[alma-config] Whats up.

[John Conway]

Hi,

 I'm sorry that I havn't yet completed a provisional design 
to submit to the imaging competition as I promised at the last 
telecon. This week I have been 100% occupied with a large VLBI
conference here at Onsala which I've been organising and the 
previous week preparing for a visiting committee, which is 
assessing Swedish astronomy. Anyway although, I am close I
havn't yet finished the design. Tomorrow I am going on vacation 
for a week, since I suspect there is only a couple of days 
of work left to finsih the design I hope to post something soon 
after I return.

In fact the design I'm working on is something of a hybrid. I'm
using the same outer ring of 32 pads as in the NRAO strawperson 
design (no point duplicating finding all those pad positions in the 
terrain). The inner 64 pad positions are also the same as in the NRAO
strawperson and hence my E-array is identical to the NRAO 
design, and my equivalent of D arrays has a maximum number of 
short spacings from antennas on the 150m diameter ring.
Howwever for intermediate arrays the pads are distributed in 
a self-similar  pattern based on a spiral pattern (though 
of course withought drawing the spriral 'construction lines'
thats not easily apparent looking at the pad positions). This
allows of course a flexibilty in the number of and resolution interval
between configurations. Arrays for looking at high North or South dec
are accomodated by filling all pads within an ellipse of axial ratio
2:1 rather than within a circle. The uv coverage is quite good
in these hybrid arrays. If one decided to operate the zoom 
array as distinct 5 arrays then one could have B, then B/C and 
C arrays with 16 moves between each array. These moves could
be done either gradually (1 or 2 moves per day) or in one 
jump whichever is most practical. 


I have made also progress in implementing a version of the Kogan sidelobe
minimisation algorithm into my own software. Instead of minsimising 
over one peak sidelobe it minismises simulataneously over many, which
helps it converge faster.
I have incorporated terrain constraints and am now running it on the 
basic design to get a final pad distribution, the main problem right now
is not the spiral pattern out to 3km (there is a degree of freedom
in the rotataion of the pattern which allows one to avoid 
most terrain problems) but in interfacing to the E-array.

Some results of optimising a spiral without terrain constraints
can be found at my web site at 
http://www.oso.chalmers.se/~jconway/ALMA/SIMULATIONS/SIM3/
The 20 beam sidelobes for snapshots and long tracks are about 3.8% and
about 1.5%  respectively. There is a -1.6% plateau in the centre, this
could  be removed by adding more short spacings to uv coverage  (this will
increase a little the far sidelobes of course, and give rise
to a few isolated -1.6% sidelobes further out), I've yet to
do this however.

  John.

P.S I will be back in the saddle a week from now. I really want to
submit something a day right then,I have lots of work to do for
lecturing etc which starts end of August so I want to produce 
something as urgently as possible. Happy July 4th
holidays for those in the US.

P.P.S Looking at the plots on the webage again
I'm not sure that the beam grayscales  for the optimised
spirals have the same dimensions as the non-optimised spirals 
and donuts, they appear to cover an area about 1/4 in size.
This is why I think the rms looks larger, I'll check this when 
I come back.