From initial conditions to galaxies

Corentin Cadiou

March 29, 2018

Introduction

Classifying galaxies

M vs v/\sigma
\(M\) vs \(v/\sigma\)
M vs v/\sigma
\(M\) vs \(v/\sigma\)

A few parameters

  1. \(M_{200}\)
  2. color / SFR
  1. ellipticity
  2. halo mass
  1. DM accretion?
  2. …

Galaxies live in an environment… and evolve with \(z\).

New Horizon simulation

What to do?

Theory

  • (too?) simplistic
  • hard to write, quick to run
  • enlightening

Simulation

  • “easy” to write, costly to run
  • “full” physics
  • not much statistics

Theory – formation in the cosmic web

Spherical collapse

Physical space Theoretical space
\(z_\text{form}\) \(\rho\)
\(M\) \(R\)

Smoothing is mass

Density is decreasing \(z\)

No need for costly simulations!

\[ M(z) = f(R, \rho(R))_{z=\infty} \]

from the initial conditions!

Encoding the cosmic web

Cosmic web: geometry of initial density field

Continuous function of space

Set of critical points

Frame of the filaments

Some new results…

Halos in filaments form later and accrete more than in voids.

Halos in nodes form later and accrete more than in filaments.

Tension with observations?

Red galaxies are inside nodes

The forgotten AGNs

Dubois+13
Dubois+13

Without AGN, massive disky-blue centrals

In agreement with our findings

Modeling galaxies

On the slope of a mountain … … next to a saddle

Theory – torque on filaments

Why?

Selection of galaxies in New Horizon Many galaxies have 3 filaments

Ok, so what?

  • Net torque on filaments?
  • Spin-up or spin-down of the galaxy?
  • Coherence scale?
  • Interaction with galaxy?

Simple setup

  • 2D model
  • 3 voids
  • 1 central peak
  • \(\Lambda\)-CDM power spectrum

Results

Cadiou et al, in prep
Cadiou et al, in prep
  • voids are “pushing” filaments
  • there is a net torque
  • can compute typical scale (WIP)

And why is that?

Simulation(s)

Objectives

  • Zoom of cluster at high \(z\sim 3\)
  • Follow filamentary accretion
  • Study interaction between filament / disk / halo

Example science:

  • effect on star formation?
  • effect on AGN?
  • angular momentum deposition?
  • shock formation?

Setup

Using RAMSES AMR-based code

  • AGN
  • SF + feedback
  • tracer particles

Specifics of tracer particles

  • Cheap (CPU + RAM)
  • Follow gas distribution
  • Lagrangian

The need for tracer particles

Good news: they are already implemented in RAMSES!

Oh wait…

Introducing Monte Carlo Tracer Particles

\[ p_{i\rightarrow j} = \frac{\text{mass flux}_{i\rightarrow j}}{M_i} \]

Basically: they are following the fluxes computed by RAMSES (exact)

Mathematically: \[ \lim_{N\rightarrow \infty} \rho_\text{tracers} = \rho_\text{gas} \] (which is great)

We just need an infinite number of tracer particles!

Comparison

Thank you!