Galaxies are factories in which gas gravitationally collapse to form stars. This process is very inefficient: the gas-to-star conversion timescale is ~200 times smaller than the free-fall time. The origin of this inefficiency is one of the main open questions in the field of galaxy evolution. In particular, a novel scenario competes with the usual picture of regulation of star formation by feedback, which is a transition between: - z~0 : regulation of star formation by small scales (stellar feedback) - z~2 (10 Gyrs back) : regulation of star formation by large scales (disk dynamics and accretion)_x0003_ If this transition exists, it should leave chemo-dynamical footprints on stars observed today: the processes dominating this regulation impose different star formation law scalings with the galaxy environment and more than half of the stars in the local Universe formed during the peak of cosmic star formation, at 1

One of the major questions of galaxy evolution is: why are galaxies so inefficient at forming stars? Indeed, we need to find a process opposing gravity, slowing down the conversion of gas to stars by a factor 100 compared to the free-fall time of the gas to understand how star formation is regulated. All processes considered so far were shown to provide insufficient support against gravity. A missing process is supersonic turbulence, injected at galactic scales (10s of kpc) and cascading down to the scale of prestellar cores (~0.1 pc). Turbulence may provide the required support, and also modify the impact of other processes, such as star formation feedback. Understanding turbulence is thus primordial, but the huge cascade dynamic range involved was intractable for numerical simulations. To undo this lock, I developed an encapsulated-zoom method for numerical simulations.We are now able to simulate, for the first time self-consistently, the self-generated turbulence cascade over four decades in spatial scales, finally bridging the gap between galactic scales and prestellar cores. To understand the role of galactic-scale turbulence, we will characterise the turbulence injection by each galaxy scale energy source (e.g disk instabilities) and determine turbulence’s interplay with the energy release from star formation feedback. We will account for the galactic environment and test our new understanding on Gyrs timescales. We will produce observable predictions of the footprint of star formation regulation by galactic scale turbulence to compare our results to the newest surveys. Furthermore, we will create models of the interaction between turbulence and stellar feedback, port our methods to emerging GPU codes, and produce predictions for the upcoming SKA and ELT. Thanks to this methodology, we are now able to determine whether or not the self-consistent treatment of galactic-scale turbulence is sufficient to regulate star formation in galaxies.

Questions like “How do galaxies grow?” and “How do they die?” are not only fundamental but capture many of the most important problems in galaxy formation. We seek comprehen- sive answers that address the nature of physical mechanisms that drive initial formation, mass assembly and size growth, and the eventual “death” of star formation via quenching. These mechanisms play out spectacularly in massive galaxies. Those centrally peaked and luminous “stellar cathedrals,” seated in the heart of enormous dark matter halos, provide unique insight into these fundamental questions. The cathedrals program will address these questions by combining observations from several exciting new facilities, including the prominent international MaNGA Survey (Mapping Nearby Galaxies at APO), an integral field spectroscopic survey of 10,000 nearby galaxies now entering its second year and of which I am the Principal Investigator. Motivating new modeling techniques, the late stages of evolution (z < 1) will be charted with unprecedented statistical precision thanks to the emerging era of wide-field surveys like MaNGA and Hyper Suprime Cam at Subaru Observatory, a precursor of EUCLID. We will study the origin of present-day trends with targeted followup of the most distant galaxies from more powerful facilities, such as MUSE on the VLT and eventually JWST and E-ELT. With a 4-year appointment at Le Centre de Recherche Astrophysique de Lyon (CRAL), I will bring my expertise in wide-field survey design and “big data” science to the French community, which is investing heavily in these areas (e.g., EUCLID and 4MOST) but has only limited access to current programs of this nature. The MaNGA survey, specifically, not only offers valuable lessons in fiber technologies and survey execution, but immediate scientific synergies with MUSE GTO science being led at CRAL. The combination of MaNGA and MUSE enables exciting exploitations of JWST and will guide development of E-ELT HARMONI, both future instruments that CRAL is playing a role in defining. Finally, by combining scientific drivers, survey experience, and the instrumentation expertise at CRAL, new project ideas on future facilities will be developed.

In the era of precision cosmology, we still need to understand how intergalactic gas flows into the potential wells of galaxy haloes, how it transforms into stars, and how stars eject gas, metals and energy feedback into the interstellar and/or intergalactic medium (IGM). Ideally, these processes have to be followed from the first galaxies and the reionization phase between redshifts 15 and 7, through the early mass assembly of building blocks, when the universe was typically 1-2 Gyr old, down to the formation of the Hubble sequence of morphological types seen in the local universe. MUSE, a 2nd generation instrument for the VLT, will probe the deep universe like never before. Thanks to its sensitivity and multiplex power, this revolutionary giant 3D spectrograph will allow us to constrain and test scenarios of galaxy formation with unprecedented precision. MUSE will start its operations in 2013, assisted by adaptive optics from mid-2015. MUSE will obtain 90,000 spectra per exposure at an average resolution R=3000 on the full visible wavelength range 4800—9300 Å in a 1 arcmin2 field-of-view, with an excellent overall throughput and stability that will facilitate very long integrations. Because no similar instrument currently exists in the world with such multiplex capabilities and sensitivity at this level of spatial and spectral resolution, we intend to make major progress in the knowledge of galaxy formation processes along the following axes: the characterization of the end of reionization (6