AIPS quarterly report, 26 September 1997
========================================


release information
===================

During the third quarter of 1997 we continued to concentrate on Space
VLBI related applications.  AIPS produced the very first images made
by an array of telescopes including Halca.  In August, we released the
15APR97 version of AIPS.  As mentioned in earlier reports, this delay
was caused by our desire to ship a version of AIPS that had shown to
be able to handle Space VLBI data.  

In spite of the delay in 15APR97, we still are planning a 15OCT97
release.  Since by then more and more varied Space VLBI data will have
passed through AIPS, we intend this release to be a more robust
version of 15APR97.  After this release, we plan to use the
experimental version of AIPS CVX as TST (15APR98).  This version of
AIPS does not currently have the robustness of 15OCT97, but offers
exciting new capabilities in the areas of 3-D imaging, on-the-fly
mapping, etc.  We are confident that this version of AIPS will be as
robust as 15OCT97 by next spring.


personnel
=========

In September 1997, Gustaaf van Moorsel stepped down as head of AIPS in
order to concentrate more on AIPS++ related matters.  Tony Beasley
will oversee the AIPS group.  Eric Greisen is planning to rejoin the
AIPS group later this fall.


New capabilities
================

OMFIT has been rewritten in many ways.  Many new models have been
included and subjected to user testing.  The error-analysis has been
considerably improved.

Baseline stacking is now supported in FRING for multiple integration
times, as required by VSOP/HALCA.

BLING has been reworked to use dynamically allocated memory rather
than the AP.  This allows it to search extremely large delay-rate
spaces as may be required during the HALCA in-orbit checkout.

JMFIT underwent a thorough update during which several bugs were
fixed.

A start was made to allow AIPS/AIPS++ interoperability, which
eventually will allow AIPS tasks to be run from AIPS++.

INDXR was completely rewritten in order to add the capacity for
merging atmospheric delay and clock offset information from VLBA
model-components (MC) tables into newly created CL tables when MC
tables are present.

For full VSOP support, it was necessary to introduce a new random
parameter.  The VLBA correlator can change correlation modes with
great flexibility, leading to time-variable rate and delay
decorrelation corrections which depend on the type of frequency and
time filtering performed in the correlator. The most general solution
to this problem was to implement a correlation_id random parameter for
VLBA datasets which points to the recorded correlation modes stored in
the existing CQ table.  This change made it possible to allow time
variable correlation mode changes in general, although only time
variable OVLB filtering is activated at present.

The new adverb ALIAS has been introduced to facilitate the calibration
of Space VLBI data.  Antennas specified via this adverb will be
treated as identical for the purposes of certain tasks.  This allows
HALCA, which appears in AIPS as a conglomerate of tracking stations,
to be calibrated as one individual antenna, and to be viewed as such
for some plotting tasks.

RESEQ is a new task which will, via the ANTENNAS input, renumber
antennas in a UV file.  Space VLBI requires the ability to alias
stations together at some points in the data reduction stream.  This
is the first step for that requirement.  RESEQ will renumber all
antennas specified in the ANTENNAS adverb to ANTENNAS(1), or, using
INFILE, to some desired order.