Abstract:We consider the averaging process on an infinite connected graph with bounded degree and independent, identically distributed starting values or initial opinions. Assuming that the law of the initial opinion of a vertex has a finite second moment, we show that the opinions of all vertices converge in $L^2$ to the first moment of the law of the initial opinions. A key tool in the proof is the Sharing a drink procedure introduced by Olle Häggström.
What problem does this paper attempt to address?
This paper attempts to address the issue of opinion convergence in nodes during the averaging process on an infinite connected graph. Specifically, the authors investigate whether, on an infinite connected graph with finite degree, all nodes' opinions will converge in the \(L^2\) sense to the first moment \(m_1\) of the initial opinions when the initial opinions are independently and identically distributed with a finite second moment.
### Main Problem
- **Research Background**: On finite connected graphs, it is known that all nodes' opinions will converge to the sample mean of the initial opinions. Many studies focus on the speed of convergence. For infinite connected graphs, a natural conjecture is that if the initial opinions are independently and identically distributed with a finite first moment \(m_1\), then all nodes' opinions will almost surely converge to \(m_1\).
- **Research Objective**: The goal of this paper is to prove that on an infinite connected graph with finite degree, when the initial opinions are independently and identically distributed with a finite second moment, all nodes' opinions will converge in the \(L^2\) sense to \(m_1\).
### Main Contributions
- **Main Result**: The authors prove that on an infinite connected graph with finite degree, when the initial opinions are independently and identically distributed with a finite second moment, all nodes' opinions will converge in the \(L^2\) sense to \(m_1\).
- **Key Tools**: The authors use the non-random pairwise averaging process (Sharing a drink, SAD) proposed by Olle Häggström to prove this result. Gollin et al. extended the upper bound of the maximum liquid amount in the SAD process from trees to general graphs, which played a crucial role in the proof.
### Method Overview
- **Definition of Averaging Process**: On a graph \(G = (V, E)\), each node carries a real value (opinion), and edges update according to a Poisson process. When an edge updates, the opinions of its two endpoints are averaged.
- **SAD Process**: The SAD process is a non-random pairwise averaging process where initially only one node has liquid (opinion), and other nodes are empty. Over time, neighboring nodes share liquid, eventually forming a stable distribution.
- **Proof of Convergence**: By analyzing the properties of the SAD process, particularly the upper bound of the maximum liquid amount, the authors prove the \(L^2\) convergence of nodes' opinions.
### Conclusion
Through rigorous mathematical proof, this paper verifies that on an infinite connected graph with finite degree, when the initial opinions are independently and identically distributed with a finite second moment, all nodes' opinions will converge in the \(L^2\) sense to the first moment \(m_1\) of the initial opinions. This result provides an important theoretical foundation for understanding the averaging process on complex networks.