ORGANISING COMMITTEE
LOCAL ORGANISATION
SPONSORS
|
|
|
|
28-29 June 2023
Prague, Czech Republic
This workshop is focussed on how to accurately model the formation of large scale structure in the Universe and how to use the next-generation of cosmological surveys to increase our understanding of dark matter, dark energy and gravity. The workshop targets experts in the field of late Universe cosmology, theory and simulations, geared not only towards senior researchers but also towards postdocs and PhD students. We have scheduled 13 talks and 4 discussion sessions, and left time for collaboration and exchange of expertise.
| SPEAKER | INSTITUTE | TOPIC |
|---|---|---|
| Matthias Bartelmann | University of Heidelberg |
Kinetic field theory for cosmic structure formation
Kinetic field theory is a new analytic approach to cosmic structure formation which avoids some of the notorious problems in this field. I will briefly review the foundations of the theory and then highlight four recent results, concerning the asymptotic behaviour of power spectra, a mean-field approximation to particle interactions, perturbation theory, and structure formation in modified gravity.
|
| Elisa Chisari | University of Utrecht |
Modelling intrinsic alignments in the Stage IV era
Galaxies align towards each other in a way that is correlated with the large-scale structure of the Universe. These intrinsic alignments are a known contaminant to weak gravitational lensing and clustering cosmology. In this talk, I will give an overview of modelling strategies for intrinsic alignments in order to prevent biased estimates of the cosmological parameters. Amongst them, I will more in depth about a new modelling strategy based on the effective-field-theory description of intrinsic alignments.
|
| Pedro Ferreira | University of Oxford |
Cosmological data analysis for the 21st century
Cosmological data analysis is undergoing a revolution of sorts which echoes, to some extent, the transformation it underwent in the 1990s. Then, with the COBE discovery and the results from the APM and IRAS surveys, new approaches to model building and testing had to be developed; Bayesian analysis emerged triumphant with likelihoods, evidence and monte carlo methods. Now we are facing a deluge of data which allows us to take our understanding of the universe much further. And we are exploring new methodologies, with Bayesian forward modelling, machine learning, emulators and field level inference. I will discuss the state of play of these methods and pose some questions about where we might be going.
|
| Flaminia, Fortuni | INAF - Osservatorio Astronomico di Roma |
FORECAST: a flexible software to forward model cosmological simulations mimicking real observations
I will present FORECAST, a new flexible and adaptable software package that performs forward modeling of cosmological hydrodynamical simulation to create a wide range of realistic synthetic astronomical images.
I will extensively talk about the methods adopted in the software: FORECAST constructs light-cone and synthetic images exploiting the output snapshots of any cosmological hydrodynamical simulation and computes the observed flux of each simulated stellar element in any chosen set of pass-band filters, including k-correction, IGM absorption and dust attenuation. Then, I will show the results of its first application, emulating the GOODS-South field as observed for the CANDELS survey exploiting the IllustrisTNG simulation. I produced images of 200 sq. arcmin., in 13 bands with depths consistent with the real data. I analysed the images with the same processing pipeline adopted for real data in CANDELS and ASTRODEEP publications, and I compared the results against both the input data used to create the images, and real data, finding good agreement with both. These tests show the makings of the software and some of its possible applications with current and future galaxy surveys.
I will present the recent release of the FORECAST code and of two datasets: the CANDELS dataset and a CEERS/JWST dataset in 10 filters (8 NIRCam and 2 MIRI) in a field of view of 200 sq. arcmin. between z=0-20.
|
| Jens Jasche | Stokholm University |
Bayesian Forward Modeling of Galaxy Surveys
The standard model of cosmology offers a wealth of opportunities to probe the fundamental physics underlying the origin and evolution of cosmic structures, the accelerating cosmic expansion, and dark matter, mainly through next-generation galaxy surveys. However, traditional data analysis methods focus on limited statistical summaries, disregarding crucial information embedded within the intricate filamentary distribution of cosmic matter in three-dimensional space. Fortunately, a new paradigm has emerged—Physics-informed field-level inference—which presents a compelling alternative for studying galaxy surveys. By leveraging nonlinear structure formation models, this approach enables the joint inference of cosmic initial conditions, the mapping of nonlinear density and velocity fields, and the derivation of dynamic structure formation histories with a detailed treatment of uncertainties. In this presentation, I will illustrate this Bayesian physical forward modeling approach through diverse data applications and projects performed by the Aquila consortium. These projects include mapping nonlinear dark matter density and velocity fields, cosmological parameter inference, and tests of fundamental physics given our inference results. Furthermore, I will present a comprehensive analysis of the Nearby Universe, conducted via the SIBELIUS project, yielding one of the most extensive data-constrained simulations to date, featuring a central pair of galaxies—Milky Way and Andromeda.
