making the approach notably accessible given the generality of the result.