15APR99 AIPS Developments

The 15APR99 version of AIPS is under development at the present time. It is run as the TST version at all NRAO sites and at a variety of other sites around the world. Changes in this version are installed on a (nearly) daily basis by the so-called midnight job. This may be run once a copy of 15APR99 is installed. There have been a number of changes to IMAGR, a variety of bugs in calibration were fixed, FITLD was improved, and a better way to write FITS-format output was created. Among the changes made to AIPS for 15APR99 are:

*VLB matters:

  1. FITLD for VLB:
      a. New proc MERGECAL to merge redundant information created when
         calibration transfer data are concatenated.
      b. Corrected handling of BIF, EIF, BCHAN, ECHAN which were honored
         even if crazy (e.g. BIF of 3 in a 2-IF data set).
      c. Corrected handling of input tables to conform to new standards
         for the contents of these tables and to allow some older
         variations.  Also changed to allow, with warnings, data out of
         time order.

  2. CVEL - new option to control whether Hanning smoothing is used.
     This is a good idea if there are narrow lines.  Otherwise, the
     shifting will produce ringing in the spectrum.

  3. Fringe fitting:
      a. Corrected bug that kept FRING from looping correctly over
         subarrays and a bug that caused it to die when rate-only
         solutions failed.
      b. MBDLY now has FREQID and SUBARRAY adverbs, allowing more than
         the first of each to be used.
      c. KRING copies multi-band delays properly to all IFs.  It uses a
         new scheme for computing SNRs detailed in AIPS Memo 101.  KRING
         has been extensively tested using a new test data set created
         by DTSIM and compared with FRING.  KRING runs faster in many
         cases, uses less memory, is able to handle larger problems then
         FRING.  MBDLY frequently does not converge on correct

*General uv data and calibration:

  4. Calibration errors found:
      a. Since 1992, old calibrations were carried forward rather than
         being replaced by failed solutions (with the output data
      b. Faraday rotation corrections were not initialized properly.
      c. Calibration was not updated often enough for good antennas when
         some antennas had large gaps between solutions (1998-08-12
         through 1998-10-17).
      d. Dispersive delay corrections are now done of there are non-zero
         values in the DISP 1 column of the CL table.
      e. QUAL and CALCODE are now used to select a list of sources some
         of which are selected or de-selected with the SOURCES adverb.
      f. 1.0 is no longer subtracted from autocorrelation data when
         bandpass calibration is applied.
      g. Calibration was not interpolated when the calibration values
         were updated unless the time also changed by "enough"; changed
         to force an update.

  5. FITAB - new task to replace FITTP gradually.  It can write more
     compressible image FITS files by allowing the user to control the
     quantization (scaling to integer form).  It writes uv data in
     binary tables forms, optionally writing them as compressed data
     (like on disk) and in pieces.  The former saves space while the
     latter provides better recovery from media error (e.g. parity,
     end-of-tape, etc.)  Older FITS readers can read the images
     produced, but cannot read the uv data tables form.  Please try
     FITAB, but not when you are taking the output to 15OCT98 or older
     AIPS versions or to packages such as difmap.  15APR99 versions of
     PRTTP, UVLOD, and FITLD can handle the new table format and PRTTP
     can now report on disk files.

  6. TECOR - new task to perform ionospheric Faraday rotation and
     dispersive delay corrections using maps of the free electron
     content of the ionosphere in standard IONEX format.

  7. CLCOR - The source position correction had its parameters changed.

  8. FLGIT - Changed meaning of APARM(4 and 5), changed FLAGVER
     defaults, improved displays.

  9. PRTUV - new adverbs were added to allow the user to specify scaling
     parameters so that the task does not have to spend time determining

 10. ACFIT - was corrected to handle any size of input problem, to write
     correct REFANTs to the output, and to count output samples correctly.

 11. UVCOP now does a better job of deleting flagged data when IF or
     channel selection is being done.

 12. UVCON - new task to generate a uv database for an array
     configuration specified by the user.


 13. IMAGR -
      a. Changed to finish a full major cycle even if NITER is
      b. Corrected bug in doing primary beam correction for multiple
         fields and fixed it to handle coordinates correctly even or 3D
      c. IMAGR, REBOX, and FILEBOX can now display boxes which are not
         completely on the visible TV area and revise only those that
         are fully visible.
      d. OVERLAP 2 is very useful and is now allowed without requiring
         DO3D to be true.  This mode uses a modified scheme for
         selecting the next field to Clean which will avoid the Cleaning
         of noise fields to some extent.
      e. The gridded subtraction of Clean components has had an error
         for a very long time.  The W term was not handled correctly
         when the first channel to be subtracted was not at the
         reference frequency.  Using BIF > 1 or the primary beam
         correction could cause significant degradation in the quality
         of the subtraction in some cases.
      f. OBOXFILE is a new option to write out the Clean boxes set
         interactively in IMAGR and SCMAP.

