SEARCH FOR PRIMORDIAL GRAVITATIONAL WAVES WITH QUBIC: DEVELOPMENT OF MACHINE-LEARNING TECHNIQUES

The quest for B-mode polarization of the Cosmic Microwave Background is among the main challenges in Observational Cosmology. Measurement of B-mode polarization in the CMB will be clear evidence of the presence of primordial gravitational waves which are theoretically expected to be produced during inflation about 10-35 seconds after the Planck epoch. The B-mode measurement is perhaps the most difficult cosmological challenge because the expected signal is very small.  It requires high sensitivity and negligible instrument systematic effects with wide frequency coverage in order to separate the primordial signal from foreground emissions.

QUBIC (QU Bolometric Interferometer for Cosmology: http://qubic.in2p3.fr) is a novel instrument concept dedicated to the search for B-modes by measuring the Q and U polarization modes.  It brings together the advantages of bolometers with high sensitivity and those of interferometers that have exquisite control of instrument systematic effects. The interferometric nature of QUBIC also allows spectro-imaging and improved spectral resolution with respect to imagers, providing a significant advantage concerning foreground removal. The Technological Demonstrator is under test at APC since 2019 and will be installed at the QUBIC site at 5000m a.s.l. in the province of Salta in Argentina in 2021.

Besides participating in data taking and exploitation of QUBIC (including travel to Argentina), the Ph.D. student will work specifically on developing data analysis techniques based on artificial intelligence and Machine-Learning for the QUBIC data analysis pipeline:

  • The separation between astrophysical and atmospheric signals

  • Map-making

  • Astrophysical component separation

  • Polarized angular power spectrum measurement

The student will work within the QUBIC team at APC, collaborating with the rest of the collaboration in France, Italy, Ireland, and Argentina. He/She may also contribute to other scientific activities in parallel to his/her specific topic.  These might include the development of the QUBIC Data Analysis pipeline from time-domain to cosmological constraints, spectro-imaging and astrophysical foreground contamination control, self-calibration and instrumental systematic effects control, machine-learning techniques applied to CMB data analysis.

Responsable: 

Jean-Christophe Hamilton, Steve Torchinsky

Services/Groupes: 

Année: 

2021

Formations: 

Thèse

Niveau demandé: 

M2

File upload: 

Email du responsable: