[daip] forwarded message from Neal Miller

Wed Aug 15 11:50:11 EDT 2007

------- start of forwarded message (RFC 934 encapsulation) -------
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII; format=flowed
Return-Path: <nmiller at skysrv.pha.jhu.edu>
Received: from polaris.cv.nrao.edu (polaris.cv.nrao.edu [192.33.115.101])
	by dropbox.aoc.nrao.edu (8.13.1/8.13.1/smtp-gateway) with ESMTP id l7FFhwps016657
	for <egreisen at aoc.nrao.edu>; Wed, 15 Aug 2007 09:43:59 -0600
Received: from revere.aoc.nrao.edu (revere.aoc.nrao.edu [146.88.1.15])
	by polaris.cv.nrao.edu (8.13.1/8.13.1/smtp-gateway) with ESMTP id l7FFhv7B024440
	for <egreisen at nrao.edu>; Wed, 15 Aug 2007 11:43:58 -0400
Received: from ipex4.johnshopkins.edu (ipex4.johnshopkins.edu [128.220.161.141])
	by revere.aoc.nrao.edu (8.13.1/8.13.1/cv-ws-8.12) with ESMTP id l7FFhPNh004546
	for <egreisen at nrao.edu>; Wed, 15 Aug 2007 09:43:41 -0600
X-IronPort-Anti-Spam-Filtered: true
X-IronPort-Anti-Spam-Result: Ao8CAB+8wkaA3OmI/2dsb2JhbAA
X-IronPort-AV: E=Sophos;i="4.19,267,1183348800"; 
   d="scan'208";a="12249556"
Received: from narya.pha.jhu.edu ([128.220.233.136])
  by ipex4.johnshopkins.edu with ESMTP/TLS/DHE-RSA-AES256-SHA; 15 Aug 2007 11:43:24 -0400
Received: from localhost (nmiller at localhost)
	by narya.pha.jhu.edu (8.13.1/8.12.11/Submit) with ESMTP id l7FFhNHx016826
	for <egreisen at nrao.edu>; Wed, 15 Aug 2007 11:43:23 -0400
In-Reply-To: <18115.4958.182280.739087 at primate.aoc.nrao.edu>
Message-ID: <Pine.LNX.4.62.0708151103050.8595 at narya.pha.jhu.edu>
References: <Pine.LNX.4.62.0708141429040.24285 at narya.pha.jhu.edu>
 <18115.4958.182280.739087 at primate.aoc.nrao.edu>
X-MailScanner-Information: Please contact the postmaster at aoc.nrao.edu for more information
X-MailScanner: Found to be clean
X-MailScanner-SpamCheck: not spam, SpamAssassin (not cached, score=0,
	required 5, autolearn=disabled)
X-MailScanner-From: nmiller at skysrv.pha.jhu.edu
X-Spam-Status: No
From: Neal Miller <nmiller at skysrv.pha.jhu.edu>
To: Eric Greisen <egreisen at nrao.edu>
Subject: Re: install help
Date: Wed, 15 Aug 2007 11:43:23 -0400 (EDT)

Hi Eric.

I have been running VLACALIB, however it does not appear to
have changed in the time between my two aips installations;
in each case, it is using the 3C48 model for the primary flux
calibrations and I've confirmed that the model image itself
is unchanged.

Here's a rundown of what I've done, along with some numbers
for each of the installations. I've done the same procedure
on each installation rather than transfer files back and
forth. Doing them side-by-side, I make sure the adverbs are
all set identically.

