Researchers identify and classify optimal weight factors for Hardy inequalities on graphs, including fractional Laplacians
The paper gives a clear description of which weight functions make a Hardy inequality tight for Laplace operators on weighted graphs. In plain terms, a Hardy inequality compares the energy of a function on a graph with a weighted sum of its square values. The authors first give a general criterion that characterizes all “positive-critical” Hardy weights for broad classes of graph Laplacians. They then apply that criterion to fractional Laplacians on graphs to find, under extra geometric and heat-kernel assumptions, an optimal Hardy weight for those operators. They illustrate the theory on several kinds of graphs, including Cayley graphs, graphs with curvature bounds, and fractal graphs.
A Hardy inequality here is an estimate of the form Q(φ) ≥ Σx w(x) φ(x)^2 m(x), where Q(φ) measures the energy of φ (a graph version of the Dirichlet integral), m is a measure on the graph vertices, and w is a nonnegative weight. A weight is called critical if you cannot make it larger and still have the inequality. Positive-critical means the inequality actually has a minimizer (a ground state that is square summable with respect to wm). Null-critical means the weight is critical but there is no square-integrable minimizer. These distinctions matter because null-critical weights are the optimal ones in the sense used by the paper.
The first main result (Theorem 1) gives several equivalent ways to recognize a positive-critical Hardy weight on a general graph. Concretely, a nonnegative function w is positive-critical if and only if it can be written as L v / v for some strictly positive function v that is superharmonic and belongs to the natural energy space D0 (this v is related to a ground state). The theorem also gives an equivalent description in terms of the graph Green’s function: such w arise from suitable charges through the Green operator, provided a certain integrability condition holds. These concrete representations make it possible to both identify and construct positive-critical weights.