SN testing

SN Search Tests

Sensitivity

The limiting magnitude of the supernvoa search software was investigated by adding artificial stars to search images and seeing which of them were detected by the search software. The test software was installed at KAIT in the ~treffers/sn/scripts directory. The magtest script generates the test. The pstar command was used to generate an image containing a random array of 70 gaussian stars with peak heights between 100 and 7 ADUs and 4.00 fwhm. (See image on the right)

Photometry was then performed on the image using the same aper_size and mconst as the search. The magnitude scale is probably correct to within a few tenths.

This test image was then added to a search image and the sn_search was run. Then a list comparing the observed magnitudes to the intrinsic magnitudes of the stars was made. A magnitude of 25.0 was assigned if the star was not detected. Here is a graph showing a test run on the image Mar7jfwf.fts. The red line has unit slope between the observed and instrinsic magnitudes.

Conclusions

It can be seen that save for one star, the search is essentially complete in this image to 17.5. It can be seen that the observed magnitudes agree to the intrinsic magnitudes within an rms error of a a few tenths . There is a also systematic displacement; I don't know the reason for this error.

Limitations and Future Plans

The sensitivity graphs should be repeated under different ambient light levels and seeing conditions as well as with differering seed values of the random number generator.

Sensitivity can be traded off with false alarms, before it runs up against photon noise. We have a ways to go.

Experiments

Here is another image Mar7jfhs run with MINSIG=20 found in the routine sn_finder.

And another with MINSIG=15 for Mar7jfsg.

Last Updated 3/21/97 - RRT