|
| Elena Kozlikin | University of Heidelberg |
Kinetic Field Theory – frequently asked questions and answers
Following the introductory talk by Matthias Bartelmann of Kinetic Field Theory (KFT), I will address three major aspects concerning the approach which have been the source of frequently asked questions in the past years. In the first part of this talk, I will show that – by using the well-known Hubbard- Stratonovich transformation – we can go from the microscopic many-body picture of KFT to a macro- scopic formulation in terms of macroscopic fields (e.g. density-fields) where the underlying microscopic dynamics of the system is still preserved [2]. In this re-formulation of the theory, we will see that KFT works perfectly well without ‘Zel’dovich trajectories’ which are typically introduced in the microscopic formulation of KFT. In the second part, I intend to show in detail how KFT avoids the difficulties of (Eulerian) Standard Perturbation Theory (SPT) by construction, thus allowing us to proceed deeply into the non-linear regime of density fluctuations [1]. I will highlight the differences between approximations made in KFT and SPT which lead to differences in the information content about a system even before we set up the perturbation theory. I will show how the perturbative terms in KFT and SPT are related and argue why KFT offers a better starting point for a perturbative approach. In the last part, I would like to explain how modified gravity theories (including screening effects) can be incorporated into the KFT formalism and show some preliminary results for nDGP and Chameleon models.
[1] E. Kozlikin, R. Lilow, F. Fabis and M. Bartelmann, JCAP 06 (2021), 035, arXiv:2012.05812. [2] R. Lilow, F. Fabis, E. Kozlikin, C. Viermann and M. Bartelmann, JCAP 04 (2019), 001, arXiv:1809.06942. |
| François Lanusse | CNRS, Saclay |
Implicit and Explicit Simulation-Based Bayesian Inference for Cosmology
Physical models in the form of simulations offer an avenue to model the data in all of its complexity, but until very recently using such models to estimate physical fields and parameters remained an open problem.
In this talk, I will discuss the two possible points of view on simulators, depending on whether they are “black-box” or “open-box” models, and the different methodologies and strategies which may be applied in each case to use these physical models within a Bayesian inference context. As both of these approaches become tractable, an interesting question for our field is to discuss which point of view will be the most effective and robust in practice.
In the case of black-box simulations (which can only be sampled from), I will discuss applications of deep generative models as a practical way to manipulate these implicit distributions within a larger Bayesian framework. I will illustrate in particular on a weak lensing example how neural compression and neural density estimation achieve theoretically optimal posterior recovery.
In the case of open-box simulations, which can be seen as differentiable probabilistic models, with an explicit joint log probability, I will discuss strategies and challenges for building large scale differentiable physical models of the large scale structure touching in particular on distributed differentiable N-body solvers and building accelerated hybrid physical/ml simulations leveraging neural ODE methodologies.
|
| Andrina Nicola | University of Bonn |
On the road towards Stage-IV cosmology
We live in exciting times for observational cosmology, with several high-precision LSS surveys about to start in the next years, such as Euclid, Roman and Rubin/LSST. These surveys offer the sensitivity required to test the fundamental assumptions underlying our cosmological model.
In this talk, I will give an overview of on-going cosmological surveys and results and then present selected Stage IV surveys. In particular, I will focus on control of observational and theoretical systematics, which will be crucial to ensure robust cosmological constraints from these data.
|
| Natalia Porqueres | University of Oxford |
Lifting the weak lensing degeneracy with a field-based likelihood
With Euclid and the Rubin Observatory starting their observations in the coming years, we need highly precise and accurate data analysis techniques to optimally extract the information from weak lensing data. However, the standard approach based on fitting some summary statistics is inevitably suboptimal and imposes approximations on the statistical and physical modelling. I will present a new method to analyse weak lensing based on a full physics model and field-based statistics. By analysing the data at the pixel level, this method lifts the weak lensing degeneracy and provides uncertainties on the cosmological parameters up to a factor 5 smaller than those from standard techniques on the same data. In addition to a gravity model, the method accounts for intrinsic alignments and baryon feedback. I will discuss the current status and ways to meet the challenges of this approach for its first real data application.
|
| Fabian Schmidt | Max Planck Institute for Astrophysics (MPA) |
EFT meets field-level inference
Standard applications of the effective field theory of LSS compute n-point correlations semi-analytically. I will discuss an alternative paradigm, implementing the EFT on a lattice as a forward model which can be used both for full Bayesian inference at the field level, and for likelihood-free inference based on summary statistics. One crucial advantage is the non-perturbative treatment of the displacement from initial position to observed coordinates, with ramifications for BAO reconstruction and redshift-space distortions.
