AIPS - 2002 Visiting Committee Report

April 2002



At the end of 2000, a frozen 31DEC00 version of AIPS was produced for general distribution. In total, 658 distributions of 31DEC00 were made, including both daily (before 31 December 2000) and frozen (after 1 January 2001) versions. A 31DEC01 version was created, updated daily during the year. The development 31DEC01 version was distributed 556 times, to about 448 sites, after December 2000. A frozen 31DEC01 version now has been released, and is available to users; past experience indicates that the total number of distributions of 31DEC01 will top 700. The AIPS version updated daily now is known as 31DEC02.

The move of the "midnight job" and the master version of AIPS to NRAO-Socorro was done in conjunction with the initiation of the 31DEC02 version. This move was completed in January 2002, with significant help from the VLA/VLBA Computer Division. The daily versions of 31DEC02 now are accessible by an ftp copy/update procedure that is run under the Unix "cvs" utility. This eliminates dependence on "secure shell" (ssh) copies, which appear to have been a significant impediment to some users in implementing the midnight job. In addition, it should eliminate the common failures of midnight jobs that have occurred when system managers on either end make modifications to ssh.

There has been considerable activity in the systems area beyond moving the midnight job and the central code repository. Linux compilers and RedHat releases are tested as available, in order to insure optimal performance, and problems are frequently found. The AIPS group has recommended installation of the GNU compiler suite version 2.95 (now 2.95.3) along with RedHat release 7.1 or 7.2. Support for file sizes greater than 2 Gigabytes now is available in both Linux and Solaris versions of AIPS. Despite a great deal of investment in installation scripts, a substantial amount of effort in the AIPS group still goes to helping users with their installations, particularly on issues such as shared disks and remote TV displays. Currently, more than half of the messages to the "daip" e-mail list concern system or installation questions. Since Pat Murphy (head of Charlottesville Computing) now has left the AIPS group, the systems issues largely fall on the AIPS staff at the AOC.

A number of changes have taken place in the tasks that load VLA and VLBA data to disk for further processing. The task FILLM now loads and makes use of VLA weather data, including application of opacity corrections. It also gives the user the option of computing channel 0 (pseudo-continuum) in AIPS rather than taking the default calculated by the VLA on-line system. FITLD has been modified to do an improved job of handling VLBI correlator information, including application of corrections for delay quantization and improved assignment of frequency identification codes.

Wide-field imaging has been improved considerably, largely to facilitate VLA observations at 74 MHz. Improvement has been made in the routines selecting and imaging multiple fields ("facets"), and in deciding which fields to clean. Along with IMAGR, various associated wide-field imaging tasks such as SETFC, CHKFC, FLATN, and FACES have been improved.

A substantial effort has been made to simplify VLBA data reduction in AIPS. This includes the completion of procedures that string AIPS tasks together in order to carry out many of the bookkeeping and data-handling tasks sometimes required for VLBI data. These procedures are critical elements of the VLBA data-calibration pipeline and service described more fully above. In addition, the ionosphere-correction task, TECOR, has been modified so that it can transparently use multiple global GPS files and make corrections for observations which cross (or nearly cross) UT midnight. All the new features are documented in an updated version of Appendix C of the AIPS Cookbook ("A Step-by-Step Recipe for VLBA Data Calibration in AIPS"), scheduled for release at about the time of the Visiting Committee meeting. Tests of the two fringe-fitting algorithms, KRING and FRING, show comparable results and robustness in most cases.

Precession subroutines had several errors fixed to enable correct precession as well as interpolation between the input and output precession dates. Calculation of relativistic light-bending and aberration has been modified. Correct matrices for inverse precession were developed and implemented.

One member of the AIPS group (Eric Greisen) has continued a major effort in defining the "World Coordinate System" (WCS), in a series of several papers written with Mark Calabretta and in conjunction with various FITS working groups. The WCS essentially is an extension of the FITS standard, with "keywords" used to specify the possible coordinates (e.g., spatial and frequency) for images and data sets. Careful definition and agreement on the WCS are essential for future multi-wavelength research and for the various possible flavors of "Virtual Observatories." It is expected that the papers, which will be the bases for formal IAU standards, will be submitted by July 2002.

Configuration studies for ALMA, the EVLA, and the SKA have been carried out using tools developed by the AIPS group. Specifically, the task CONFI now can handle as many as 2,000 stations, an important capability for the SKA. Designs of the ALMA compact configuration and of the EVLA E-configuration, which take into account road access and fiber-optics constraints, also have been carried out using the capabilities added to the AIPS configuration tasks.