- Computer Maintenance--
R. Cutri announced
that karloff and lugosi would be unavailable for one
hour on Wednesday (9/13) because of scheduled maintenance. Lugosi has been
more stable ever since a number of component replacements were installed in an
attempt to correct its previous tendency to turn off CPUs (apparently the new
power supply was the critical ingredient).
- Protocamera Data Product Status--
R. Cutri reported
that we appear to be on schedule for finishing the
protocamera data products. This means we expect that in two weeks all data
from photometric nights will have been processed for bandmerging and
confirmation and will be available via XCATSCAN. The data products for the
two different cameras will be stored separately but both will be
accessible. T. Evans
reported that a new version of XCATSCAN should be online by the end of this
two-week period also, so that the data products will be accessed through the
new version. This will include a 2MASS option that will require a password. A
separate 2MASS version of XCATSCAN will no longer be required.
- Twilight Flat Stability--
R. Cutri reported that
an analysis of the twilight flat stability has been
performed. This involved computing a mean twilight flat for the entire
observing period from the individual nights' twilight flats, and the standard
deviation about the mean in each pixel was computed and viewed in image format.
There was structure in the standard-deviation image, with extreme values of
about 5% tending to be in the corners, where light-leak variations may have
had some impact. For about 98% of the pixels, the standard deviation was 1% or
less. The conclusion is that the twilight flats appear to be quite stable for
this observing run.
- Obs/IPAC Interface Document--
J. White reported
that M. Rudenko had sent comments on the Obs/IPAC
Interface Document Rev. H. The suggested modifications appear useful and have
low impact on 2MAPPS. J. White and J. Fowler will go over them in detail and
J. White will incorporate them in the next version of the document.
- CsFlat Divide-By-Zero Error--
A divide-by-zero
error was intercepted by the user-installed IEEE error
handler in CsFlat running some April '95 J scans on karloff (CsFlat is the
protocamera pipeline version of the frame flattener that uses the flattening
method anticipated for 2MAPPS in the DFLAT subsystem). The handler prints a
message showing the error code and (supposedly) program location and then
terminates the program. Past problems with IEEE error handlers under Solaris
have led to the discovery that the information passed to the handler is
generally erroneous in optimized programs, and only the nature of the call is
informative (e.g., in this case the divide-by-zero handling routine was
called). A search of the CsFlat code had revealed that the program was
vulnerable to unmasked zero pixels in the responsivity image. J. Fowler
and J. White agreed to pursue this after the meeting (installing a trap
for this condition subsequently revealed that there was indeed one
unmasked zero pixel in the responsivity image; code was installed to
mask the pixel at that point, bypass the division, and print a message
identifying the fixup and the pixel involved; the set of scans then ran
successfully).
- Flowchart Report--
J. White reported that
flowcharts produced by the graphics department in
Spalding were fine in hardcopy form but apparently unusable in postscript file
format. J. Fowler reported that he had examined the disks and it appeared that
Macintosh files had been written on DOS disks and vice versa. All attempts to
view the flowcharts with ghostscript or to print them on local printers
failed. This approach will therefore be abandoned.
- Extended Source Processing Near the Galactic Plane--
S. Terebey reported further on the extended source processing in or near
the galactic plane (see last week's minutes). GALWORKS generates too many
false-alarm extractions in regions of significantly varying background. It
appears that the main problem is background removal under these conditions. It
does not appear likely that setting thresholds differently can help
sufficiently to achieve the related science goals. She and T. Jarrett
will study alternative methods for handling the background.