Clustering–Based Time Resolved Spectral Investigations Of Bursts From Magnetar Sgr J1550−5418

Warning The system is temporarily closed to updates for reporting purpose.

Demirer, Mustafa (2023) Clustering–Based Time Resolved Spectral Investigations Of Bursts From Magnetar Sgr J1550−5418. [Thesis]

PDF
10612532-demirer.pdf

Download (1MB)

Abstract

Magnetars, strongly magnetized neutron stars, are the sources of short duration but extremely energetic hard X-ray bursts. This thesis presents a time-resolved spectral analysis of 42 bursts originated from SGR J1550−5418. Our study introduces an innovative approach to time-resolved spectral analysis: Initially, we created overlapping time segments and fitted them using three models: a comptonized model, a double blackbody model, and a modified blackbody model with resonance cyclotron scattering. Subsequently, we tested four distinct algorithms for clustering overlapping time segments, namely; K-means clustering, DBSCAN, agglomerative clustering, and Gaussian mixture. The K-means algorithm was ultimately selected for its effectiveness. After that, we created non-overlapping time segments by fitting the clustered time segments. We employed the Bayesian Information Criterion (BIC) for model comparison. As a result, we found that the COMPT model is most favorable for the most fits, with approximately half of the time segments also being favored by the other two models. Additionally, we observed a deviation from Stefan- Boltzmann trend in kT vs R2 plot of the double blackbody model. The most notable aspects of this study are that it is the first extensive application of the MBB−RCS model and our novel method combining overlapping time segments with clustering analysis.
Item Type: Thesis
Uncontrolled Keywords: Neutron star, magnetar, X-ray burst, machine learning. -- nötron yıldızı, X ışını patlaması, makine öğrenmesi.
Subjects: Q Science > QC Physics
Divisions: Faculty of Engineering and Natural Sciences > Basic Sciences > Physics
Faculty of Engineering and Natural Sciences
Depositing User: Dila Günay
Date Deposited: 03 Sep 2024 15:00
Last Modified: 03 Sep 2024 15:00
URI: https://research.sabanciuniv.edu/id/eprint/49872

Actions (login required)

View Item
View Item