STAR CLUSTER MEMBERSHIP IDENTIFICATION BY SUPERVISED MACHINE LEARNING MODELS APPLIED TO N-BODY SIMULATIONS

Loading...
Thumbnail Image

Date

2023

Authors

Bissekenov, Abylay

Journal Title

Journal ISSN

Volume Title

Publisher

School of Sciences and Humanities

Abstract

This thesis investigates possible ways to apply supervised machine learning algorithms on N-body simulations. Because of the limitations of observational data, there is a motivation to research star clusters by the N-body simulations. The simulations used for the study are based on the Plummer model, and each has its star formation efficiency (SFE) and several random realizations. A random forest model was trained on the simulation with 15% star formation efficiency on a 20-100 Myr timeframe. The model was tested on the other N-body simulations with 17-25% SFEs and showed high classification accuracy throughout the whole dynamic evolution of tested simulations. The majority of mistakes of the model were the false positives (FP) that turned out to be within a 2 Jacobi radius, indicating that they might be gravitationally bounded to center of cluster. Framework and learning strategy can be considered effective and further applied for the mock observations of N-body simulations.

Description

Keywords

Type of access: Embargo, Star clusters, N-body simulation, Machine Learning, Supervised Learning

Citation

Bissekenov, A. (2023). Star Cluster Membership Identification By Supervised Machine Learning Models Applied To N-Body Simulations. School of Sciences and Humanities