Large Astronomical Data sets - Analysis and tools
Star/Galaxy Separation
In an image of the sky, what makes a single star look different from a galaxy, a system of millions (sometimes trillions) of stars? To the human eye, and to many telescopes, they both look like single points of light in the night sky.
This seemingly very simple question hides the much more complicated issue of allocating a size and a scale to objects observed in the sky, which has concerned observers and theorists throughout the twentieth century. Perhaps the most dramatic illustration of this long-standing issue is Heber Curtis and Harlow Shapley so called great debate in the 1920s, which solved the question of the size of our Galaxy in relation to cosmic scales; whereas Shapley was arguing in favor of the milky way embracing the entirety of the universe and spiral nebulae being part of it, Curtis saw our galaxy as one object among many other island universes.
One common denominator of the wide variety of observational probes constraining Dark Energy is the necessity to select pure samples of galaxies. More specifically, all the surveys must differentiate galaxies at cosmological distances from local objects, to obtain pure, or at least well-understood, samples. In the area of ''precision cosmology'', any source of systematic error is likely to play a decisive role and needs to be taken into account in order to refine the standard inflationary Big Bang picture.
During my PhD, I have been leading the star/galaxy separation task force in the Dark Energy Survey. In particular, I have designed a new tool for star galaxy separation, proved its performance on simulations (Soumagnac et. al 2014) and tested it on the DES data (publication in preparation).
Fast accessing and cross-matching large astronomical data sets
Fast access to large catalogs is required for some astronomical applications. In Soumagnac & Ofek 2018, we introduce the catsHTM tool, consisting of several large catalogs reformatted into HDF5-based file format, which can be downloaded and used locally. To allow fast access, the catalogs are partitioned into hierarchical triangular meshes and stored in HDF5 files. Several tools are provided to perform efficient cone searches at resolutions spanning from a few arc-seconds to degrees, within a few milliseconds time. The first released version includes the following catalogs (by alphabetical order): 2MASS, 2MASS extended sources, AKARI, APASS, Cosmos, DECaLS/DR5, FIRST, GAIA/DR1, GAIA/DR2, GALEX/DR6Plus7, HSC/v2, IPHAS/DR2, NED redshifts, NVSS, Pan-STARRS1/DR1, PTF photometric catalog, ROSAT faint source, SDSS sources, SDSS/DR14 spectroscopy, SkyMapper, Spitzer/SAGE, Spitzer/IRAC galactic center, UCAC4, UKIDSS/DR10, VST/ATLAS/DR3, VST/KiDS/DR3, WISE and XMM. We provide Python code that allows to perform cone searches, as well as MATLAB code for performing cone searches, catalog cross-matching, general searches, as well as load and create these catalogs.
Constraining cosmological models with BAOs
The Universe is about 13.7 billion years old. Between two very particular ages of the very young universe (when it was only a few hundreds of thousand years old) called inflation and recombination, the Universe is filled with an ionized plasma, hot and dense, in which photons and baryons are strongly bound, "coupled". Under the effect of scattering between the photons and the charged particles of the plasma, the photons are ''trapped'' in the plasma. During this time, the interplay between the plasma pressure and the radiation pressure results in ''sound waves'': spherical perturbations of the density (and pressure) propagating rapidely around each initial overdensity, similar to the waves that form on the surface of a lake when we throw a stone in it.
As the Universe expands, the baryonic matter cools down, eventually allowing the nuclei and electrons to bind into stable, neutral atoms at a time called recombination (approximately 370000 years after the Big Bang). The photons suddenly liberated from the matter. The Universe becomes ''transparent'' and the baryonic shells propagating in the form of sound waves freeze, leaving an imprint in the distribution of matter.
It is quite an amazing fact that this signature, known as Baryon Acoustic Oscillations (BAOs), of a phenomenon that happened more the 13 billion years ago, is still visible in the sky and measurable. In particular, astronomers are able to recognise this signature in the large-scale distribution of galaxies. The distance traveled by the over-density shell, provides a feature of known physical size and as such, it acts as a standard ruler. Standard rullers are very important to astronomers, because they give us a way to track the expansion of the universe, and to learn more about the mysterious energy that drives this expansion, called Dark Energy
In Soumagnac, Barkana, Sabiu et al. 2016 and Soumagnac, Sabiu, Barkana et al. 2018, we used the BAO probe to constrain the ratio of total mass versus luminous matter in the BOSS DR10 and DR12 survey.