Polarimetry discussions in ASAC: Making a fresh start
       =====================================================

                Johan Hamaker, ASTRON, DWINGELOO
                        February 6, 2001


 Past and present
 ----------------

	ASAC states some target figures for polarimetric performance but fails to
clearly indicate how these targets can be realised. There is nothing wrong with
the latter, because it is very much a technical matter that should be left to
others. Here is the situation as I see it. I hope I am well enough informed.

	Two important boundary conditions appear to be firm for engineering reasons:
   		
	1. The antennas will be altaz-mounted.
	
	2. The feeds will be linearly polarized and non-rotatable.
	
In addition we may expect that

	3. Polarization leakages will be larger than in existing instruments.
	
	4. Source polarization may be stronger.
	
This combination of conditions is without precedent. The >>capital question<<
is if and how one can do reliable polarimetry under them.

	Discussions in ASAC have revolved mostly about the alternative question
whether the design could be changed for polarimetry's sake. My reading is that
the answer is effectively 'no'. So we want an answer to my capital question.
This requires an expertise different from that in ASAC.

	At least a number of ASAC members must be more or less familiar with the
traditional approximate linearised polarimetry equations; but the problem in
its entirety is hardly if at all tractable in that framework. Pecking at it
from various angles fails to produce a good overall picture, let alone a
solution. The repeated inconclusive discussions in ASAC demonstrate this.


 Opening up a new front by staging a matrix attack
 -------------------------------------------------

	I contend that the only way to get to grips ALMA polarimetry is by
abandoning the piecemeal quasi-linear approach and mounting a frontal
matrix-based attack instead. In the matrix framework I can see the contours of
a solution that I will be able to write down and sdiscuss in a precise form.
But to critically review it, others will have to familiarise themselves with
the matrix formalism.

	It is certainly not realistic to expect ASAC as a whole to do this, and it
would be equally unrealistic to ask them to believe me alone on my word.


 Of course: a commission!
 ------------------------

        To recruit outside help, I propose that ASAC move to intitute a small
task group. I would be happy to see in it at least two sparring partners who
have between them

	1. some hands-on experience in polarimetric aperture synthesis and

	2. enough mental flexibility and mathematical inclination to invest in
familiarising themselves with the matrix approach. (This is really not a very
big deal, but I find that many are unwilling to take the trouble.)

I hope ASAC members can identify suitable candidates in their own environments.

        The group could be charged with

	a. Assessing the suitability of the matrix formalism as framework for
discussing polarimetry.

	b. Working out the basic method for solving the selfcal problem in ALMA and
assess the likelihood that it can be made to work.

	c. Indicate how the variation of instrumental polarization over the primary
beam may be handled.

	d. Recommend changes to the ALMA design that would be worth pursuing, such as
rotating part of the feeds over 45 deg.

	e. Reporting in writing at least one month before the next ASAC plenary
meeting and attend that meeting for a discussion.


 Results NOT to be expected
 --------------------------

	I must emphasise that this proposed study can do no more than generate
confidence that ALMA as planned can be made to work as a decent polarimeter.
The final proof will have to come from practice (or perhaps from an extensive
simulation).

        Apart, perhaps, from d above, I do not expect any result that
would call for a major change in the design. Within the boundary conditions
listed, ASAC has already indentified design constraints that are crucial, such
as stability and minimal beam squint, and these are well understood by the
engineers.