Please use this identifier to cite or link to this item:

http://hdl.handle.net/10609/107807
Title: Distant galaxies analysis with deep neural networks
Author: Cacho Martínez, Raúl
Director: Bosch Rue, Anna
Tutor: Solé Ribalta, Albert
Keywords: galaxies
stellar decomposition
inversion problem
deep neuronal network
deep learning
stellar populations
Issue Date: 8-Jan-2020
Abstract: In this work we face a very common problem in Astrophysics. One of the first parameters to obtain from a galaxy spectrum is the redshift. The redshift at which a galaxy is, can tell us a lot of things about the large scale structure of the universe. However, the telescope time is limited, and it would take a lot of time to survey the whole sky observing the spectrum of galaxies. This is the reason why surveys using narrowband photometry (for example ALHAMBRA or JPAS) are arising. These surveys allow to observe a large number of galaxies in much less time than using spectroscopy, thus making astronomers able to disentangle the structure of the Universe and the features of very distant galaxies. Traditionally, the features have been derived using the technique known as SED-fitting, which consists in deriving the features of the galaxy from its spectrum. This is not an easy problem, not only because of the large number of variables in play (velocity, velocity dispersion, age and metallicity for each single stellar population, or SSP), but because of the degeneracies. A degeneracy happens when two different SSPs show almost undistinguishable spectra. For example, a degeneracy exists between age and metallicity, with and old and metal-rich1 SSPs showing similar spectrum to that of a young and metal-poor SSPs. In this Master Thesis we evaluate the ability of Deep Neural Networks, using as input the observations of a galaxy, to obtain the parameters of the galaxy (redshift, mass and galaxy type).
Language: English
URI: http://hdl.handle.net/10609/107807
Appears in Collections:Research papers

Share:
Export:
Files in This Item:
File Description SizeFormat 
raulrcmTFM0120presentation.pdfPresentation of TFM6.01 MBAdobe PDFView/Open
raulrcmTFM0120memory.pdfMemory of TFM1.18 MBAdobe PDFView/Open

This item is licensed under a Creative Commons License Creative Commons