An Introduction of A Data-driven Model for Analyzing Stellar Spectra
(2019) In Master's Theses in Mathematical Sciences NUMM11 20191Mathematics (Faculty of Engineering)
Centre for Mathematical Sciences
- Abstract
- This thesis describes a method for concluding attributes of a star (including ages, temperatures, chemical element abundances) from its spectra. The method was developed by the 4MOST (4-metre Multi-Object Spectroscopic Telescope) consortium. How the model outputs stellar attributes by receiving spectra is too complicated to be included in one thesis, so only introductions of numerical methods which are used during the process will be describe in detail here.
The first part of the thesis is about how to determine the coefficients of the model by minimising a cost function. The least square method is used to make up the main part of the cost function and the Lasso regularization term is added in order to avoid the overfitting. Due to the... (More) - This thesis describes a method for concluding attributes of a star (including ages, temperatures, chemical element abundances) from its spectra. The method was developed by the 4MOST (4-metre Multi-Object Spectroscopic Telescope) consortium. How the model outputs stellar attributes by receiving spectra is too complicated to be included in one thesis, so only introductions of numerical methods which are used during the process will be describe in detail here.
The first part of the thesis is about how to determine the coefficients of the model by minimising a cost function. The least square method is used to make up the main part of the cost function and the Lasso regularization term is added in order to avoid the overfitting. Due to the huge size of matrices during the iteration process, Cannon2 adopts low memory BFGS to find the minimum of this cost function.
The second part is about how to output the attributes of stars by their spectra data with this model. Minimising a cost function is still the core idea and a variant the Gauss-Newton method, the Levenberg–Marquardt algorithm, is applied to realize the minimising process. Afterwards, some figures will be given to visualize these output solutions. (Less)
Please use this url to cite or link to this publication:
http://lup.lub.lu.se/student-papers/record/8998056
- author
- Shi, Yiyu LU
- supervisor
- organization
- course
- NUMM11 20191
- year
- 2019
- type
- H2 - Master's Degree (Two Years)
- subject
- keywords
- Cannon2, Stellar spectra, chemical abundance, the least square method, Lasso regularization, the Newton’s Method, quasi-Newton method, BFGS method, low memory BFGS method.
- publication/series
- Master's Theses in Mathematical Sciences
- report number
- LUNFNA-3030-2019
- ISSN
- 1404-6342
- other publication id
- 2019:E58
- language
- English
- id
- 8998056
- date added to LUP
- 2024-09-30 14:42:11
- date last changed
- 2024-10-22 15:02:18
@misc{8998056,
abstract = {{This thesis describes a method for concluding attributes of a star (including ages, temperatures, chemical element abundances) from its spectra. The method was developed by the 4MOST (4-metre Multi-Object Spectroscopic Telescope) consortium. How the model outputs stellar attributes by receiving spectra is too complicated to be included in one thesis, so only introductions of numerical methods which are used during the process will be describe in detail here.
The first part of the thesis is about how to determine the coefficients of the model by minimising a cost function. The least square method is used to make up the main part of the cost function and the Lasso regularization term is added in order to avoid the overfitting. Due to the huge size of matrices during the iteration process, Cannon2 adopts low memory BFGS to find the minimum of this cost function.
The second part is about how to output the attributes of stars by their spectra data with this model. Minimising a cost function is still the core idea and a variant the Gauss-Newton method, the Levenberg–Marquardt algorithm, is applied to realize the minimising process. Afterwards, some figures will be given to visualize these output solutions.}},
author = {{Shi, Yiyu}},
issn = {{1404-6342}},
language = {{eng}},
note = {{Student Paper}},
series = {{Master's Theses in Mathematical Sciences}},
title = {{An Introduction of A Data-driven Model for Analyzing Stellar Spectra}},
year = {{2019}},
}