Título:New approaches to galaxy clustering

Descrição:The large-scale clustering of galaxies contains a wealth of information on the geometry and expansion history of the universe, on gravity, and on the initial conditions. In order to extract this information, we need to deal with the complex formation process of galaxies. As a result, current observational constraints are largely based only on the robust BAO feature. However, thanks to significant advances in o ur theoretical understanding of galaxy clustering, we now have a well-defined approach for absorbing all the complicated, incompletely understood physics of galaxy formation into a set of free parameters (the bias parameters and stochastic amplitudes), which opens up considerable additional constraining power. The next challenge to tackle then is how best to connect this theory with data, as delivered by ongoing and future surveys such as BOSS, DESI, and Euclid. In my talk, I will review these developments, and discuss the prospects galaxy clustering as a robust probe of cosmology in the coming decade.

Palestrante:Fabian Schmidt

Instituto:Max Planck Institute - Garching

Data:

Hora:

Local:Página do LIneA - http://www.linea.gov.br/seminarios/

Título:Identifying RR Lyrae in the Dark Energy Survey

Descrição:RR Lyrae variable stars (RRL) are powerful distance indicators which can be used to locate stellar structures within the Milky Way halo. The Dark Energy Survey (DES), with its large observing footprint and depth, presents a rich opportunity to discover these variables to distances near the edge of the Galactic halo. However, the small number of observations in the DES wide-field survey data (fewer than 20 multiband observations over the first three years) pose a challenge for the traditional RRL detection techniques. In this talk, I will describe the multiband template fitting method and machine learning classifier we developed to identify 5783 RRL candidates in the first three years of DES data. These methods will enable RRL detection in other sparse multiband data sets that are not optimized for time series analyses.

Palestrante:Katelyn Stringer

Instituto:Texas A&M University

Data:2019-05-16

Hora:

Local:Página do LIneA - http://www.linea.gov.br/seminarios/

Título:Bayesian Deep Networks for parameter derivation: from Strong Lensing to the borehole image logs.

Descrição:Deep Learning algorithms have been proposed to several complex computer vision tasks. In several cases it has reached or exceed human level performance. In this contribution we address the issue on how to use those methods for regression tasks and how to derive consistent uncertainties by applying a Bayesian framework in the convolutional Deep Neural Networks. We present a science case based on the modelling strong lensing features such as Einstein radius, velocity dispersion and lens redshift. We were able to retrieve those parameter with fractional deviation up to 15% in Dark Energy Survey quality level simulated images. We were able to derive a PDF in a couple of seconds outperforming the traditional MCMC methods in computational time. We also present a second case applied to geosciences in which we derived permeability and porosity from image ultrasound and resistivity image logs.

Palestrante:Clécio R. De Bom

Instituto:CEFET-RJ

Data:2019-04-24

Hora:14:00

Local:Auditório do prédio Emmanuel Liais

Título:Is morphology telling us the truth about quenching?

Descrição:Passive and star forming galaxies have different morphology and structure, suggesting a link between the star formation activity and the formation/growth of internal components. Despite the numerous efforts, the details of how galaxies form and evolve, and in particular the mechanisms responsible of the decreasing/cessation of the star formation and the morphological transformations are still not fully understood. In the talk I will present results we obtained combining the morphology and the star formation activities for galaxies in the CANDELS survey. We first studied structural properties of the four main morphological classes (bulge dominated systems, pure spiral disk dominated, intermediate 2-component and irregular disturbed galaxies) and how their fractions change at different epochs (Huertas-Company et al,2016) to then move forward and analyze in more details the galaxy structure exploring bulge and disk properties and how they correlate with the morphology or the star formation activities of the host galaxies (Dimauro et al, 2018, 2019).

Palestrante:Paola Dimauro

Instituto:Observatoire de Paris

Data:2019-04-18

Hora:11:00

Local:Página do LIneA - http://www.linea.gov.br/
Seminários - http://www.linea.gov.br/seminarios/