[daip] CPASS (fwd)

[Eric Greisen]

Miller Goss writes:

 > AIPS 1: BPASSPRM      0           0        Control information:
 > AIPS 1:               0           0        1: if > 0 use only the
 > AIPS 1:               0           0           autocorrelation data.
 > AIPS 1:               0           0        2: print level.
 > AIPS 1: ** press RETURN for more, enter Q or next line to quit print **
 > #
 > AIPS 1:               1           1        3: If > 0 do not divide data
 > AIPS 1:               0                       by source model
 > AIPS 1:                                    4: If > 0 store phases only
 > AIPS 1:                                       in the BP table.
 > AIPS 1:                                    5: If = 0 divide by 'channel
 > AIPS 1:                                       0' before determining BP.
 > AIPS 1:                                       If > 1 switch off the
 > AIPS 1:                                       channel 0 divide option.
 > AIPS 1:                                    6: min. amp. closure err
 > AIPS 1:                                    7: min. ph. closure err
 > AIPS 1:                                    8: > 0 => scalar average
 > AIPS 1:                                    10: > 1 => normalize ampl.
 > AIPS 1:                                       portion of bandpass by
 > AIPS 1:                                       area under BP function.
 > AIPS 1:                                       Useful for VLBI.
 > AIPS 1:                                    11: > 1, mode to allow
 > AIPS 1:                                       calibration and divide
 > AIPS 1:                                       by channel 0. Useful for
 > AIPS 1:                                       VLBI. Do not use if you
 > AIPS 1:                                       don't understand. See HELP
 > AIPS 1: CPARM        10         200        Fit parameters:
 > AIPS 1:              0.02        60        1: No. of terms.
 > AIPS 1:               2        *rest 0     2: Max. no. iterations.
 > AIPS 1:                                    3: Convergence tolerance.
 > AIPS 1:                                    4: Pre-average interval (sec)
 > AIPS 1:                                       Default: Auto -  5 min.
 > AIPS 1:                                                Cross - 15 min.
 > AIPS 1:                                    5: Fit type (1=Re/Im; 2=A/P)
 > AIPS 1:                                    6: BP table for initial val.
 > AIPS 1:                                    7: Fit phase only (1=> with
 > AIPS 1:                                       ampl.; 2=> without)
 > AIPS 1:                                    8: >0 => Autoscale
 > AIPS 1: ** press RETURN for more, enter Q or next line to quit print **
 > #
 > AIPS 1:                                    9: >0 suppress data weighting
 > AIPS 1:                                    10: >0 => save as coefficient
 >          s
 > AIPS 1:                                       rather than normal BP tabl
 >          e
 > note that BPARM 9 is not mentioned in the input file for CPASS ?

   It is also not mentioned in the help and in the Fortran.  It was
some option deprecated some time ago.
 > 
 > I want to use amp and phase cparm(5)=2 and thus I need to calibrate
 > the data. I do have docalib=2
 > 
 > CPASS1: DOCALIB=1 & CPASSPRM(5)=0 incompatible, setting DOCALIB=-1

Athol apparently thought this a bad idea - but there is BPASSP(11)
(see above) which if set will allow you to override Athol's odd idea.

 > then the BP table does look plausibe - but I need to calibrate
 > the data to fix the delays . There is a good chance I am
 > misunderstanding the help file for this old program.
 > .
 > 
 > then I try something else that also seems to fail.
 > I try to apply an initial BP table with CPARM(6)=2 - made in
 > BPASS and a table that looks quite nice.
 > 
 > ANd I get this error message. Must be making some blunder in
 > the inputs ?
 > 
 > CPASS1: Solving polarization LL, IF channel    1
 > CPASS1: BPVAL: INCOMPATIBLE INITIAL DATA IN BP:   2
 > CPASS1: Solving polarization RR, IF channel    1
 > CPASS1: ZOPEN: LUN = 43 ALREADY OPENED IN FTAB
 > CPASS1: TABINI: I/O ERROR FROM ISTAB ON BP
 > CPASS1: BPINI: ERROR   3 FROM TABINI OPCODE = READ
 > CPASS1: BPINI: ERROR INITIALIZING BANDPASS TABLE FOR READ
 > CPASS1: BPVAL: ERROR   3 OPENING BP TABLE:   2
 > CPASS1: Appears to have ended successfully
 > CPASS1: mizar        31DEC07 TST: Cpu=      25.0  Real=      59
 > AIPS 1: Resumes
 > >
 > 
 > When I plot this BPVER in possm it is fine.
 > 
That is suprising since the incompatible message leaves a file open
and then the task cannot open files properly after that.

Note - this task now writes a "normal" BP table of Reals and
Imaginaries, rather than Chebyshev coefficients by default.  This
means that a previous CPASS "BP" table cannot be used for initial
values into the next round.  There is CPARM(10) to allow you to keep
the file as Chebyshev coefficients.  Note that time smoothing of BP
tables smoothed whatever was there and I do not believe that smoothing
Chebyshev coefficients produces necessarily the correct result - hence
the new default in the task.

Note CPASS was written by Athol - it contains many mistakes, not the
least of which is that it never converges.  I have attempted to work
on it upon occasion but have never succeeded in making it much better.

Eric Greisen