Hi John,
We did look into this. If I remember correctly:
1) I think the Wandelt algorithms rely on precalculating maps folded with each
"beam mode" - with frightening storage and memory implications. Presumably
they need to be full sky rather than the cutdowns we normally use.
2) As you know the way the pipeline works is to perform an interpolation off
the input map per trajectory timestep and per detector. The computational
expense of including in this operation on-the-fly convolution of an arbitrary
beam shape with arbitrary rotation angle on the sky presumably scales with the
number of modes necessary to describe the beam to the desired precision - even
if we are just adding up precalculated numbers it will be a high cost.
Many approximations are possible to speed up but they will all involve
fundamental changes to the basic mode of operation (like precalculations for a
limited set of angles on the sky etc). For the specific purpose of
approximating the response to far sidelobes such approximations may of course
be acceptable.
Clem
--
**********************************************************************
Clem Pryke - Associate Professor - Physics
University of Minnesota,
Room 313 Tate, 116 Church Street S.E. Minneapolis, MN, 55455
Tel: 612-624-7578 Fax: 612-624-4578 email: pryke(a)physics.umn.edu
**********************************************************************
Show replies by date