Please use this identifier to cite or link to this item:
https://www.um.edu.mt/library/oar/handle/123456789/92886
Title: | C-FORGE : clustering stellar formations in galaxy evolution simulations |
Authors: | Gatt, Benjie (2011) |
Keywords: | Space simulators Astrophysics Galaxies Galaxies -- Clusters Algorithms |
Issue Date: | 2011 |
Citation: | Gatt, B. (2011). C-FORGE : clustering stellar formations in galaxy evolution simulations (Bachelor’s dissertation). |
Abstract: | There is a desire to find the sister stars of the Sun - those stars born in the same stellar cluster. A stellar cluster is a structure found in a galaxy, consisting of a high density group of stars which were born at roughly the same age, and are moving with a similar velocity. By studying the behaviour of such clusters, it is hoped to locate suitable candidates for the sister stars. Simulations are frequently run to model the birth and subsequent evolution of a galaxy, and stars can be seen to form in clusters in these simulations. Manually locating the clusters is difficult and time consuming - a tool able to automatically find these clusters would be a useful aid to researchers. This is not a simple clustering problem for two main reasons - the large scale of the data, as the number of stars in the simulations is typically hundreds of thousands, or millions, and the fact that stellar clusters may have interlopers which are not part of them but simply passing through. C-FORGE is an attempt to provide a solution to these problems, by identifying stellar clusters in a reasonable amount of time relative the amount of data it has to process. A cluster comparison tool was also devised to aid the tracking of clusters across different time steps. The clustering algorithm used is inspired by DBSCAN, a density based clustering algorithm. In this approach, clusters are defined as areas of high density. Certain stars, called core stars, are used as seeds, and stars which are density connected to the cluster are assigned to it. A star is assigned a core star by virtue of having a high enough number of stars in its neighbourhood (of a predefined distance). Density connectedness implies that the stars are at worst within the neighbourhood of a star itself in the neighbourhood of a core star. C-FORGE was able to successfully cluster stars in the data it is provided with, and eliminate stars in those clusters found to differ significantly from a given cluster. The comparison tool is also able to compare different time steps, classifying clusters according to the number of previously clustered stars. |
Description: | B.SC.(HONS)COMP.SCI.ARTIFICIAL INT.&PHYSICS |
URI: | https://www.um.edu.mt/library/oar/handle/123456789/92886 |
Appears in Collections: | Dissertations - FacICT - 2011 Dissertations - FacICTCS - 2010-2015 |
Files in This Item:
File | Description | Size | Format | |
---|---|---|---|---|
BSC(HONS)COMPSCIARTIFICIAL INT_PHYSICS_Gatt_Benjie_2011.PDF Restricted Access | 17.38 MB | Adobe PDF | View/Open Request a copy |
Items in OAR@UM are protected by copyright, with all rights reserved, unless otherwise indicated.