OPEN CLUSTER MEMBERSHIP DETECTION FROM GAIA DR3 DATA USING THE SELF-ORGANIZING MAPS METHOD

• Main Author: Ikrar Yulistiawibawa, Muhammad
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/77120
• Institution: Institut Teknologi Bandung

Open clusters are a group of stars consisting of dozens, and even up to thousands, of stars that share the same age, formed from the same molecular cloud, and are gravitationally bound to each other. Open clusters are found in spiral and irregular galaxies, where new stars are currently forming, but not in elliptical galaxies. Within the Milky Way galaxy, open clusters are found both inside and between the spiral arms. Gaia is a European space mission providing astrometric, photometric, and spectroscopic data for over one million stars in the Milky Way galaxy. Gaia data is also available for a significant sample of extragalactic and solar system objects. Additionally, Gaia data includes information about positions, parallaxes, proper motion, radial velocities, and brightness. In this current era of digitization, rapid technological advancements have brought about changes in various fields, including Astronomy. Artificial Intelligence is one branch of computer science that utilizes machine technology, specifically computer systems, to simulate human intelligence processes. Given these conditions, the author is interested in harnessing the progress of machine learning-based computational technology to detect membership in open clusters using the Self-Organizing Maps (SOM) method from Gaia DR3 data. SOM requires input on the number of clusters to be created, hence the Silhouette Coefficient validation cluster is used to evaluate SOM's performance. In this final project, the author uses three different input parameter variations for each open cluster. The two clusters under examination are a closely spaced pair of clusters, NGC 5617 - Trumpler 22, and a pair of clusters with a small separation, King 16 - Berkeley 4. The results from SOM clustering demonstrate satisfactory grouping of clusters. For the first two parameters related to the issue of very close pair clusters, SOM successfully separates the NGC 5617 - Trumpler 22 pair but encounters some difficulty in distinguishing NGC 5617 from another object, Pismis 19. However, in the last parameter, SOM is able to separate NGC 5617 from Pismis 19. For the issue of clusters with small separation, SOM works very well in separating the two pairs, King 16 - Berkeley 4, along with other objects that are not part of the pair, such as Dias 1. The results indicate that SOM effectively groups closely spaced double clusters, while for pairs with small separations, SOM performs exceptionally well.