253 Inferring model of the Galaxy with compact binaries observed by LISA

Job title:

253 Inferring model of the Galaxy with compact binaries observed by LISA

Company:

Centre National d’Etudes Spatiales

Job description

25-253 Inferring model of the Galaxy with compact binaries observed by LISAPostuler25-253 Inferring model of the Galaxy with compact binaries observed by LISA

• Doctorat, 36 mois
• Temps plein
• Indifférent
• Maitrise, IEP, IUP, Bac+4
• Fundamental Physics

PostulerMissionLaser Space Interferometer Antenna (LISA) will observe our Galaxy — Milky Way in the gravitational wave spectrum. It will be possible due to millions of compact galactic binaries which will be accessible to the detector within its sensitivity. Galactic binaries mostly comprise of double white dwarf binaries which will be emitting quasi-monochromatic gravitational waves in a very narrow frequency band.The objective of this PhD project is to infer the model of the Galaxy informed by the population of galactic binaries from LISA observation.Let us look at the steps and approaches we are going to take to reach this objective.The product of the global data analysis for LISA will consist of posterior distributions of physical parameters which describe binary systems, such as their masses, orbital periods, location (distance and sky position) and orientation of the orbit relative to the observer.The purpose of the global analysis (or global fit as we usually say) is to take all the signals mixed in the LISA data stream and simultaneously estimate their parameters. The simultaneous fit will allow to avoid biases in the values of parameters which can be caused by overlapping signals. The main portion of these signals will be coming from galactic binaries, some of them will be resolved (with the current estimate of the order of tens of thousands of resolvable signals). Other millions of signals will constitute the stochastic background.Population synthesis models provide us with a catalogue of galactic binaries with associated parameters: masses, orbital periods, sky positions and distances. To create such models researches usually rely on assumptions for star formation rate, milky Way potential and synthetic population of galactic binaries. At the moment there exist a couple of detached models which describe the population of galactic binaries in the Galaxy.In this PhD project we will focus on inferring the population of galactic binaries based on the observations as a final result of the inference. The most consistent way to infer the population will be to incorporate different models into the process of the global fit. This approach is sometimes called hierarchical inference, when in addition to the inference of signal parameters, we infer the population from which the sample of sources originate.One of the ways to include the information on source populations is to feed it to the bayesian inference algorithm through priors or proposals while sampling. To set up an MCMC algorithm for parameter estimation we need a prior on all parameters. At the same time, we need to define proposals for the Markov Chain. There can be different approaches explored here. One is when we take existing population models, fit density estimation model to them, for example, such as neural density estimators which are very flexible and can easily model an arbitrary complicated distribution. As the result of the analysis we will be able to say which model is most probable by looking at the rate of proposals that were accepted for each model.

The next step will be to go further and construct a continuous parameterised model of the galaxy population which will contain all discreet models and also can interpolate between them. Here we will be discovering new unknown models which later will have to be interpreted to explain the physics underlying the observation. In this case we will be doing hierarchical inference of the parameters that define the overarching phenomenological model.Another question that will be addressed in this PhD project is how to incorporate the information that we get from galactic binary background. We can start by interpreting it as the density of the sources distributed on the sky. This can provide the information on the volume of signals for each direction and can give additional useful information for astronomers.The methodological part of this PhD will require the implementation of techniques that originate from the following approaches. One is the artificial intelligence methods for density estimation.We can implement the neural density estimation with the normalising flows, for example, which is a deep learning technique that has well demonstrated its performance for such tasks. The other other methodological task will focus on accelerated MCMC sampling techniques. Moreover we will explore the combination of these techniques for better inference.We will tackle this problem by creating simulated datasets for LISA data which will represent a sample of galactic binaries from different populations in realistic LISA noise.For more Information about the topics and the co-financial partner (found by the lab !); contact Directeur de thèse –Then, prepare a resume, a recent transcript and a reference letter from your M2 supervisor/ engineering school director and you will be ready to apply online before March 14th, 2025 Midnight Paris time !

Expected salary

Location

Nice, Alpes-Maritimes

Job date

Wed, 05 Feb 2025 05:47:54 GMT

To help us track our recruitment effort, please indicate in your email/cover letter where (vacanciesineu.com) you saw this job posting.