1) FILLM. I've set doweight=10, douvcomp=-1, cparm 0 3 16
    (with EVLA conversion, Rick has suggested setting these
    cparms and manually flagging data)
2) TVFLG on each source within the multisource data. The
    day of data in question are pretty nice, so the flagging
    has amounted to ~10-15 seconds at the start of each scan
    of my primary calibrator, removal of an antenna here and
    there, and the usual occasional bad "pixel".
3) SPLIT out the calibrator (apply flags, but not any calibration).
    Run CALIB on it to do a self-cal (phase only, use smodel 1 0
    to model it as a 1 Jy point source, uvrange = 15 9999 as per
    VLA Calibrator manual). Run BPASS on the self cal'd data,
    copy the resulting BP table over to the multisource data
    (after inspecting with POSSM).
4) SETJY, calculating the flux for 3C48 (optype=calc)
5) VLACALIB, first on 3C48 (use model, so antennas, uvrange,
    etc. set to zero). VLACALIB has doband=1 and uses the BP
    table. A second run of VLACALIB is done on the calibrator,
    with uvrange restricted as indicated.
6) GETJY for calibrator, referenced to 3C48.
    -> New computer, 31DEC07 aips, August 3 install:
       IF 1  0.86611 +/- 0.00471
          2  0.87004 +/- 0.00426
    -> Old computer, 31DEC06 aips, July 19 2006 install:
       IF 1  0.84248 +/- 0.00518
          2  0.84527 +/- 0.00439
7) VLACLCAL
8) SPLIT off data (docal=2, doband=1, douvcomp=-1)
9) UVFIX with default params
10) IMAGR. For same adverb settings, the rms of the
    central facet is:
    -> New computer, 31DEC07 aips, August 3 install: 47.8
    -> Old computer, 31DEC06 aips, July 19 install:  43.3
11) CALIB self cal data. Use images from all imaged facets
     as in2, calibrate on phase. Identical adverbs set for
     each installation.
12) IMAGR on self cal data. The rms for the central facet
     is (from IMEAN, rms from peak in histogram):
    -> New computer, 31DEC07 aips, August 3 install: 33.6
    -> Old computer, 31DEC06 aips, July 19 install:  29.9
    (there's a fairly bright source in one of the facets
    adjacent to the central facet; self calibration helps
    a lot)
13) UVSUB. Remove CC's from all imaged facets from the self
    cal data. UVMTH, in and in2 both set to the UVSUB data,
    opcode=sub. This step is a way to handle the slight
    correlator offset errors of this multichannel continuum
    L band program. Use IMAGR to make a dirty image of just
    the central facet for each the UVSUB and UVMTH data,
    confirm that the "double V" pattern at the phase center
    is removed. The rms for the UVMTH data in this central
    facet is:
    -> New computer, 31DEC07 aips, August 3 install: 30.2
    -> Old computer, 31DEC06 aips, July 19 install:  27.1
    UVSUB to return CC's from cleaned images to the UVMTH data.

If CALIB hasn't changed, I'm running out of possible differences.
I have seen that the tables format seems to have changed, such
that attempting to save out fits tables created on one
installation and TACOPing them over to loaded data on the
other installation doesn't produce the intended result. I
have found that "final" uv data - such as that I produce after
step 13 above - can be written to file, loaded into the other
installation, and imaged to yield the same results as were
obtained in the initial installation of aips. I've also done
the simple check of applying the flags created in TVFLG and
inspecting the data again with TVFLG to confirm that they were
applied (both directly in the multisource data, and in data
that I've SPLIT out and had SPLIT apply the flags).

I've got 31DEC04 installed on the older machine, and am now
going through the same procedure on it to see how things
compare.

Thanks for your time.

Take care,
Neal

On Wed, 15 Aug 2007, Eric Greisen wrote:

> You do not say precisely what you are doing so it is hard to make many
> comments.
>
> Some things have changed - if you use VLACALIB you will find that it
> does different things.  It now uses a model for 3C286 and other
> primary flux calibrators which should make some change to the fluxes
> you find for other sources.  In general they should be better.
>
> CALIB itself has not changed and IMAGR is its simpler modes has not
> really changed.  I must admit to being somewhat stumped by this at
> least wiothout more information.
>
> Eric Greisen
>
------- end -------