- Subsystem Interfaces with EXEC/PCP --
The previous meeting included discussion of J. White's recommended
guidelines for consistency between EXEC/PCP and the other subsystems.
The CogEs were requested to review these guidelines and design to them
unless negative effects were discovered. Most of the subsystems are
apparently in good shape already on this score, and several others have
been discussed in splinter sessions. The guidelines as defined by J.
White are now considered to be in effect.
- PIXPHOT & BANDMERGE Runs --
G. Kopan reported that PIXPHOT has now executed in a simulated
2MAPPS environment. While analysis is just beginning, the execution
appears to be correct. The throughput time was within 50% of the FDD
guideline. Some additional speedup may be possible to reduce the
excess, which is caused by processing tasks that have been added since
the FDD guidelines were set up. The greater capacity of the latest
2MAPPS hardware configuration should absorb any residual excess.
As part of this exercise, PROPHOT produced its first output source
list. J. Fowler reported that BANDMERGE has also produced its
first output source file. The latter was based on subsystem simulation
test input, not protocam data; the special test bed is needed so that
all types of confusion can be precisely set up and tested. R. Cutri
reported that K. Lester's protocam data conversion program was just
becoming capable of generating protocam based input for BANDMERGE to
use for testing. This will be used once all code branches have been
exercised via subsystem simulation test data; the current test set only
verifies the mainline processing, i.e., input, band-merging, position
refinement, and output. All confusion processing code is in place but
has not yet been tested. Statistical processing is not yet installed.
Large test sets will be generated later for timing purposes.
- CONMAN Status --
L. Fullmer reported that design of CONMAN is underway. Some methods
have been developed for doing the scan-to-scan confirmation in a
more optimal way than described in the FDD.
- R1 Cutoff Magnitudes --
Data generated by PIXCAL/DFLAT/FREXAS have been used by T. Evans in
MAPCOR development. She reported that some R1 magnitudes appear
peculiar in that they are too faint to be real (e.g., 16), and
requested information about whether there would be a magnitude cutoff
in FREXAS. G. Kopan reported that there is currently a detection
threshold, but that the photometric processing can still come up with
unexpectedly faint magnitudes or no magnitudes at all, and that
applying magnitude cutoffs would have to be a MAPCOR job, where the
correlation with R2 extractions will occur.
- R1/R2 Position Correlation in MSX Field --
T. Evans and R. Cutri showed a plot of R1/R2 position dispersions
derived from the MSX field. In this particular case, the correlations
were encouraging, raising hopes that confusion problems in the galactic
plane and other high density regions may not be as bad as expected. It
should be noted that H. McCallon has found the R1/R2 position offsets
to be variable, however, so the good correlation may not persist over
multiple scans. The excellent correlation within the data set studied
was nevertheless considered a promising sign.
- Expected Number of Sources --
R. Cutri reported that he had recalculated the number of point
sources expected to be extracted in the survey and obtained a
higher value than the previous one, about 540 million instead of 300
million. The value applies to J band, but the total number of
band-merged point sources should be close to this.
- Asteroid and Optical Associations --
R. Cutri reported that M. Sykes is developing software that will
take the four corner definition of a scan and produce a list of
known asteroids contained in the scan area (with some margin) at the
mid-scan time. The current FDD shows this process taking place inside
the PCP scan processing, but it will probably be moved outside; all
scans would be processed in a single execution prior to DBMAN, which
would then do the processing to associate these "catalog" objects with
2MASS point sources. This would eliminate the option of GALWORKS
employing the asteroid information, but the usefulness of this has not
been established anyway, and the requirements on GALWORKS have recently
been modified to reduce its avoidance of areas expected to yield
difficulties for extended source processing.
This raised the question of whether the optical catalog association
processing would be done in DBMAN or in POSPTS, or whether POSPTS might
produce a file for use by DBMAN. The current plan is to include optical
association columns in the point source file generated by POSPTS,
eliminating the need for such processing in DBMAN.