The paper tackles a classical problem in mathematical analysis called Hardy inequalities. These inequalities relate the "energy" of a function to a weighted version of that function, and they appear throughout physics and mathematics, particularly in quantum mechanics and the study of differential equations. The central challenge is finding the best possible "weight" in such an inequality, meaning the largest function you can use while the inequality still holds. The authors work in a fairly general geometric setting, considering not just flat space but curved spaces called Riemannian manifolds, as well as discrete networks called graphs.
The main technical contribution is a new way to construct these optimal weights using something called a Green potential, which is a tool borrowed from classical potential theory. Roughly speaking, the Green potential encodes how a differential operator "spreads" influence across a space, similar to how a point source of heat diffuses outward. By choosing this potential carefully, with a strictly positive charge that fades away at the boundary of the region, the authors build a whole family of Hardy weights controlled by a single parameter ranging from 0 to 1. A key discovery is that as this parameter crosses zero, the nature of the inequality changes qualitatively, transitioning between two different regimes called positive-criticality and null-criticality. This kind of sharp transition had not been observed in this construction before.
The authors also pin down the best possible constant in their Hardy inequality, which turns out to have a clean formula depending only on the exponent p in the underlying operator. At the special parameter value of zero, they prove their weight is optimal in the strongest possible sense, meaning no larger weight could work. The fact that identical results hold on graphs, which are discrete rather than continuous objects, shows the underlying ideas are quite robust and not tied to smooth geometry. This work advances the understanding of how geometry and operator theory interact to constrain the form of fundamental analytic inequalities.