[daip] [Fwd: cross check DFTPL-IMAGR]

[Eric Greisen]

Hanno Spreeuw wrote:
> Sorry, I just wanted to add that I did set CMETHOD='DFT' in IMAGR.
> 
That would not enter with a dirty image...

> -------- Original Message --------
> Subject: cross check DFTPL-IMAGR
> Date: Wed, 07 Jan 2009 15:42:36 +0100
> From: Hanno Spreeuw <J.N.Spreeuw at uva.nl>
> To: daip at nrao.edu
> 
> Hello, I am comparing the outputs of IMAGR and DFTPL for a residual
> dataset with one source in the center.
> DFTPL (SHIFT=0) sums the real parts of the visibilities of all
> baselines, according to the helpfile. I guess this means the mean, but
> is it the weighted mean? If so, are the weights used as is?
> 
> I selected the smallest possible time intervals (one sampling) and I
> used natural weighting in IMAGR and NITER=0. The central pixel values
> from the dirty images do not agree exactly with the output from DFTPL.
> Should there be an exact match between the central pixel values and the
> fluxes from DFTPL for some setting of adverbs in IMAGR? I tried
> UVWTFN='NO','NZ', 'NV' and 'NS'.

DFTPL goes through straight disk reads to get the UV data while IMAGR 
uses the calibration package via the OOP package.  In general one would 
not expect much difference, but the latter will apply flagging and data 
selection that may well be a bit different than that of DFTPL.

You should also check how recent your updates to 31DEC08 have been. 
There was a bug fix on Dec 9 affecting multi-channel data in DFTPL.

Eric Greisen