The Cosmos in a Computer

1
The Cosmos in a Computer

• Joop Schaye

2
History of the Universe
3
The Composition of the Universe
4
The Composition of the Universe
5
Length Scales (cm)
6
Galaxy Formation
Cold Dark Matter, baryons Primordial fluctuations
Hubble Expansion Pressure
Gravity
Dark matter halos shock-heated baryons
7
Galaxy Formation
Dark matter halos, shock-heated gas
8
(No Transcript)
9
(No Transcript)
10
z14 t0.3
z5 t1.2
z1.4 t4.7
z0 t13.8
11
Density evolution
12
Why do we need simulations?

• Astrophysical phenomena are generally
• too complex solve analytically
• too extreme to replicate in the lab
• Simulations allow one to
• Gain physical insight through controlled
  experiments
• Test theories
• Measure model parameters
• Create virtual observations, used to
• Test data analysis methods
• Reveal selection biases
• Design instruments/telescopes

13
Ingredients

• Coordinates comoving
• Hubble expansion solved analytically and scaled
  out
• Cubic, cartesian for cosmological simulations
• Only place where General Relativity plays a role
• Boundary conditions
• Periodic for cosmological simulations

14
Ingredients

• Initial conditions
• Seed fluctuations from inflation
• Gaussian distribution
• Scale-dependent
• Linear evolution solved analytically, start at
  redshift z50-1000
• Components
• Cold dark matter gravity
• Gas gravity hydrodynamics, source/sink of
  radiation
• Stars (form during simulation) gravity, source
  of radiation, chemical elements
• Radiation radiative transfer

15
Ingredients

• Minimum and maximum length scales
• Much greater than scale of interest
• Resolution set by computational limitations
• (Semi-)analytic models of unresolved, (subgrid)
  physics
• Emission/absorption of radiation (cooling)
• Star formation
• Stellar mass loss (winds/supernovae)
• Black holes

16
Ingredients

• Discretization
• Mass (Lagrangian) particles or
• Length (Eulerian) cells
• Time step such that changes are small (time step
  ltlt spatial resolution / velocity)
• Adaptivity
• Spatial
• Built-in for particles (N-body, SPH)
• Grids within grids (AMR)
• Mass Varying particle masses
• Time Individual time steps

17
Computational demands

• Memory scales with number of resolution elements
• 10243 floats 4 GBytes
• 3 positions, 3 velocities, mass, density,
  temperature, chemical composition,
• CPU time scales with number of time steps, which
  scales with spatial resolution
• Parellelization
• Multiple CPUs compute simultaneously
• Distributed memory CPUs only fast access to
  local memory only

18
A slice of the z0 universe
19
Simulation videos

• Simulation of growth of structure
• Evolution of a cluster of galaxies
• Formation of an elliptical galaxy
• Formation of a spiral galaxy
• Dwarf galaxy with feedback

20
Simulation videos

• Simulation of growth of structure
• Evolution of a cluster of galaxies
• Formation of an elliptical galaxy
• Formation of a spiral galaxy
• Dwarf galaxy with feedback

21
OWLSOverWhelmingly Large Simulations
30 M CPU hours awarded
22
Main Questions

• Where is the ordinary matter and how can it be
  detected?
• What role does environment play in the formation
  and evolution of galaxies?
• Where are the heavy elements and how were they
  dispersed from the stars in which they were
  created?

23
Help

• Claudio Dalla Vecchia Oort 434
• Marcel Haas Oort 436