Projects completed in Spring 2018 Semester

The next generation of large spectroscopic surveys of the Milky Way (WEAVE, 4MOST, DESI to quote a few) which will deliver millions of spectra that need to be analyzed with methods developped in the framework of big data tools. 

This project aims at developing and testing a machine learning code capable to deliver the stellar atmospheric parameters (effective temperature, gravity, metallicity) out of high resolution spectra.

The upcoming large sky surveys will deliver the deepest views on the cosmic web via the analysis of photometric and spectroscopic redshifts.

This project aims at testing and improving different algorithms that should reconstruct the large structures of the Universe, in particular its filaments, from incomplete datasets as robustly as possible. The tests will be conducted on existing observational sample as well as on simulations of the ESA mission Euclid.




We enter an era of extragalactic research which benefit from an unprecedented wealth of  multi-wavelength sky surveys. Even of good quality, the images that are gathered will never have a spatial resolution of comparable quality as the Hubble Space Telescope ones. However, these are mandatory to address some of the most crucial questions related to galaxy evolution.

Hence,  one solution is to apply deconvolution techniques to these ground-based images. This allows us to reach an HST-like spatial resolution.  

The Laboratory of Astrophysics has worked on an efficient code, FireDec, which has been tested already on ground-based images of distant galaxy clusters. The image quality is improved by a factor of 10.

The goal of this master project is to improve further the treatment of the noise in the images, make it run efficiently on sets of large images, and to make the code “user-friendly »

During the last two decades, our understanding of the physical mechanisms driving globular clusters (GCs) formation and early evolution has been seriously challenged by the discovery of multiple populations (MPs) with spreads in He and other light elements(e.g. Na, O, Al). This phenomenon appears to be ubiquitous, as it is observed in all GCs in the Galaxy, in nearby galaxies, and in massive early type galaxies.

In the most popular MP formation models, second population (SP) stars are formed from a combination of the ejecta of evolved massive stars from a first population (FP) and from unprocessed material.  Crucial constraints on the nature of the stellar polluters can be obtained from a detailed investigation of the MP chemistry.

For this project we propose to study the “template” GCs NGC 2808 to characterise its MP chemical pattern. State-of-the-art techniques will be employed to derive accurate abundance measurements from high-resolution spectra for all the elements relevant for MPs, like light-, alpha-, and heavier elements. Different NLTE corrections will be applied to  to assess the impact of NLTE on the measured abundance spreads.

Strong gravitational lensing offers the most powerful way to constrain the mass distribution of massive structure in the universe (galaxies and cluster of galaxies). By combining information collected with the Hubble Space Telescope and spectroscopy using for example the MUSE instrument on the ESO-VLT, we will explore fast reconstruction mass models(based on GPU infrastructure) that can probe the mass reconstruction of galaxies and/or galaxy clusters. We will apply these techniques to data challenges and real observations.



Strong lensing happens when the path of the light coming from a background galaxy or quasar is distorted by a foreground massive and compact object.

The project aims to search for strong lensing systems in the ~2 millions galaxy and quasar spectra of the BOSS and eBOSS catalogs.

Spectra are provided with PCA templates of galaxy or quasar automatically fitted by the BOSS pipeline. Strong lensing candidates are identified by looking for emission lines in the spectra not associated with the template, indicating the presence of an additional galaxy along the line of sight of the object.

The goal is to conduct a systematic search for such emission lines in both galaxy and quasar samples and will use machine learning to obtain an automated classification of the lens candidates.





One of the main cosmological probes is galaxy clustering (where both the Baryon Acoustic Oscillation (BAO) and the Redshift Space Distorsion (RSD) can be measured).

The on-going SDSS/eBOSS program – currently the largest redshift survey program for cosmology – uses several tracers to measure the galaxy cluster, one of which are the Emission Line Galaxies (ELGs): star-forming galaxies at redshift ~0.85. 

The proposed master project is to work on the clustering analysis for the ELGs. This requires understanding from the ELG spectra observations down to the cosmological requirements. The developed tools will be tested against the complete, well-understood BOSS data and with the current ELG data, the observations of which will be completed in June 2018.

Quasars are supermassive black holes living at the centre of galaxies. The Dark Energy Spectroscopic Instrument (DESI) will obtain 700’000 quasar spectra to map the Universe at redshift z>2.1 using a recent technique called the Lyman-alpha forest.

The detection of atypical spectra including broad absorption line and damped lyman-alpha systems will be essential for cosmological analyses.

The project is to exploit the large quasar spectra database from BOSS and eBOSS (~100’000 object using machine learning algorithm to setup an automatic finder for atypical quasars to be used for DESI.




One of the main cosmological probes is galaxy clustering (where both the Baryon Acoustic Oscillation (BAO) and the Redshift Space Distorsion (RSD) can be measured). The on-going SDSS/eBOSS program – currently the largest redshift survey program for cosmology – uses several tracers to measure the galaxy cluster, one of which is the Emission Line Galaxies (ELGs): star-forming galaxies at redshift ~0.85.

The goal of this project is to improve the sky subtraction in the pipeline estimating the redshift from the galaxy spectra. This step is critical for the ELG in SDSS/eBOSS project – and future projects as DESI and 4MOST: ELGs are star-forming galaxies whose redshift is mainly estimated thanks to the OII emission line.

