Resumen de Pronòstic de mapejat d'intensitat lyman-alpha amb PAU survey
In the upcoming years, the amount of observational data available for astro- physics as a whole, and particularly cosmology, is expected to increase by orders of magnitude due to the new observational programs and technological advance- ments. To fully capitalise on these upcoming observations, new statistical methods to draw scientific results have to be developed and applied. Intensity Mapping is one of such new methods, which consists on the tracing of a sharp spectral feature (e.g., a emission line) in angular coordinates and redshift, without considering re- solved objects.
This thesis is focused on the simulation and evaluation of Intensity Mapping with the Lyα line by cross-correlating two kinds of optical surveys: the narrow- band photometric survey PAUS, and the spectroscopic surveys eBOSS and DESI.
The former should contain extended Lyα emission from the intergalactic medium in the background of the images, while the latter contains the Lyα forest (i.e., the set of absorption lines generated by the intergalactic medium) in their quasar spectra.
Cross-correlating both then should yield a signal of the diffuse Lyα emission, despite the foregrounds and noise in PAUS images.
In order to simulate this, a hydrodynamic simulation specifically designed for the study of Lyα in the intergalactic medium has been used to simulate both Lyα emission and absorption, while the foregrounds in PAUS images have been mod- eled using a deep lightcone mock catalogue and fitting spectral energy distribu- tions to its objects. Instrumental/atmospheric noise has been added to the simu- lated PAUS images by directly measuring the noise from sets of masked and stacked images, and introducing it on the simulation as a Gaussian distribution. Besides, a cross-correlation code in Python has been developed from scratch, optimised and validated in the framework of the thesis to compute the cross-correlations.
The results are presented in a probabilistic manner: for different cases (PAUS- eBOSS, PAUS-DESI and two hypothetical PAUS extensions with DESI) 1,000 cross- correlations are computed with different realisations of the instrumental/atmospheric noise of PAUS images, as well as quasar positions. With the optimistic approxima- tion of uncorrelated instrumental noise (which would require further work in data reduction), even the largest PAUS extension considered only yields a probability of detection of ∼ 15%; using the actual correlated noise all the probabilities become negligible.
In spite of these negative results, some valuable conclusions are extracted. The different kinds of two-point correlation functions (monopole, parallel and perpen- dicular to the line of sight) show complementary behaviours: monopole and par- allel perform better at larger scales, while perpendicular samples better the scales smaller than 10 Mpc/h. Besides, the redshift smoothing in the Lyα signal due to the photometric filters has been properly implemented in the theoretical model, and can even be used to predict the scales with better SNR for each correlation function.
Therefore, these findings indicate that this methodology may be more suitable for broad-band surveys; we conclude the thesis giving some basic guidelines on how this study could be replicated for broad-band data.
