T. Jarrett pointed out that some aspects of the galaxy processing would benefit if the coadd overlap in-scan were made the same as the scan-to-scan cross-scan overlap, i.e., 52 pixels. G. Kopan reported that this is controlled via a NAMELIST variable, and it would be set up in the pipeline NAMELIST to provide this overlap. J. Fowler mentioned that in the survey, precession will cause scans along of-date lines of constant right ascension to yield variable cross-scan overlap between scans taken a year or two apart. T. Jarrett indicated that this would not produce any significant problems for the galaxy processing.
L. Fullmer has begun working with the galaxy processing developers; she is using NED and the digitized POSS to provide galaxy information to aid in developing the code. T. Evans is assisting B. Light in the time-variable PSF analysis. G. Laughlin is preparing to assist R. Beck in the pipeline operations for the full April '95 data processing; she will prepare the darks and run other scripts that perform miscellaneous setup tasks that must be accomplished before a night's data can be run through the pipeline.
As stated in last week's minutes, csflat will not be used in the production pipeline (it will be tested in the analysis pipeline), but the stability of the responsivity image derived from twilight flats will be monitored to see if it would have supported the use of csflat in place of cflat.
Since analysis will be running on karloff with undelivered code, care must be taken to keep straight which versions of the software are being picked up in any given version of the pipeline or private development scripts. The production pipeline will use delivered code in the /ops/bin directory. The /dev/bin directory contains code generated as part of the delivery process. G. Kopan suggested used of the /del/bin directory as a "waiver" area and described how to override default path information at the top of a script to pick up specific undelivered versions of code for analysis and testing. J. Fowler pointed out that most programs used in the pipeline require command-line parameters, and in the absence of a full set of such parameters, they display a "tutorial" screen that shows how the command line should be used. This tutorial screen includes the version of the program. It is therefore suggested that (a.) all programs display a tutorial screen that includes the version number and either quit afterwards if command-line parameters weren't given or else permit a no-execution command-line/NAMELIST mode, and (b.) all programs to be used in a script could be executed without command-line parameters at the top of the script to get the version numbers for whatever the path in effect picks up. This would allow instant verification that (even automated testing whether) the desired versions will be invoked further down in the script.
B. Light reported that pfrac (the "enpixelated energy" parameter provided by G. Kopan's bumps program for use in kamphot to select an appropriate PSF) varies from 0.33 to 0.6 over the night of 95-04-25 for various bands (J has the lowest values, Ks the best, and temporal variations are nicely correlated). The kamphot PSFs (currently being computed by kamphot, not selected from a seeing-dependent set, since the latter does not yet exist and is in fact being generated in these runs) correlate well over time with the pfrac values. It has become clear, however, that the PSF/aperture-photometry hybrid technique for determining point-source fluxes will have to depend on the seeing, since the worst acceptable seeing involves some loss of flux outside the aperture. Either the aperture size must vary with seeing, or else the linear-combination weights for combining PSF and aperture-photometry values must become functions of seeing. Further study of this issue will be conducted.
R. Cutri reported that some posrec divergences have been observed on Ks scans near the galactic center (e.g., scans 111 and 112). GSC matchup confusion is suspected; this will be investigated further.
Since the pipeline operations are the highest priority at the moment, we will postpone dedicating working-group time to the system monitor until the middle of next month. The meeting of June 13 will be tentatively planned as the time when the working group will focus specifically on the system monitor. Members with ideas or requirements for this function should prepare to discuss them at that time.