The sky emission lines are numerous and close to the OII line for the ELGs at redshift ~1. An incorrect subtraction of the sky introduces spurious features which confuses the redshift measurement. Furthermore, a clean sky subtraction would enable many science projects, as the search for supernovae in the galaxy spectra.

The goal is to use wavelet filtering or other advanced signal processing techniques.



The expansion rate of the Universe, also called the Hubble constant – or H0 – is one of the most sought-after prizes in modern cosmology. The currently best model to explain our cosmological observations is called LambdaCDM, and postulates notably the existence of dark matter and dark energy as the main components of the Universe. A change in the nature of dark energy would produce a change in the predicted value of H0, that is to be compared with direct measurement.

Time-delay cosmography, or measuring the time delays between the images of a background source strongly lensed by a  foreground galaxy is one of the two most precise techniques currently used to directly measure H0.

LASTRO is a leading force in this field, with more than a decade of monitoring and time-delay measurements of strongly lensed quasars. In the scope of future large-scale surveys of the sky that will discover an unprecedented amount of strongly lensed Supernovae, the proposed project aims at implementing the physical and numerical tools into the current time-delay measurement codes in order to analyse precisely and accurately these new data.



Measuring the physical properties of galaxies accurately is key for understanding their evolution in cosmic history. One particularly interesting galaxy property is its star formation rate (SFR) which depends on various parameters like the gas density, the activity of the central supermassive black hole, and the interaction rate with neighbouring galaxies.

In this project, we aim to determine the SFR from radio observations of synchrotron emission which originates from cosmic rays, i.e. highly energetic charged particles that spiral around magnetic field lines in the interstellar medium. As cosmic rays are produced in supernova remnants, the remains of massive stars, there is a direct link to the SFR.

The goal of this project is to model the cosmic ray spectrum in detail with GALPROP ( https://galprop.stanford.edu/ code.php ) and use it as an input for a semi-analytical galaxy model. In particular, we are interested in applying such models to nearby low metallicity galaxies, like Haro 11, which seem to be local counterparts to the first galaxies in the Universe.

Related papers:

http://adsabs.harvard.edu/abs/ 2017MNRAS.468..946S

http://adsabs.harvard.edu/abs/ 2016ApJ…827..109S



Heating of the hydrogen and helium gas by the diffuse UV background revealed to be a crucial element for the evolution of galaxies, responsible in particular of the star formation quenching of the faintest galaxies, namely, the dwarf galaxies.In this master project, we want to go one step further by estimating the UV flux generated by a Milky Way like galaxy and its possible impact on a nearby dwarf galaxies. Among different questions, we would like to understand in which conditions the Milky Way UV flux dominates over the diffuse UV background. In a first step, relying on the pyrates and pNbody library, the student will have to build a simple Milky Way model including a realistic stellar population to estimate the UV flux, taking into account absorption processes. In a second step, he/she will build a self-consistent dynamical model based on GEAR, where the UV fluxes are computed on the fly, during the the simulation. In addition to the predictions of the hydrogen and helium heating, this work will also shed light on the molecular hydrogen content and its evolution with time.

Related papers:




In this work, we would like to explore a new star formation receipt allowing the realisation of ultra-high resolution simulations of galaxies, overpassing current numerical limitations. This new scheme  which is already implemented in our code GEAR relies on the grouping of particles tracing gas to ensure a complete initial stellar mass function. During his/her master, the student will have to test it through different kind of simulations that will help to constrain some free parameters. The final goal will be to apply this new method in a cosmological context to compute the formation of the first stars giving birth to the first galaxies. An important piece of the master will be to explore the impact of those first stars on the chemical evolution of the galaxies, owing to their metals ejection.

Related papers:




Dwarf galaxies are the least luminous galactic systems with the largest dark to stellar mass ratio in the Universe. Due to they small mass, they are very sensitive to any perturbation. As such they represent a unique laboratories for testing different key ingredients to understand the formation and evolution of galaxies. Based on new high resolution numerical simulations of dwarf galaxies formed in a cosmological context, we would like to review our understanding of they interaction with a Milky Way-like galaxy, in particular the so-called tidal stripping effect. During the master thesis, the student will have to run a bunch of simulations where the dwarf galaxy orbit around a disky potential representing the host galaxy. He/her will have to understand for which mass and which kind of orbits the dwarf sees its star formation being quenched and when it is completely stripped. Understanding this process is a key ingredient to reconcile the formation of dwarf galaxies with the standard cosmological model.

Related papers:




Gravitationally lensed quasars have wide applications ranging from 1) measuring the expansion rate of the Universe, 2) measuring the stellar content of the galaxies, 3) probing dark matter structures/sub-structures in galaxies, 4) constraining spatial structure of line emitting region of quasars, and 5) diagnosis of the inner structure of quasars via micro-lensing. The detection of lensed quasar also provides us a chance to study distant objects, which cannot be achieved without gravitationally lensing effect. The proposed project aims at detecting more lensed quasars with public database from Gaia (space) and HSC (ground-based) surveys.