[Pafgbt] PAF beam spacing on the GBT for a pulsar survey

[Paul Demorest]

Rick,

At this level, I think it's kind of a matter of opinion.  I know we had a 
few emails expressing varying thoughts on the topic already ;)  However, 
survey speed is definitely the most commonly used metric for these 
comparisons.  We could always try it both ways, maximize survey speed vs 
maximize expected number of sources and see how different the answer is..

-Paul

On Tue, 13 Apr 2010, Rick Fisher wrote:

> Hi Paul,
>
> Do we want to maximize the integral of (G/T)^2, or should the pulsar 
> population as a function of flux density be factored in?  If 
> completeness is an issue, a steeper log(N)-log(S) curve would favor 
> closer spacings since more pulsars would be discovered near the 
> sensitivity limit.  This would be particularly true for new pulsars.
>
> I realize that PALFA uses interlaced pointings to fill in closer 
> spacings, but their "47 pointings to cover one square degree" still 
> implies -3 dB crossings (assuming 3.3 arcmin HPBW).
>
> Rick
>
> On Tue, 13 Apr 2010, Paul Demorest wrote:
>
>> Hi Rick,
>> 
>> There is a nice picture of the ALFA beam shape at the top of this page:
>> 
>> http://www2.naic.edu/alfa/gen_info/info_obs.shtml
>> 
>> maybe you've seen this already, but the beams are widely spaced, crossing 
>> at about -6dB.  They get -3dB spacing in the actual survey by filling in 
>> the holes with multiple pointings.
>> 
>> To determine an optimal beam spacing (for a given array+telescope) I think 
>> we want to maximize the integral of (G/T)^2 over the field of view.  That 
>> should result in the highest possible survey speed.  Then we should check 
>> the figures we used for the comparison with PMB and Efflesburg surveys and 
>> see if any claims need to be revised.
>> 
>> I'd guess the survey speed goes down by a factor of ~2 vs ideal 
>> fully-spaced beams, but maybe the optimization could make this only ~1.5 or 
>> so?
>> 
>> -Paul
>> 
>> On Tue, 13 Apr 2010, Rick Fisher wrote:
>>
>>>  In writing an MRI proposal for the construction of a PAF for the GBT 
>>> we've
>>>  run into a conundrum that we should have anticipated much earlier. Any
>>>  thoughts on the following will be appreciated.
>>>
>>>  The problem is that the "plate scale" (linear feed offset distance per
>>>  angular beam offset in HPBWs) is roughly twice as large on the GBT as it
>>>  is on the 20-meter telescope where we have done our PAF tests.  Hence, 
>>> our
>>>  19-element array can accommodate only half as many HPBW offsets as the
>>>  20-meter before the focal spot loses significant power off the edge of 
>>> the
>>>  array.  The attached plot calculated by Karl Warnick shows
>>>  Tsys/aperture_efficiency as a function of beam offset for the GBT. Going
>>>  to 37 elements (or even more) is something that we clearly want to do in
>>>  the long run, but it's a big step up in all aspects of the array system
>>>  (Dewar size, number of receiver channels, real-time beamformer size,
>>>  etc.).
>>>
>>>  I am wondering if the best strategy for this proposal will be to stay 
>>> with
>>>  the 19-element array as the next logical step and to form 7 beams on the
>>>  GBT with the 6 outer beams spaced about 0.6 HPBW from the center beam.
>>>  This is essentially Nyquist spacing, but it is different from the 
>>> strategy
>>>  used in the Arecibo PALFA survey.
>>>
>>>  The PALFA web site says that they are using 47, 7-beam pointings to cover
>>>  one square degree of sky.  This would imply that their beams cross at
>>>  about the 3-dB level, which means that most of the sky is covered with
>>>  sensitivity considerably below peak beam sensitivity.  If we use 0.5 or
>>>  0.6 HPBW spacing we won't cover as much sky in beam areas, but the
>>>  relative average sensitivity within this area will be considerably 
>>> higher.
>>>  Does this sound like a reasonable enough trade-off to justify putting a
>>>  19-element array on the GBT as the first science instrument?
>>>
>>>  Rick
>> 
>> 
>