[d2n-analysis-talk] SAMC and Acceptance Cut Studies

[Brad Sawatzky]

Hi David,

I ran the acceptance losses at each aperture by John LeRose (slide 4),
and they basically make sense.  Q1 rolls the distributions off at the
edges, the dipole entrance has a small effect, the dipole _exit_ has a
large effect, and Q3 has a fairly small effect -- all pretty consistent
with what you show.

Assuming reactz_gen is what we usually call z_targ (ie. the target
length that the HRS can see), then that looks OK too.  Nominal
acceptance is +-6 cm at 90deg, or about +-8.5cm at 45deg.

I don't understand the losses for very tight cuts shown on slide 6
though.  If you point a beam of particles right along the central axis
of the spectrometer, and with p = p_0, then 100% should pass through to
the focal plane....   Is this some straggling effect due to scatters
between the generation point and the Q1 vacuum window?  What happens if
everything is vacuum (no windows, no glass target cell, etc)?

If I understand how SIMC/SAMC works, then it is essentially a ray tracer
once the particle reaches Q1.  So "statistical errors" don't apply (at
least the won't follow a gaussian counting statistics).  If it's a
cross-section weighted random walk from the generation point, through
the target cell, air/He4, etc up to Q1, then you'll get some run-to-run
variation in what makes it because you're "rolling the dice" in
mean-free-path steps propagating the particle up to the Q1 entrance
window.  A simple counting-statistics uncertainty model still isn't
valid though.

I think what you should be doing is tightly constraining the generated
particle kinematic distribution (ie. the black lines on slide 4).
 - narrow those up so you're producing a mono-energetic beam pointing
   into the middle of Q1
 - disable (set to vacuum) the cell walls, air, etc so there is no
   scattering
Then every single particle had better make it to the center of the HRS
acceptance.  Then 'turn on' the cell walls, what happens.  Turn on the
air (he4), what happens?

I'm not sure I completely how the sigma_i is computed for the
cross-section weighting you describe on slide 11+, but the 'weighted'
distribution certainly show better agreement.  Is that something built
into SAMC, or does the input come from another program?

-- Brad