| Version 1.9 Build 1556
|
|
Next: Data-Loaders
Up: NOTE 200 - VLBI REQUIREMENTS FOR AIPS++
Previous: Introduction
Subsections
Most of this document deals with required functionality of AIPS++
``tasks''. However, the VLBI community is also very concerned about
the programmability of the package. Traditionally, many VLBI
astronomers have (necessarily) been involved in algorithm development.
And it is clear that this is a continuing effort in all aspects and
stages of VLBI processing. It is recognized that classic AIPS has
imposed a high threshold for non-specialists to test and implement
new methods.
In AIPS++ there are many layers of increasingly complex software. It
is hoped that the user interface and script language (Glish) will
provide a platform for simple data operations and straightforward
processing. However, it seems unavoidable that at least some fraction
of algorithm implementation by VLBI astronomers requires the use of
compiled code. Because of the large datasets involved in VLBI
processing, speed is a consideration for these projects. Easy access
to class libraries, interfaces to C and FORTRAN, and documentation on
the appropriate level seem to be required to ensure programmability.
Skeleton tasks, tutorials and even training courses should be
considered.
There need to be flexible and simple ways for the users to manipulate
their u,v datasets, for example:
- The user should be capable of using the host system's
capabilities and utilities to manage hisher datasets. The datasets
should use a normal file name, and the user should be able to use
the host directory hierarchy to best organize the data.
- Tasks should generally take multiple input u,v datasets where
this makes sense. For example, the map making program should be
capable of taking multiple input datasets, all of which contribute
to the output images with user defined weights.
- There needs to be a flexible way for the user to select the
particular subset of data, in a dataset, to be processed. As well as
selection based on time, antenna number, frequency, subarray, etc,
it should be possible to select based on the values of other data
(including monitor data).
- It should be possible to extract a subset of data from a dataset
and manipulate it in some powerful command language. This would
include displaying the data and optionally replacing it in the
dataset (e.g. multiply the amplitudes for some antenna by an
arbitrary factor).
- It should be possible to extract a subset of data from a dataset
in a variety of formats (e.g. FITS or plain text) in order to
transfer the data to other programs or packages. It should also be
possible to read the modified data back into AIPS++ using the same
formats.
- For applications where the built-in tasks and command language
features are insufficient, there needs to be a program interface to
allow the casual programmer reasonable access to the data. Some
flexibility and efficiency can be sacrificed in making this
interface comparatively simple. FORTRAN programmers should be
supported.
Another point we want to emphasize is one of documentation for
non-specialist users. Our users often have no previous experience in
synthesis imaging and start directly with VLBI. They need a
cookbook-like introduction as well as tutorials.
For the VLBI user community it is of great importance that AIPS++ is
portable to all systems that have significant support in the global
VLBI community. At the moment these are mostly UNIX systems; besides
Solaris and SunOS, especially DEC Alpha, HP/UX and Linux platforms are
popular around the VLBI world.
To get a feel for the possible data types that VLBI processing has to
take into account, we list below possible VLBI observations for which
AIPS++ processing andor application development should be possible.
The first part of this list concentrates on data types that we
consider part of (future) standard VLBI practice:
- continuum observations in single or dual polarization
- spectral line observations in single or dual polarization
- observations in which polarization, frequency, andor pointing
centre may be rapidly switching in time.
- simultaneous observations in multiple frequency bands (e.g. for
observing multiple lines simultaneously or multi-frequency
synthesis or S/X), with variable numbers of channels within each band
- pulsar-gated data
- mosaiced observations with severalmany pointing centres
(e.g. large line sources, gravitational lenses). This must be
supported at both the u,v dataset and image dataset level
- polarization data with unequal uv-sampling, where all four
polarization parameters are not available simultaneously, as might
occur for networks with inhomogeneous polarization sampling or
time-switching of the recorded polarization
- combination of data from different observations that have
different (but overlapping) spectral sampling
- time-series data of profiles and visibilities (e.g. pulsar
data with bin number as a data axis)
- multi-array datasets (e.g. MERLIN+EVN)
- multiple correlations from multi-field centre observations,
for which the calibration should be identical (e.g. gravitational
lenses)
More specialized dataset types which could be taken into
consideration in the design of AIPS++ are:
- space VLBI observations
- observations during which the source changes structure (e.g. SS433)
- cluster-cluster data (e.g. WSRT-VLA multi-antenna VLBI)
- burst sampling for mm-wavelength VLBI
- triple correlation (including the case where one of the visibilities
has a different frequency)
- combinations of the above (e.g. pulsar gated data in
cluster-cluster mode)
Figure 1:
A possible model for data
flow for VLBI data in AIPS++. The correlator specific format
may contain all the calibration data. Examples of some of
these are denoted in this figure by their classic AIPS table
two letter codes: e.g. PC for phase cal data, SN for amplitude
calibration for example based on state counts, TY for system
temperatures, FG for flagging, MC for correlator model
components. Standard reduction has a specific VLBI part which
involves calibration based on external data and fringe fitting.
This allows the Measurement Set to be averaged in frequency and
time. Following this, standard imaging and (self)calibration
techniques are available to improve the image quality and the
model of the source and sky.
|
Next: Data-Loaders
Up: NOTE 200 - VLBI REQUIREMENTS FOR AIPS++
Previous: Introduction
  Contents
Please send questions or comments about AIPS++ to aips2-request@nrao.edu.
Copyright © 1995-2000 Associated Universities Inc.,
Washington, D.C.
Return to AIPS++ Home Page
2006-10-15