I am a Beecroft fellow at Oxford University, working on astrostatistics and, more specifically, field-level analysis of weak lensing data. My work focuses on understanding the large-scale structures of the Universe. For that, I develop statistical and data analysis tools to optimally extract the information from the observations.

â€‹

Before Oxford, I was a postdoc at Imperial College, working at the Imperial Centre for cosmology inference. Formerly, I was a PhD student at the Max-Planck Institute for Astrophysics (MPA) and the Excellence Cluster Universe.

## Research

### Bayesian forward modelling of cosmic shear data

### Large-scale structure inference from the Lyman-Î±Â forest

The next-generation of surveys will provide unprecedented precision in cosmic shear measurements, posing a challenge on how to extract as much information from the data as possible. Harvesting the full information content of the data will require a field-based approach, in which every data point is used, rather than relying on summary statistics that do not capture all the information and whose distributions are not well known.

I developed a Bayesian hierarchical model to infer the cosmic matter density field, and the lensing and the matter power spectra, from cosmic shear data. This approach employs a physical description of gravity, improving upon previous methods that assumed Gaussian distributions of the shear field, and allowing us to sample from the initial conditions. With this more complex data model, we get a better representation of the data, and we extract information beyond the two-point statistics, exploiting the full information content of the shear fields.

Most of the analyses of the cosmic large-scale structure are based on galaxy observations. However, galaxies mostly trace the overdense regions of the Universe. A complementary probe of the large-scale structure is the Lyman-α forest, which sensitive to the low-density regimes. Also, by probing small scales down to 1 Mpc, the Lyman-α forest is sensitive to neutrino masses and dark matter models.

â€‹

A usual approach to extract information from the Ly-α forest is to analyse the power spectrum. However, the non-linear dynamics transport the information from the two-point statistics to high-order correlations.

â€‹

I developed a Bayesian framework to infer the 3D dark matter density and its dynamics from the Lyman-α forest. This method employs a physical forward model that encodes the high-order correlations of the dark matter density.

### Robust data models for contaminated datasets

Next-generation of galaxy surveys will not be limited by statistical noise but by systematic uncertainties. In the past, such effects have been addressed by generating templates for such contaminations and accounting for their effect in a Bayesian framework.

â€‹

Since the future surveys can be subject to yet unknown contaminations, I developed a robust likelihood to effectively deal with effects induced by unknown foreground and target contaminations. This likelihood recovers an unbiased power spectrum at all scales, even from data subject to contaminations.

â€‹

This likelihood has been applied to the analysis of SDSS-III data in Lavaux, Jasche & Leclerqc (2019), recovering the unbiased power spectrum even at large scales (k<0.01) where the state-of-the-art analyses show deviations due to contaminations in the data.

### Formation and evolution of AGN

Active galactic nuclei (AGN) are black holes at the centre of galaxies that accrete material from the galaxy’s central region. As the material falls in towards the black hole, it spirals in and forms into a disk. This accretion disk heats up due to the gravitational and frictional forces and emits strong radiation in X-ray, gamma-rays, and radio. AGN are one of the most powerful sources of the Universe.

I investigated how AGN formation and evolution depends on their environment. I characterized the dark matter environment with a dark matter density field inferred from a galaxy survey. This analysis provided evidence of an evolutionary sequence between two types of AGN.

"Field-level inference of cosmic shear with intrinsic alignments and baryons"

NP, Heavens, Mortlock, Lavaux, Makinen (Submitted to MNRAS, 2023)

â€‹

"LyAl-Net: A high-efficiency Lyman-α forest simulation with a neural network"

Boonkongkird, Lavaux, Peirani, Dubois, NP, Tsaprazi (Submitted to A&A, 2023)

â€‹

"The Cosmic Graph: Optimal Information Extraction from Large-Scale Structure using Catalogues".

Makinen, Charnock, Lemos, NP, Heavens, Wandelt (OJA, 2022).

â€‹

"Lifting weak lensing degeneracies with a field-based likelihood"

NP, Heavens, Mortlock, Lavaux (MNRAS, 2022).

"Bayesian forward modelling of cosmic shear data"

NP, Heavens, Mortlock, Lavaux (MNRAS, 2021)

NP, Hahn, Jasche, Lavaux (A&A, 2020)

"Inferring high redshift large-scale structure dynamics from the Lyman-alpha forest"

NP, Jasche, Lavaux, Ensslin (A&A, 2019).

â€‹

"Explicit Bayesian treatment of unknown foreground contaminations in galaxy surveys"

NP, Kodi Ramanah, Jasche, Lavaux (A&A, 2019).

â€‹

"Imprints of the large-scale structure on AGN formation and evolution."

NP, Jasche, Ensslin, Lavaux (A&A, 2018).

â€‹

"NIFTy 3 - Numerical Information Field Theory."

Steininger, Dixit, Frank, Greiner, Hutschenreuter, Knollmuller, Leike, NP, Pumpe, Reinecke, Sraml, Varady, Ensslin (Annalen der Physik, 2018)

"Cosmic expansion history from SNe Ia data via information field theory"

NP, Ensslin, Greiner, Boehm, et al. (A&A, 2017)

## Aquila Consortium

I am a member of the Aquila Consortium, a group of people interested in developing and applying data science methods to the analysis of cosmological data.

â€‹

You can find more information about our science in here.

## Contact

## â€‹

Denys Wilkinson Building,

University of Oxford, Keble Rd,

Oxford OX1 3RH, UK