IPAC 2MASS Working Group Meeting #106 Minutes

IPAC 2MASS Working Group Meeting #106 Minutes 10/15/96

Attendees: R. Beck, T. Chester, R. Cutri, T. Evans, J. Fowler, L. Fullmer, T. Jarrett, G. Kopan, B. Light, J. Mazzarella, H. McCallon, S. Wheelock, J. White


  1. Epoch of Coordinate System in Image Headers
  2. First 2MAPPS Executions on Rodan
  3. Disk Reconfigurations
  4. Simulation of 3-Channel Data


  1. Epoch of Coordinate System in Image Headers A question arose concerning what coordinate-system epoch was supposed to be used in the FITS headers of images. The documented epoch is 2000, i.e., the coordinate system is J2000. Exceptions to this had been seen and had caused some confusion. The problem is that the J2000 standard is not yet in effect, because the Obs/IPAC Tape Interface Specification Rev. J has not yet been implemented, although it has been agreed upon and signed off.

    The celestial coordinates passed with the scan data from the observatory to IPAC are written into the FITS headers of all data frames. From there, they are copied into the FITS headers of coadded images. These coordinates are a priori position angles, however, not reconstructed position angles. The latter are to be computed by POSFRM, and these will also be written into the image headers (in fact, PICMAN already does this, but the information it receives is not yet the real POSFRM output).

  2. First 2MAPPS Executions on Rodan R. Beck reported on the first attempt to compile and link 2MAPPS on the new Sparc 3000, ``rodan''. The software all compiled correctly, but a problem occurred while MAPCOR was being linked. Some math library routines apparently were not available. T. Evans and R. Beck will look into this.

    2MAPPS was run on rodan through PIXCAL, and generally the CPU times were cut in half compared to karloff and lugosi. CPU performance should improve even more when the new compilation options are used (in this test, the existing makefiles were used as on karloff/lugosi, not taking advantage of some new ultrasparc capabilities).

  3. Disk Reconfigurations The data base development activity on karloff will require that the machine be dedicated to this work starting on about November 1. Plans to move directories needed for other 2MAPPS development off of karloff were discussed. The plan is to begin configuring karloff as the data base machine, lugosi as the analysis machine, and rodan as the 2MAPPS software development and test (and soon, production) machine. The following conclusions were reached.

    1. Team members will be responsible for moving their own personal files from karloff to lugosi, but must wait for R. Beck to prepare room to do this, since he has to make some major reconfigurations first.

    2. Production data currently on karloff is all from nonphotometric nights and will be moved to tape and deleted.

    3. Production data currently on lugosi is from photometric nights and will be moved to karloff and deleted from lugosi.

    4. Development and delivery directories will be copied from karloff to rodan and deleted from karloff (i.e., /2massc/dev and 2massc/del). This will be done by R. Beck; developers should continue to use karloff until R. Beck notifies the team that the move to rodan is ready to be made.

    R. Cutri will consult with S. Lo about options for moving disk drives physically between karloff and lugosi as part of the reconfiguration. This is complicated by the recent I/O errors and outright failures on certain disks attached to karloff.

  4. Simulation of 3-Channel Data In order to address a concern about differences in source positions and photometry between the BIGPOS and 2MAPPS pipelines, a splinter group had been formed to investigate expanding the simulation capability so that tests of 2MAPPS could be done with absolute knowledge of what the results ought to be. This spinter group consisted of B. Light, T. Jarrett, R. Cutri, J. White, and J. Fowler. The group found that to enhance the existing simulation code to the point needed would involve a significant expense.

    Meanwhile, G. Kopan and B. Light determined that the photometric discrepancies could be attributed to the fact that BIGPOS used the older frame flattening technique which is known to be inferior to that in DFLAT (used in 2MAPPS). The position discrepancies were also explainable as due to improvements in 2MAPPS; three independent checks on the positions yielded this same conclusion. Further pursuit of simulation enhancement will be abandoned at this time.