Simulation-based Large Scale structure cosmology

The Good, the Bad and the Ugly

Carol Cuesta-Lazaro

\{ \Omega_\mathrm{M} , \sigma_8, ... \}

Forward Theory

Gravity

N-body

Initial Conditions

Simulations

Effective Field theories

Galaxies

Empirical

Hydro

Forward Survey

Mask, fibres...

Summarise

Initial Conditions

Observation

25 \, h^{-1}\mathrm{Mpc}

Empirical halo-galaxy connection

Identify host Dark matter haloes

p(N_g|M_h)

p(N_g|M_h, f_\mathrm{env})

Central galaxies : CoM largest sub-halo

Satellite galaxies : particles in halo

p(

)

|

\mathrm{Cosmology}

N-point functions

Counts-in-cell

Wavelets

Marked tpcfs

kNNs

Voids

Optimise information on cosmological parameters

Robust Surprises

Summarise

Density-split clustering

\xi^{qg}(s, \mu)

Quintile-galaxy crosscorrelations

Voids

Clusters

Quintile autocorrelations

\xi^{qq}(s, \mu)

Voids

Clusters

Fisher Information

Dark Matter halos Quijote Simulations

BOSS CMASS DR12

0.45 ≤ 𝑧 ≤ 0.6

The Good

The Bad

The Ugly

Simulations can model non-linear clustering

Forward modelling + ML = constraints from any summary

Ad-hoc galaxy-halo connection

Halos are social constructs

Limited number of simulations, prior limited

Difficult to estimate robust model errors