[d2n-analysis-talk] Gas Cerenkov Cut Efficiencies (Background Corrected)

[Brad Sawatzky]

On Thu, 11 Mar 2010, David Flay wrote:

> I've attached the gas Cerenkov cut efficiencies as they stood before
> corrections for the first two plots.  I've fit the pion background
> from my selected electron sample (see 2/18/10 talk:
> http://www.jlab.org/~flay/analysis_2010.html) an subtracted it off.
> The last two plots show the corrected gas Cerenkov efficiencies.
> 
> The correction brings all efficiencies to better than 95% across most
> of the cut range up to ~500 channels (2.5 photoelectrons).
> 
> All the values are much more consistent with each other, but across
> the full kinematic range, the variation is about 1-2%.

Those look pretty reasonable to me.  There's still a small systematic
effect as you note, but it's not too bad.

> The correction at the low end of the kinematic range corrects the
> efficiency by up to ~15-20%, and at the highest momenta, the
> correction is on the order of 1-2%.  Most of the (statistical) error
> bars are ~1-2% -- I have not added more runs to my previous studies
> (in other words, I have chosen runs that have one particular target
> configuration -- transverse-270, as an example -- I did not mix target
> configurations.)
> 
> The high momenta data have the largest error bars (in particular p =
> 1.70 GeV) as we have a very short list of runs to use (I am currently
> using all of them).  

Sanity check the run-list.  We spent most of our time running at the
highest momenta.  The total statistics should be comparable to the lower
momentum runs (this was by design).  Your "short list of runs" statement
doesn't make sense...

Note that you can bin all runs for a given beam energy + LHRS momentum
setting together in order to parametrize the shape of the pion
contamination (the LHRS just sees a pion energy distribution, and that
energy distribution doesn't care about polarization).  You then fit that
shape to the residual pions for each unique kinematic to get the
associated contamination (grouping runs by polarization state too for
this bit).

> Also, the background is very very difficult to fit at the highest
> momenta (p = 1.51,1.60,1.70 GeV).  This is the region where the
> contamination is on the order of ~2%.  There is barely any structure
> to the background distribution, and I don't know if it's terribly
> useful to even fit this distribution -- any suggestions?

The better question is what is the error on your fit/subtraction?  That
is, you say your total subtraction is 2% -- a pretty small number.  Say
your relative uncertainty on that 0.02 correction is 50%.  That sounds
pretty bad, but in the end it is only a 1% absolute contribution to the
error budget and that's what matters.

-- Brad

-- 
Brad Sawatzky, PhD <brads at jlab.org>  -<>-  Jefferson Lab / Hall C / C111
  Ph: 757-269-5947 -<>- Pager: 757-584-5947 -<>- Fax: 757-269-5235
The most exciting phrase to hear in science, the one that heralds new
  discoveries, is not "Eureka!" but "That's funny..."   -- Isaac Asimov