 14. Images may now be up to 16384 on a side, a parameter which should
     be honored throughout AIPS.

 15. FLATN - has another adverb to delete pixels around the edge of each
     input image.  Changed it to use windows to speed up the operation.

*General matters:

 16. TAPES - new verb to display what tapes are available on the local
     or on a remote host.  It will even try to go to a remote TPMON over
     the Internet to get this information if it is not available
     locally!  (Of course, it works only to 15APR99 sites.)

 17. The use of INTEGER*2 was reduced in AIPS to the minimum possible
     since it is a source of compiler error (e.g. the newer egcs
     compiler versions).

 18. XAS no longer has a small uninitialized row at the bottom of the

15OCT98 AIPS Developments

The 15OCT98 version of AIPS is the standard "release" now available to all sites. See the release notes for details on how to get a copy. The release is described in the 15OCT98 AIPSLetter. Among the changes made to AIPS for 15OCT98 were:

*VLB matters:

  1. FITLD improvements:
      a. Calibration transfer from the correlator into AIPS data sets
         has been implemented.  It should no longer be necessary to
         copy log files into text files which are then read into AIPS.
         See Chris Flatters' memo or web page for details:
(memo) (web page)
      b. The computation of the reference frequency for lower sideband
         data was corrected.
      c. The renumbering of sources was improved to avoid excessive
         source numbers and confusion over CALCODEs.

  2. Fringe-fitting in KRING was improved 
      a. to attempt forward and backward extrapolation of solutions in
         order to avoid excess computation.
      b. The SNR is now computed in a new and self-consistent way
         For details, see Ketan's
(AIPS Memo 101)
  3. Delay calibration was not previously applied if it was zero for the
     first IF/channel.  Now all values are checked and used.

*General uv data and calibration:

  4. SPLAT - is a new task to apply calibration and average the output
     over time and/or spectral channels.

  5. SPLIT - was revised to average spectral channels after calibration,
     to handle multiple subarrays as advertised, to count samples
     better, and to omit the REMOVED random parameter.

  6. The interactive editing tasks (EDITR, EDITA, SNEDT and the editing
     in SCMAP) were revised to allow editing in crowded displays and to
     provide a number of other improvements in the speed with which data
     may be flagged.

  7. Numerous errors in BPASS were corrected.  Channel-dependent data
     flagging led to improper weighting of channels, bad interpolation
     across missing channels and other problems.  The application of
     BPASS solutions from autocorrelation data to cross-correlation data
     and vice versa was fundamentally in error.  A missing antenna in an
     AN table caused addressing failures.

  8. FLGIT - is a new task to remove RFI-affected samples from
     low-frequency, multi-channel data sets.

  9. UVCOP can now apply up to 50,000 flags at a single time (ten times
     more than the rest of AIPS).

 10. Times in SN and CL tables no longer need to be the same for all

*Imaging and modeling:

 11. CONFI - is a new task to optimize array configuration designs.

 12. The NVSS web site may now be used to prepare field parameters for
     input to IMAGR.  See
 13. FLATN can now apply geometric corrections while flatten a
     3D multi-field set of images.

 14. Radial bandwidth smearing corrections are available in SAD, IMFIT,
     and JMFIT.

 15. OMFIT has steadily acquired the ability to model the universe.

*Data display:

 16. XAS is now able to do 24-bit TrueColor displays on monitors which
     support such things.  This allows true 3-color images and full
     hue-intensity displays.

 17. Split screen and roam were added to XAS as well.  They work best in
     the 24-bit mode but are usable in 8-bit.

 18. You may now have as many TV displays, message terminals, etc as you
     want.  Only one may be connected to the Internet but one can have
     1 or more local Unix socket displays.  This means that XAS runs on
     the same computer as, for example, IMAGR and AIPS.  The images may
     appear on any computer or X-terminal display selected by the
     $DISPLAY variable.

 19. Printing to the terminal may be set to detect the size of the
     current terminal (e.g. xterm) window and to pause when that size is
     filled.  Previously, the official window size was stored in a file
     and the actual size was ignored.

 20. POPS now has a much larger area for storing procedure, new
     variables, etc.

Eric W. Greisen