**AAS 197, January 2001**

*Session 107. Galaxy Clusters and Large Scale Structure II*

Display, Thursday, January 11, 2001, 9:30-4:00pm, Exhibit Hall
[Previous] |
[Session 107] |
[Next]

## [107.03] Finding SDSS Galaxy Clusters in 4-dimensional Color Space Using the False Discovery Rate

*R.C. Nichol, C.J. Miller (CMU), D. Reichart (Caltech), L. Wasserman, C. Genovese (CMU), SDSS Collaboration*

We describe a recently developed statistical technique that
provides a meaningful cut-off in probability-based decision
making. We are concerned with multiple testing, where each
test produces a well-defined probability (or p-value). By
well-known, we mean that the null hypothesis used to
determine the p-value is fully understood and appropriate.
The method is entitled False Discovery Rate (FDR) and its
largest advantage over other measures is that it allows one
to specify a maximal amount of acceptable error.

As an example of this tool, we apply FDR to a
four-dimensional clustering algorithm using SDSS data. For
each galaxy (or test galaxy), we count the number of
neighbors that fit within one standard deviation of a four
dimensional Gaussian centered on that test galaxy. The mean
and standard deviation of that Gaussian are determined from
the colors and errors of the test galaxy. We then take that
same Gaussian and place it on a random selection of n
galaxies and make a similar count. In the limit of large
n, we expect the median count around these random galaxies
to represent a typical field galaxy.

For every test galaxy we determine the probability (or
p-value) that it is a field galaxy based on these counts.
A low p-value implies that the test galaxy is in a cluster
environment. Once we have a p-value for every galaxy, we
use FDR to determine at what level we should make our
probability cut-off. Once this cut-off is made, we have a
final sample of galaxies that are cluster-like galaxies.
Using FDR, we also know the maximum amount of field
contamination in our cluster galaxy sample. We present our
preliminary galaxy clustering results using these methods.

If you would like more information about this abstract, please
follow the link to http://ranger.phys.cmu.edu/sdss/SDSS_cluster.
This link was provided by the
author. When you follow it, you will leave the Web site for this
meeting; to return, you should use the Back comand on your
browser.

The author(s) of this abstract have provided an email address
for comments about the abstract:
nichol@cmu.edu

[Previous] |
[Session 107] |
[Next]