Friday, November 6, 2009

Mock Slice

The next step in trying to figure out what is wrong with my 3D correlation functions is to try to actually do a full reconstruction on the mock data. I'm starting with taking a redshift slice of the mock data and calculating a 3D correlation function on the slice to see if I get a logical result. This appears to be working:

Matching 3D correlation functions

To be more explicit about what I am doing to create the above. I am first taking the mock data box and selecting a "redshift slice" of the data:
For my correlation function:

Redshift slice (green) of all data points in
comoving distance range of 350-450 Mpc/h

What this looks like in Cartesian coordinates:

It is hard to see in 2D projections,
but this is essentially a thick spherical shell with inner radius
350 Mpc/h and outer radius 450 Mph/h

For Alexia's correlation function I am taking a slice in the z dimension:

Redshift slice (green) for Alexia's data

I chose the slice such that both mine and Alexia's data subsets had approximately the same number of objects (~100,000).

The distributions of the randoms continue to match the data:
For my correlation function:

Number of randoms (blue) and data points (red)
for different dimensions (ra, dec, redshift, x, y, z)

This gives me the confidence to try to do the full reconstruction on this mock data set, as that would simply be just repeating the above on several redshift bins.

I seem to be running Alexia's and my correlation functions side by side a lot to compare the results, I have made the following scripts:

svn +x *.script
./091105jess.script 091105jess.out 2> 091105jess.err &
./091105alexia.script 091105alexia.out 2> 091105alexia.err &

Cool link of the day:
Now my parents can understand xkcd:

Update (from Kevin):
As far as is concerned, is useful,
but is funny,
and exists.

No comments:

Post a Comment