Note: a new task -- "FLOPM" -- is available in recent versions of AIPS
      (31DEC07 and later), which also reverses the order of spectral
      channels, making UVFLP as described below unnecessary.
      However, if you run FLOPM on the lower sideband, then DBCON
      does not like to join it to the upper sideband (I think
      because the headers have been reordered/restructured by FLOPM,
      and are no longer recognised as the same sort of data by DBCON).
      To get around this, run FLOPM *once* on the lower sideband, but
      *twice* on the upper sideband, so you get back to where you
      started, but the headers are in the same format, having gone
      through FLOPM. Then you should be able to use DBCON to
      combine the sidebands.

                      -----------------------------------

       UVFLP (to re-order channels in GMRT LSB uv data files within AIPS)
       ==================================================================

To flip the oder of spectral channels in uv databases in AIPS. Specifically for
LSB data from GMRT, datasets, so they can be combined with the corresponding
USB dataset, using DBCON.

Files:

  UVFLP.FOR    <- source code (adapted from AIPS template, FUDGE.FOR)
  UVFLP.HLP    <- help text

To compile (as per Section 12.5.2 of the AIPS Cookbook):

  source /opt/aips/LOGIN.CSH    <- to setup AIPS etc (or whatever, in
                                                      your setup)
  $CDTST                        <- to setup logicals

  LIBS $APLPGM > UVFLP.OPT      <- to create linking options file

  COMLNK UVFLP UVFLP.OPT        <- to compile and link

To use:

  setenv MYAIPS `pwd`           <- in directory where compiled

and then within AIPS say

  VERSION 'MYAIPS'

to make UVFLP available.

                      -----------------------------------

For GMRT data, the LSB data have a negative frequency increment, and the USB
data a positive frequency increment, so cannot be combined as they stand in
AIPS with DBCON. However, the GMRT data are on a consistent, single, linearly
incremented frequency scale, so can be combined, after re-ordering (and padding
with empty channels).

It is also possible to combine GMRT uv datasets which have had a number of
channels averaged together, provided the number of channels chosen and averaged
mean that the channels are still on a single, linearly incremented frequency
scale.

For example, consider data which has had 10 channels averaged together, shown
below in terms of frequency increasing left to right.

1) Original data:

        128 <--- channels <--- 1|1 ---> channels ---> 128
        ------------------------|------------------------
                   LSB          |         USB
        ------------------------+------------------------
                             610 MHz

(which channel width 0.125 MHz).

Note: "+" indicates reference frequency.

2) After selecting channels 6 to 115 (inclusive) in each sideband, and
averaging 10 channels together using SPLAT:

         11 <-- channels <-- 1|   |1 --> channels --> 11
         ---------------------|   |---------------------
                  LSB         |   |        USB
         ---------------------+   +---------------------
                      608.75 MHz 611.25 MHz
                    centre of ch1 in each file

Note: 10 (original) channels around 610 MHz are omitted, which correspond
      to the width of a (new) averaged channel

3) After re-odering the LSB data using UVFLP:

         1 --> channels --> 11|   |1 --> channels --> 11
        ----------------------|   |---------------------
                   LSB        |   |        USB
        ----------------------+   +---------------------

4) After running data through BLOAT, to pad with empty channels:

        1  --> channels --> 11|12 13 --> channels --> 23
        ----------------------|-------------------------
                   LSB        | empty channels padding
        ----------------------+-------------------------

i.e. after running BLOAT in previous re-ordered LSB data with APARM=23,1,1,11,0

        1  --> channels --> 11 12 |13 -> channels --> 23
        --------------------------|---------------------
          empty channels padding  |         USB
        --------------------------+---------------------

i.e. after running BLOAT in previous USB data with APARM=23,13,1,11,0

These datasets can now be combined with DBCON, since the frequencies of
channels 1 to 23 in each file match precisely.

Notes:

(1) The output from UVFLP also (optionally) changes the sign of the channel
width in the 'FQ' table, as this is negative for GMRT LSB datasets. Use
DOINVERS=1 to change the sign of the channel width in the output FQ table.

(2) UVFLP copies associated tables, except for any 'FG' tables (since the
channel numbers will change). Apply any 'FG' tables using UVCOP, before
using UVFLP.

                      -----------------------------------

Dave Green (dag9@cam.ac.uk) -- MRAO -- 2007 April 2nd