Fast detection of any-size communities based on van der Waals potential

Detecting communities with different sizes is essential for capturing the full topology of complex networks. However, the existing community detection algorithms, are unable to detect communities with small-sizes. In this paper, we propose a simple but effective and efficient community detection algorithm, called Potential. Unlike existing label propagation variants that struggle to identify detailed structures between nodes, Potential overcomes this limitation by incorporating the van der Waals potential, thereby enabling accurate detection of communities with any sizes. The Potential algorithm initializes a unique label for each node. Then, a node is treated as a molecule or an atom, and it is labeled by assigning it the same label as its neighbor that has the maximum van der Waals potential with it. Experimental results demonstrate that Potential can effectively detect both small and large communities, outperforming the state-of-the-art methods in terms of accuracy, stability, and running time on real and synthetic networks in most cases. Additionally, its parameter is easy to set, and it exhibits good convergence.