New classification method of dynamical state of galaxy clusters using Gaussian Mixture Model

Galaxy clusters are the largest gravitationally bound systems in the universe, nodes of Large-scale filaments, and environment of galaxy evolution. Therefore, understanding the evolution of galaxy clusters is a very important issue for both cosmological and astronomical perspectives. One effective way of studying the evolution of galaxy clusters is analyzing the dynamical state of galaxy clusters through various components such as galaxies, gas, and stars.

We have been researching a more detailed and merger-process-based analysis for classifying the dynamical states of galaxy clusters. As a result, we developed an observation-based, enhanced dynamical state classification method using a Naive Gaussian Mixture Model (GMM) applied to the N_cluster run simulation data. The GMM classifier was designed to categorize two merger states (merger and relax) as well as three merger states (recent merger, ancient merger, and relax) to provide a more detailed understanding of the merger processes. The results demonstrate improved performance and ease of application to observational data. Our findings indicate that using a larger number of indicators yields better results, and we can get the order of important dynamical indicators for the GMM method. Additionally, our analyses show that a projected classifier (which uses six indicators projected down to five or two) consistently produces better outcomes than non-projected classifiers. Furthermore, the new GMM classifier outperforms our previous results. This enhanced dynamical state classification method and its findings will be utilized in numerous related studies to better understand the evolution of galaxy clusters and the mass assembly history of the universe.