|
| Leonardo Senatore | ETH Zurich |
Recent Development on the EFTofLSS
I will present several developments in the theory of the EFTofLSS on the theory and the data sides.
|
| Hans Winther | University of Oslo |
Non-linear clustering observables in models beyond LCDM
I will give an overview over numerical simulations in models beyond LCDM: modified gravity, dark matter and dark energy models. What has been done in the past and what is currently being done. The biggest challenge with constraining such models with future surveys is that it often requires a big simulation effort. I will show how we can get around this with very little computational resources (compared to what is being done for LCDM) and allow us to produce both cheap mock galaxy catalogues and/or emulators for the non-linear matter power-spectrum that will allow us to constrain your favourite model with observations from current and future galaxy surveys.
|
| Matteo Zennaro | University of Oxford |
The bacco emulator project
I will present the bacco emulators, a suite of matter and galaxy power spectrum emulators obtained combining linear predictions, perturbation theory, state-of-the-art simulations, and artificial neural networks. These emulators include baryonic effects, neutrinos and dynamical dark energy and are well suited for the analysis of upcoming spectroscopic and photometric galaxy surveys. After introducing the emulators themselves I will show, as an application, how they can be used to re-analyse DES Y3 data including smaller scales, leading to interesting results in terms of cosmology, baryonic phyisics and S8 tension.
|
| Main Workshop | |||
| Tuesday 27th June |
Wednesday 28th June |
Thursday 29th June |
Friday 30th June |
|---|---|---|---|
| 8:30 - 8:55 | Registration | ||
| 8:55 - 9:00 | Opening | ||
| 09:00-09:30 | P. Ferreira | F. Lanusse | Free discussion and collaboration |
| 09:30-10:00 | A. Nicola | J. Jasche | |
| 10:00-10:30 | Discussion - Data | F. Fortuni | |
| 10:30-11:00 | Break | ||
| 11:00-11:30 | Free discussion and collaboration | L. Senatore | F. Schmidt |
| 11:30-12:00 | M. Bartelmann | N. Porqueres | |
| 12:00-12:30 | Discussion - Theory | Discussion - Outlook | |
| 12:30-14:00 | Lunch | ||
| 14:00-14:30 | E. Kozlikin | Free discussion and collaboration | |
| 14:30-15:00 | M. Zennaro | ||
| 15:00-15:30 | Break | ||
| 15:30-16:00 | E. Chisari | ||
| 16:00-16:30 | H. Winther | ||
| 16:30-17:00 | Discussion - Methods | ||
| NAME | INSTITUTE |
|---|---|
| Thomas Bakx | Utrecht University |
| Matthias Bartelmann | Institute for Theoretical Physics, Heidelberg University |
| Sante Carloni | UTF Charles universty in Prague |
| Elisa Chisari | Utrecht University |
| Tristan Daus | Institute for Theoretical Physics, Heidelberg University |
| Despoina Farakou | FZU, Institute of Physics of the Czech Academy of Sciences |
| Pedro Ferreira | University of Oxford |
| Peter Filip | Institute of Physics, Prague |
| Bartolomeo Fiorini | Queen Mary University of London |
| Flaminia Fortuni | INAF - Osservatorio Astronomico di Roma |
| Dražen Glavan | CEICO, FZU - Institute of Physics of the Czech Academy of Sciences |
| David Heyrovský | Institute of Theoretical Physics, MFF, Charles University |
| Stéphane Ilic | IJCLab |
| Jens Jasche | Stockholm University |
| Elena Kozlikin | Institute for Theoretical Physics, Heidelberg University |
| Pavel Kůs | CEICO |
| Francois Lanusse | CNRS |
| DIANA LOPEZ NACIR | DF-UBA/FZU |
| Andrina Nicola | AIfA - University of Bonn |
| Natalia Porqueres | University of Oxford |
| Sabir Ramazanov | CEICO, Institute of Physics of Czech Academy of Sciences |
| Rome Samanta | CEICO |
| Ignacy Sawicki | CEICO, Prague |
| Fabian Schmidt | Max Planck Institute for Astrophysics |
| Ashim Sen Gupta | Queen Mary University of London |
| leonardo senatore | ETH Zurich |
| Constantinos Skordis | CEICO - FZU, Institute of Physics of the Czech Academy of Sciences |
| Georg Trenkler | CEICO |
| Leonardo Trombetta | CEICO, Institute of Physics of the Czech Academy of Sciences |
| Alexander Vikman | CEICO-FZU |
| David Vokrouhlický | CEICO |
| Hans A. Winther | ITA University of Oslo |
| Matteo Zennaro | University of Oxford |
|
|
|
|
|
|
|
|