[daip] clean component filtering

[Andy Biggs]

Hi Amy. Yes, the IMAGPRPRM(8,9) thing was some other problem; I tried to 
run a 'normal' IMAGR after I emailed you and it similarly hung.

My inputs for CCEDT are:

AIPS 2: CCEDT:    Select CC components in BOXes and above minimum flux.
AIPS 2: Adverbs     Values                 Comments
AIPS 2: ----------------------------------------------------------------
AIPS 2: USERID     2002                    User number.
AIPS 2: INNAME     'TEST'                  Input file (name).
AIPS 2: INCLASS    'ICL001'                Input file (class).
AIPS 2: INSEQ         1                    Input file (seq. #).
AIPS 2:                                       0 => high
AIPS 2: INDISK        1                    Disk unit #.       0 => any
AIPS 2: INVERS        1                    Input  file version no.
AIPS 2: OUTVERS       0                    Output file version.
AIPS 2: BCOUNT        1                    Beginning row to copy
AIPS 2: ECOUNT        0                    Last row to copy: 0 => end
AIPS 2: CUTOFF       0.05                  Minimum merged flux.  Use
AIPS 2:                                    large negative for no edits.
AIPS 2:                                    0.0 cuts off at zero.
AIPS 2: BOXFILE    ' '
AIPS 2:                                    Text file with Clean boxes
AIPS 2: NBOXES        1                    Number of boxes in CLBOX
AIPS 2: CLBOX         1           1        Clean windows to apply
AIPS 2:            1024        1024        *rest 0
AIPS 2: NCCBOX        0                    Number of boxes
AIPS 2:                                    < 0 -> write multiple CCs
AIPS 2: CCBOX      *all 0                  Four coordinate values for
AIPS 2:                                    each box in arcsec.
AIPS 2: CPARM         0.7         1.4      (1,2) = half size of rectang.
AIPS 2:            *rest 0                   in X,Y(arcsec) in which
AIPS 2:                                      to get total flux for
AIPS 2:                                      CUTOFF.
AIPS 2:                                    (1) < 0 => (2) is radius of
AIPS 2:                                      circular aperture in CUTOFF
AIPS 2:                                    Automatic model splitting:
AIPS 2:                                      3 = Max number of output
AIPS 2:                                      tables. >0 => automatic
AIPS 2:                                      model splitting enabled.
AIPS 2:                                      4 = Min fractional flux in
AIPS 2:                                      the output models. 0=>0.95.


In tests, I run this with gradually increasing values of CUTOFF, 
resulting in decreasing numbers of CC components in the output file. 
That's all perfectly fine, but at the point where all the CC components 
are rejected it reports:

localh> CCEDT2: Task CCEDT  (release of 31DEC08) begins
localh> CCEDT2: Merged    271 components to make    271
localh> CCEDT2: CEDTAB: NO COMPONENTS SELECTED
localh> CCEDT2: Purports to die of UNNATURAL causes
localh> CCEDT2: localhost    31DEC08 TST: Cpu=       0.0  Real=       0

I would say that this is the wrong behaviour i.e. it shouldn't 'die', 
but should at least write out a new CC table with no components in it. A 
user presumably doesn't want the original set of clean components if 
they've been rejected and if they don't like the empty file they can 
just delete it. What do you think?

By the way, my attempts to experiment with the CC filtering in IMAGR 
were somewhat thwarted, as I said, by some other unknown problem. I was 
wondering though, do IMAGRPRM(8,9) have the same meanings as CUTOFF and 
CPARM(1,2) in CCEDT i.e. does IMAGR sum all clean components within 
radius IMAGRPRM(9) and reject them if this sum is less than IMAGRPRM(8)? 
It's not terribly clear from the help which may have a typo:

(8) If non-zero, select only those Clean components having
                > ABS(IMAGRPRM(8)) Jy within a radius of IMAGRPRM(9)
                cells of the component.  If IMAGRPRM(8) < 0, the abs
                value of the flux near the component is used.

Is ABS(IMAGRPRM(8)) correct? The next sentence seems to say that ABS is 
only used when this parameter is -ve.

Thanks.

Andy

Amy Mioduszewski wrote:
> Hi Andy,
> 
> What are your inputs for CCEDT?
> 
> Also, I have used IMAGRPRM(8,9), and it worked, so maybe you didn't do 
> something quite right there, although it is hard to tell from you 
> description.
> 
> Amy
> 
> Andy Biggs wrote:
>> Hi there. I'm having a difficult time at the moment with some map 
>> making. I have a lot of maps to make and I want to do it in as 
>> *automated* a fashion as possible. I am doing a wide-field deep clean 
>> at 21cm and am using a fly's-eye faceting approach. The problem is 
>> that some of the facets do not contain any sources which results in 
>> them ending up with lots of weak, spurious clean components which I 
>> want to exclude from CALIB (this is before each epoch is combined 
>> together with STUFFR).
>>
>> My favoured way of dealing with this would be to run CCEDT. However, 
>> it seems that in those fields where all the clean components would be 
>> (correctly) rejected, CCEDT doesn't do what I would like it to, and 
>> which I think it should, which is to write out a new *empty* CC table.
>>
>> I've tried various other approaches, such as flagging the entries 
>> using TAFLG, but that caused CALIB to do some very strange things, 
>> mainly only using a small and somewhat random number of the many 
>> fields. I've tried running IMAGR with the filtering option 
>> (IMAGRPRM(8,9)) but this didn't seem to do anything at all i.e. IMAGR 
>> didn't even begin (nothing appeared in the MSGSRV and no activity was 
>> reported by 'top'). One thing that sort of worked was to set field 
>> cards as 'F -1 -1 0 0' in BOXFILE which prevented IMAGR placing a box 
>> in that field. However, you then need to tell IMAGR the size of the 
>> field through IMSIZE and you can only do this once i.e. if you have 
>> different sized fields (which I do) you can't really do this. Also, 
>> you need to know in advance which fields aren't going to have any 
>> sources.
>>
>> Anyway, I was wondering if you agree that CCEDT should write out an 
>> empty CC table if no components are selected? I think that that would 
>> help me a lot.
>>
>> Thanks,
>>
>> Andy
>>
>> p.s. I'm running a very recently midnight-jobbed version of 31DEC08 on 
>> a MacBook Pro.
>>

-- 

Andy Biggs
ESO, ALMA Regional Centre
Karl-Schwarzschild-Strasse 2
D-85748 Garching, Germany

tel. +49-89-32006471
fax. +49-89-32006898