[Paper Review] Lamplighter graphs do not admit harmonic functions of finite energy

Agelos Georgakopoulos|ArXiv.org|Jul 8, 2009

Graph theory and applications12 references3 citations

TL;DR

This paper proves that lamplighter graphs $G \wr H$, where $G$ is a locally finite graph and $H$ is a finite connected graph with at least one edge, do not admit non-constant harmonic functions of finite Dirichlet energy. The proof relies on a strengthened version of a structural lemma on edge-disjoint paths between rays, showing that such graphs lack the geometric conditions required for finite-energy harmonic functions to exist, extending known results on grids and trees.

ABSTRACT

We prove that a lamplighter graph of a locally finite graph over a finite graph does not admit a non-constant harmonic function of finite Dirichlet energy.

Motivation & Objective

• To resolve a question posed by A. Karlsson on whether lamplighter graphs over regular trees admit non-constant harmonic functions of finite Dirichlet energy.
• To generalize known results on grids ($\mathbb{Z}^d$, $d \geq 3$) to arbitrary locally finite graphs $G$ and finite graphs $H$.
• To establish a sufficient condition—based on path growth and edge-disjointness—under which graphs do not support non-constant finite-energy harmonic functions.
• To provide a general framework applicable beyond lamplighter graphs, using a strengthened version of a result by Markvorsen, McGuinness, and Thomassen.

Proposed method

• A new lemma (Lemma 3.1) is developed that strengthens a result on edge-disjoint paths between rays in graphs, linking path growth rates to the non-existence of finite-energy harmonic functions.
• The proof constructs a sequence of pairwise edge-disjoint paths $P_i$ between two rays in the lamplighter graph $G \wr H$, with lengths growing at most linearly in $i$, satisfying the condition in Lemma 3.1.
• The construction uses the blow-up structure of $G \wr H$, where each vertex of $G$ is replaced by a copy of $H$, and paths are built using switching edges and tree-based navigation in the base graph $G$.
• The length of each path $P_i$ is bounded by a linear function of $i$, using diameter and edge-count bounds on finite subtrees of $G$, and leveraging the fixed size of $H$ to control switching path lengths.
• The argument relies on the fact that the Dirichlet energy of a harmonic function is finite only if such path constructions do not exist, leading to a contradiction if a non-constant finite-energy harmonic function existed.
• The method applies to general graphs, not just Cayley graphs, and does not require group structure, making the result broadly applicable.

Experimental results

Research questions

• RQ1 Does the lamplighter graph $T \wr \mathbb{Z}_2$, where $T$ is a regular tree, admit a non-constant harmonic function of finite Dirichlet energy?
• RQ2 Can the non-existence of non-constant finite-energy harmonic functions be extended from infinite grids to arbitrary locally finite graphs $G$ when $H$ is finite and connected?
• RQ3 Is there a general graph-theoretic condition—based on ray structure and path growth—that implies the non-existence of non-constant harmonic functions of finite energy?
• RQ4 Can the result be extended to cases where $H$ is infinite and locally finite, as posed in Problem 3.1?

Key findings

• Lamplighter graphs $G \wr H$ do not admit any non-constant harmonic function of finite Dirichlet energy when $G$ is locally finite and $H$ is a finite connected graph with at least one edge.
• The non-existence result holds even when $G$ is not a Cayley graph, indicating a purely graph-theoretic obstruction.
• The proof establishes that the existence of a sequence of edge-disjoint paths between two rays with linearly bounded length implies the non-existence of non-constant finite-energy harmonic functions.
• The construction of such paths in $G \wr H$ relies on the blow-up structure and the use of spanning trees in finite subgraphs of $G$, ensuring path disjointness via lamp state differences.
• The length of each constructed path $P_i$ grows at most linearly with $i$, satisfying the condition in Lemma 3.1, which is sufficient to rule out finite-energy harmonic functions.
• The result generalizes known results on grids and provides a new tool for analyzing harmonic functions on product-type graphs.