290 likes | 404 Views
Simulations of dust in interacting galaxies. Patrik Jonsson, UCSC In collaboration with TJ Cox, Joel Primack, Jennifer Lotz, Sandy Faber…. Purpose. Make realistic “simulated observations” of merger simulations: Broadband images Spectral Energy Distributions
E N D
Simulations of dust in interacting galaxies Patrik Jonsson, UCSC In collaboration with TJ Cox, Joel Primack, Jennifer Lotz, Sandy Faber…
Purpose • Make realistic “simulated observations” of merger simulations: • Broadband images • Spectral Energy Distributions • Requires radiative transfer to take dust effects into account
Monte-Carlo method “Photons” are emitted and scattered/absorbed stochastically
Outputs • Data cube for each camera, typically 300x300 pixels x 500 wavelengths • Can be integrated to give images in broadband filters • Or look at spectral characteristics • Absorbed energy in grid cells • Determines FIR luminosity reradiated by dust • Devriendt FIR template SED is added to integrated spectra
To Date: • @ 20 merger scenarios completed • @ 50 snapshots/scenario • 11 viewpoints/snapshot • 10 filters/viewpoint = • Many images… 100,000 images, 10,000 SEDs Total of 1TB data
Sbc vs. G-series galaxies G3G3b-u1 Sbc201a-u4 G-series has less gas and hence less star formation and less dust. (urz color)
With dust Without dust (urz color)
Integrated energy UV/vis brightness practically constant
Magnitudes & Colors Rapid change of attenuation and color at coalescence
All simulations Mostly different orbital configurations Looks good…
Comparing to Heckman et al (98) Explored correlations between quantities for starbursts Also Looks Pretty good
Real Selection effect
But this is not so good… correlation is in the wrong direction!
The effect of mass Dust direction Mass direction Fiducial 3 different Sbcs with different masses
The effect of IMF Slope -3.3 Slope -2.35 Attenuation peaks at 60% instead of 80%
The effect of orbit Fiducial (prograde-prograde) Retrograde-retrograde RR is about 50% brighter, but only in IR
The effect of dust model Milky-Way-type dust SMC-type dust
Future • Morphological analysis (Jennifer) • SCUBA source comparison (Chapman) • Improve SAM burst recipe • … • What are we going to do with all the data?
3 steps For every GADGET snapshot: • SED calculation • Adaptive grid construction • Radiative transfer
Adaptive grid 200kpc size with max resolution 2pc, equivalent to a 1e5^3 uniform grid but with only 100k cells.
Adaptive Grid construction • Start with uniform grid (10^3) • Recursively subdivide cells into 2^3 subcells, until • Maxlevel is reached • Cell size < min(r_i)*fudge • Recursively unify cells as long as • (Sigma gas/<gas> < gas tolerance AND • Sigma L/<L> < L tolerance) OR • “cell is uniform enough that < 1 ray will be affected by unification”
SED calculation • Convolve SFR history with stellar model • Disk stars uniform SFR for 8 Gyr • Bulge stars instantaneous burst 8 Gyr old • Single metallicity for SEDs • Formed stars expand • 1km/s velocity dispersion • End up with SED (500 points) for each particle
MC input parameters • M_dust/M_gas • Effectively determines metallicity of gas at the start of the simulation • M_dust/M_metals • From metals produced during the simulation • Dust model (Draine 03 MW) • Dust opacity, albedo and scattering characteristics And the info from the grid, of course, luminosity and density of gas & metals in the cells
Radiative transfer stage • Run entire SED at once without scattering • Run with scattering for a single wavelength • 10^6 rays per wavelength, 11 view points • Repeat for 20 wavelengths between 20nm and 5um • And for lines (H alpha and H beta) • Interpolate SED to full resolution