Phase transition in preferential attachment-detachment through embedding

Michael Hinz,Angelica Pachon
2024-10-14
Abstract:We study a random graph model with preferential edge attachment and detachment through the embedding into a generalized Yule model. We show that the in-degree distribution of a vertex chosen uniformly at random follows a power law in the supercritical regime but has an exponential decay in the subcritical. We provide the corresponding asymptotics. In the critical regime we observe an intermediate decay. The regimes are clearly defined in terms of parameter ranges.
Probability,Combinatorics
What problem does this paper attempt to address?
The problem that this paper attempts to solve is to explore the asymptotic behavior of the in - degree distribution of nodes and its phase - transition characteristics after introducing the edge - detachment mechanism in the network graph model. Specifically, the researchers constructed a random graph model with preferential attachment and edge - detachment, and analyzed its behavior under different parameter ranges by embedding it into the generalized Yule model. ### Research Background and Problem Description 1. **Preferential Attachment Model**: The Preferential Attachment model is a discrete - time random graph model, in which new nodes are connected to existing nodes with a probability proportional to their in - degree. This mechanism can well simulate the concentration of resources in the market environment or the concentration of links in the communication network. 2. **Edge - Detachment Mechanism**: In order to be closer to the real situation, this model also introduces the edge - detachment mechanism, that is, some connections may disappear due to changes in market conditions or sudden loss of attention. This mechanism makes the model more complex, but also more in line with the actual situation. 3. **Phase - Transition Phenomenon**: The researchers found that by adjusting the parameters in the model, different asymptotic behaviors of the in - degree distribution can be observed. Specifically, in three different parameter ranges: supercritical, subcritical and critical, the in - degree distribution of nodes shows power - law decay, exponential decay and decay behavior between the two, respectively. ### Main Contributions - **Embedding in the Generalized Yule Model**: By embedding the preferential - attachment - detachment model into the generalized Yule model, the researchers provided a concise method to describe these different asymptotic behaviors. - **Complete Parameter - Range Description**: The paper gives a complete description of the model behavior under different parameter ranges, including that the exponent of power - law decay can take any value in the range of \((1, +\infty)\). - **Asymptotic Behavior Analysis**: Analyzed in detail the asymptotic behavior of the in - degree distribution of nodes under different parameter ranges and provided the corresponding mathematical expressions. ### Formula Summary For given parameters \(\lambda_1, \lambda_2>0\) and \(\mu_2\geq0\), the limit behavior of the in - degree distribution of nodes is as follows: - **Supercritical Region (\(\mu_2 = 0\) or \(\lambda_2>\mu_2\))**: \[ p_j\sim c_2j^{-(1 + \frac{\lambda_1}{\lambda_2-\mu_2})}, \quad \text{where} \quad c_2=\frac{\lambda_1}{\lambda_2}(\frac{\lambda_2}{\lambda_2-\mu_2})^{\frac{\lambda_1}{\lambda_2-\mu_2}}\Gamma(1 + \frac{\lambda_1}{\lambda_2-\mu_2}) \] - **Subcritical Region (\(\mu_2>0\) and \(\lambda_2<\mu_2\))**: \[ p_j\sim c_3(\frac{\lambda_2}{\mu_2})^{j + 1}j^{-(1 + \frac{\lambda_1}{\mu_2-\lambda_2})}, \quad \text{where} \quad c_3=\frac{\lambda_1}{\mu_2}(\frac{\mu_2}{\mu_2-\lambda_2})^{-\frac{\lambda_1}{\mu_2-\lambda_2}}\Gamma(1 + \frac{\lambda_1}{\mu_2-\lambda_2}) \] - **Critical Region