There is a "chunk of the Virgo consortium universe" available for you here. The data come from a massive simulation of a cube of the universe measuring 140 Mpc on a side. Details of the simulation and the galaxy creation can be found at http://www.mpa-garching.mpg.de/Virgo/data_download.html The data give the x, y, and z coordinates in Mpc and star formation rate in solar masses per year of 8384 simulated galaxies.

We are going to define subsets of galaxies as "late types" (ie Sb/Sc spirals) and "early types" (ellipticals and S0's) based on their star formation rates. Let's say late types are things with SFR's > 1 Msun/yr, and early types are things w/ SFR's < 0.1 Msun/yr. (Does this definition make sense?)

Plot up x vs y for all galaxies to give a feel for what the data look like. Then do the same thing just for E's and just for S's. Describe any differences and explain whether or not it makes sense.
Now calculate the two point correlation function for all the galaxies (see below for how to do this). Do this for seperations between 2 and 20 Mpc, then fit a power law and see what slope you get. Describe qualitatively what this is telling you about how galaxies are distributed.
Now do the same thing just looking at 2 point correlation function for spirals only, and again for ellipticals only. Plot 'em up, fit a slope, and describe the differences. ie what does this tell you about the clustering of different types of galaxies in the universe, and does this make sense qualitatively based on what you know about galaxies?
Compare your values for the clustering properties of different galaxy sets (all, early, late) to those published in the literature.

Here now is a galaxy redshift catalog taken from the latest SDSS results (courtesy Prof Zehavi). This now gives you RA, Dec, redshift, and g-r color for SDSS galaxies (RA and Dec are both in degrees). This is a volume limited sample of L* galaxies, and I have trimmed the volume so that it is a rectangular volume (which will make life much easier for the analysis).

Go through the same steps as you did for the Virgo Simulation. In this case, let's divide our galaxies by color: make our "early types" galaxies with (g-r)>0.85, and our "late types" galaxies with (g-r)<0.65.

First we need to transform our dataset from observed spherical coordinates (RA, Dec, z) into physical cartesian coordinates (x,y,z). Turn redshift into distance using Hubble's constant (H0=72), and then convert spherical to cartesian. You can easily find spherical->cartesian conversions online, but just be careful of definitions: most definitions have an Dec-like angle measured from the z-axis, while Dec itself is actually measured from the celestial equator.

Once you've done this, you should be able to run your code on the SDSS data.

Plot up the different distributions of galaxy types, as you did before. Describe the different galaxy distributions, and also differences between the real distribution and that of the simulated catalog.
Calculate the 2pt CF for both early and late types galaxies, and for the dataset as a whole, just like you did for the simulation.
Compare your theoretical results with your observational ones. How are they similar? How are they different? Why are they different?
What do the results tell us about clustering of galaxies of different type?

Remember the 2ptcf describes the probability of finding two galaxies seperated by a distance r over and beyond that expected from random distribution.

The simplest estimator for this is given by

1+xi(r) = DD(r)/RR(r)

where xi(r) is the 2ptcf, DD(r) is the number of pairs in the dataset with seperation r, and RR(r) is the number of pairs with seperation r that you'd expect just from random points. Note that this expression assumes equal total numbers of data points and random points.

This is easy to calculate, but very time-consuming (computing time ~ N²). I strongly recommend running your code using a compiled language (Fortran, C, etc), not an interpreted one (Matlab, Supermongo).

Astr 328/428 Group Project #2