[daip] Re: AIPS question: MCUBE

[Eric Greisen]

Gary A. Welch writes:
 > I'm hoping you're still Designated AIP...

     we suggest you use daip at nrao.edu - it reaches everyone when I am
out of town.

 > 
 > I've used COMB to make a cube which is the weighted mean of 4 cubes derived from 
 > observations over 4 days.  The instrument setup was the same on all days.  I 
 > found that DOALIGN -2 was needed.

     I find that most peculiar - doalign=-2 is needed only when the
coordinates do NOT match and one wants to ignore coordinates more or
less entirely.  This situation should NEVER arise in interferometry
data, although optical data may lack coordinates enough that one has
to resort to this stategy.

     Send me the IMHEAD output for the input cubes.

 > 
 > The cube looks fine, except for channel 36, which in 2 of the 4 days shows an 
 > objectionable noise pattern (Ch 36 is in the middle of the cube. Ispent much 

          You should be able to see the pattern in the individual
cubes anyway - it may be due to interference, but POSSM then should
show that too - either interference in the source or in the calibrator
(try plotting the BP table)

 > time trying to determine the cause of the problem, without success.).  I want to 
 > replace channel 36 in the original cube with the mean derived from the 2 days 
 > where the noise pattern was not present.  MCUBE seems to be the task I need.  
 > 
 > I've used COMB to produce a replacement "cube" comprising the desired 2-day mean 
 > for just channel 36.  The axis order is the same as in the original cube. The 
 > AIPS Cookbook, section 8.5.1, makes the replacement process seem 
 > straightforward.  "Explain MCUBE" states that when the output cube already 
 > exists, as in my case, the inputs AXREF, AX2REF, and NPOINTS, are ignored.  
 > 
 > I have set INSEQ and IN2SEQ in MCUBE to the sequence number of the replacement 
 > cube (I've tried giving it sequence numbers 1 and 36 when producing it), and 
 > have tried various values of DOALIGN: -2, -1, 0, +1.  In every case MCUBE 
 > reports "Axes similar and cannot build a 'SEQ.NUM.' axis." before dying of 
 > unnatural causes.    
 > 
 > My impression is that the software doesn't know I want it to replace channel 36 
 > in the original cube.  INSEQ apparently doesn't identify the channel number to 
 > be replaced, so MCUBE must look in the header to find that information.  
 > However, that hypothesis hasn't helped me fix the problem.
 > 

Send me the IMHEAD for the two inputs and the adverbs you chose.

Eric Greisen