Back to ​http://svn.pan-starrs.ifa.hawaii.edu/trac/ipp/wiki/Zoology

Outline

This page outlines an investigation into the cross-matching used by the Durham group to generate the 'true positives' and 'false positives' data subsets. Durham used TOPCAT to perform the cross-match, and presented the results as FITS files. Our aim is to reproduce the same results as Durham then experiment with different cross-matching parameters.

Coverage

As can be seen from the plot below, the section of Sloan stripe 82 used by Durham does not fully cover the SAS footprint. This needs to be taken into account when performing crossmatching and determining false positives.

Durham-supplied data

Durham supplied the following files (FITS binary table format) for analysis.

File	Rows	Comments
footprint-stripe82.fits	449,254	Sloan Stripe 82 master file
footprint-g-all-psphot.fits.bz2	282,604	all IPP g filter
footprint-r-all-psphot.fits.bz2	278,280	all IPP r filter
footprint-i-all-psphot.fits.bz2	320,655	all IPP i filter
footprint-z-all-psphot.fits.bz2	258,130	all IPP z filter
footprint-y-all-psphot.fits.bz2	175,578	all IPP y filter
footprint-g-not-matched-psphot.fits.bz2	171,973	all IPP false detections in g
footprint-r-not-matched-psphot.fits.bz2	67,455	all IPP false detections in r
footprint-i-not-matched-psphot.fits.bz2	59,973	all IPP false detections in i
footprint-y-not-matched-psphot.fits.bz2	46,729	all IPP false detections in z
footprint-z-not-matched-psphot.fits.bz2	51,391	all IPP false detections in y

Durham cross-matching

Durham used TOPCAT to perform the cross-matching. They used the following settings:

• 'Algorithm' : Sky
• 'Max Error' : 1.0 arcsec
• 'Match Selection' : Best Match Only

It is thought (by Nigel Metcalfe) that Peter Draper cut the Sloan data on BINNED1 = true. I applied this cut, and it only removed 428 sourced out of a total of 449,254.

Tiger-team cross-matching

We attempt to first reproduce the Durham results, then experiment with different cross-matching constraints.

Attempts to reproduce Durham results

Using TOPCAT with the same data and same settings as Durham, while also first removing all IPP detections outwith Stripe 82, we are able produce the the same results (we cut the Sloan data using BINNED1 = true, as above).

Results of all cross-matching are shown on the spreadsheet ​here

Cross-matching with different error radii