The paper studies a question in number theory and algebraic geometry about how rational points on geometric spaces are distributed when you track certain algebraic information about them. Specifically, when you have a point in projective space or a Grassmannian (a space whose points represent higher-dimensional subspaces rather than just lines) defined over a number field, you can attach to it an "ideal class," which is an object from algebraic number theory that measures how far the number field is from having unique factorization. The paper investigates which ideal classes actually arise and how the points achieving each class are spread out.
One of the main results is a "weak approximation" theorem for these ideal class fibers. Weak approximation is a classical principle saying, roughly, that rational solutions to a system of equations can be found that simultaneously approximate any desired behavior at finitely many primes. The authors show this holds within each fixed ideal class, meaning that rational points achieving a given ideal class are not just sparse or clumped but are genuinely dense in a suitable sense within linear subspaces and Grassmannians.
The paper also introduces a new local invariant called a "local defect" of a morphism between projective spaces. This is a measure of how a map between these spaces behaves prime by prime. Using these local defects, the authors give a precise description of exactly which ideal classes appear in the image of a linear subspace under such a map: the answer turns out to be a finite union of cosets inside the class group, which is a clean and structured algebraic answer to what could have been a complicated geometric question.