IPAC 2MASS Working Group Meeting #63 Minutes

IPAC 2MASS Working Group Meeting #63 Minutes, 6/06/95

Attendees: R. Cutri, T. Evans, J. Fowler, T. Jarrett, G. Kopan, B. Light, S. Wheelock, J. White

AGENDA

  1. 2MAPPS Schedule Review
  2. Pipeline Processing Success Criteria
  3. Pipeline Processing Status

DISCUSSION

  1. 2MAPPS Schedule Review --- A brief review of the 2MAPPS development schedule was conducted to see what milestones might have been affected by the ongoing developer involvement with the pipeline processing of the April '95 data. The nearest milestones are July 1 deliveries of TAPELOAD and PIXCAL/PIXPHOT. Assuming that "delivery" is taken to mean a Subsystem Design Specification document, Software Interface Specification documents, regression test baseline, user's guide, and executable software modules, then the July 1 milestones will probably have to be converted to phased deliveries. Intermediate deliveries of PICMAN, PROPHOT, and GALWORKS were due on 1 June, 1 May, and 1 June, respectively, and software deliveries have taken place which partially satisfy these milestones; design documents and the other aspects mentioned above were not involved, however.

    One obstacle to delivering documentation has been the previously stated intention of doing it in a TeX-compatible format. This has introduced a significant amount of drag, and it was suggested that it be dropped. It was also stated that the classic approach of documenting before coding was inapplicable because latter-day revelations about the instrumental behavior have necessitated prototyping well into the stage where algorithm design was supposed to be over and implementation was to be the major activity.

    The relaxation of the documentation format is one attempt to speed up the process of preparing subsystems for design reviews. It should also be noted that the entire reason for scheduling TAPELOAD and PIXCAL/PIXPHOT deliveries on July 1 was so that the developers could proceed to work on their other subsystems, not because any system integration is scheduled to take place in the near term.

    The discussion then turned to the question of whether the current pipeline processing was still continuing to divert developers' attention from 2MAPPS design and implementation. It appeared that such distraction was indeed subsiding rapidly, and it was agreed that we should make every effort to disengage immediately, let the DMT run the pipeline, and let the analysis specialists take over worrying about the April '95 data.

  2. Pipeline Processing Success Criteria --- In the process of checking output from the pipeline processing of the April '95 data, an unfortunate amount of confusion was generated by the large number of error messages, many of which were not actually indicative of anything meriting concern. As a result, it was decided that a well defined procedure was needed for certifying that each observation night's data had been processed with a high probability of success, that this procedure should be automated to the extent feasible, and that a standard notification should be provided that indicates which observations have been processed.

    It turned out that G. Laughlin had already been working on such a procedure and has distributed an email memo inviting developers to contribute information usefule for the automated checking, if they so desire. The processing status is updated regularly and is available for perusal on the World-Wide Web via http://www.ipac.caltech.edu/2mass/2mass.html (select the Protocamera Processing Status option); access is restricted to 2MASS personnel via the user-ID/password documented elsewhere.

  3. Pipeline Processing Status --- The current processing rate amounts to two days per observation night. The use of time-variable PSFs in KAMPHOT is generally seen as successful, validating the 2MAPPS design. There have been some problems, but these are expected to be solved by a more densely populated grid of PSFs and a better determination of the corresponding variances. Most of the POSREC failures have been diagnosed and fixed, but a few remain that are not understood yet. For example, failures have been observed in calibration fields, where the source density is neither very high nor very low, and occasionally the failures occur only while processing KAMPHOT sources, while other times it is only while processing DAOPHOT sources, with occasional flipflops in reprocessing.

    Some photometric dispersion is seen that is not yet understood. There appears to be a variable offset between KAMPHOT and DAOPHOT magnitudes, and this increases the dispersion in the hybrid photometry. It is hoped that an improved PSF grid will permit the DAOPHOT part to be jettisoned. Meanwhile, the nature of the variable offset will be studied.