This thesis debates the removal of artifacts arising in radio interferometric observations in the context of particle dark matter (DM) signals in the radio spectrum. After introducing the problem, the search for indirect, non-gravitational DM signals by cross-correlating EM signals with gravitational tracers and a description of the statistical methods used will be provided. Of particular interest in that regard is the 2-point correlation function. Interferometric basics and image synthesis that lead to said artifacts will be explained. Two different approaches will be reviewed; firstly the full reconstruction of radio interferometric maps using UNets and secondly the direct calculation of the two-point correlation function from maps using a Convolutional Neural Network (CNN). It is found that the first approach fails reconstructing the 2-point correlation function whereas the second approach shows to be more promising in that regard. The thesis will close with discussing possible further developments and the applicability to actual radio data.
Ricerca di segnali di materia oscura in immagini radio interferometriche attraverso Reti Neurali Convoluzionali
THOMAS, LEO
2018/2019
Abstract
This thesis debates the removal of artifacts arising in radio interferometric observations in the context of particle dark matter (DM) signals in the radio spectrum. After introducing the problem, the search for indirect, non-gravitational DM signals by cross-correlating EM signals with gravitational tracers and a description of the statistical methods used will be provided. Of particular interest in that regard is the 2-point correlation function. Interferometric basics and image synthesis that lead to said artifacts will be explained. Two different approaches will be reviewed; firstly the full reconstruction of radio interferometric maps using UNets and secondly the direct calculation of the two-point correlation function from maps using a Convolutional Neural Network (CNN). It is found that the first approach fails reconstructing the 2-point correlation function whereas the second approach shows to be more promising in that regard. The thesis will close with discussing possible further developments and the applicability to actual radio data.File | Dimensione | Formato | |
---|---|---|---|
823428_thesis-leo-thomas.pdf
non disponibili
Tipologia:
Altro materiale allegato
Dimensione
17.56 MB
Formato
Adobe PDF
|
17.56 MB | Adobe PDF |
I documenti in UNITESI sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/20.500.14240/146168