[BLAST_SHIFTS] Shift summary 07/12/2004 A (1-9)

From: Electronic Log Book (elog@blast05.lns.mit.edu)
Date: Mon Jul 12 2004 - 09:44:33 EDT


Operator: tancredi

beam polarization: 70.5+-0.7 / -66.7 +- 0.7 false asym: 1.8 +- 0.5
average beam polarization: 68.6 +- 0.5 %

Polarized ABS runs starting from sunday 07/11 at 11.20 am

        Runs 8773 8775-8782 8784-8791 8793 8795-8798

        fitted hPz : 0.61 +- 0.05 (Left,perp) 0.53 +- 0.05 (Right, para)
                Cut set1 |Mm-Mproton|<0.06 0.3<betaProton<0.8 + CCcut
                file: asym1.gif
        fitted hPz : 0.57 +- 0.04 (Left) 0.55 +- 0.04 (Right)
                Cut set2 |Pm|<0.2 |W-1.91| <0.1 no betaProton cut + CCcut
                file asym2.gif

In both case the L-R average target polarization is 82 +- 4 % +- 5 %
The first error is the fit (mostly statistics). The second error is *my* estimate
of the systematic/analysis-dependent error at this point (see below).
There is no montecarlo or model error.

Until all montecarlo issues are solved all polarization results are obtained
ONLY by comparison with a "point-calculation" of the asymmetry. Note while
we had some problems with the D2 montecarlo (inherently much more complex) , if I remember correctly the difference between point-calculation/montecarlo
was not a big effect (for BLAST) in hydrogen runs.

================

Shown in the next plots are the cuts, applied to the data :

I prefer cut set #2, but I just want to emphasize what systematic uncertainty to expect due to different choices. This is discussed below.
Note that with cuts #2 the L/R sector agree much better (and this statement has
a 70 % chance to be significant)

3rd attachment: (Mmass - Mproton) distribution (Mtgt= Mdeut, Mx = Mproton)

                black = all data with +1, -1 charges
                red = black + proton beta cut
                blue = black + |pmiss|<0.2 cut

4th attachment: Invmass distribution, colors as above

Note that Mm cuts of 0.06 (green) and 0.1 have been used as the default, with
a beta cut (i.e. acting on the red curve). An alternative is to cut pmiss<0.2
and select events on invariant mass W, independently of the beta detemination.
Note that the calculation of the pathlenght and hence beta had an error (bug) that only recently (june) has been discovered and tried to fix (thanks to chris)

For yaks, I also show the Z distribution. Note no explict z-cuts and all events
reconstruct nicely in the cell.

Finally, the phi-e- distributions for the L, R sector show a potential problem: it
is asymmetric which is likely to be a recon efficiency problem. However this becomes an analysis problem too since you can't rely on phi-symmetries and cancellations (e.g the tensor spin correlation parameter). Cutting in a narrow
+- 5 deg range should show no change.

However I do get a lower average polarization result (last attachment, asym3.gif)
I am forced to keep this result, which yields <Pz> of 75 +- 5 %

I think it safe to say that there is 5 % systematic uncertainty on top of any hPz
determination. Better statistics will help too.

P.S.
Program used: macros/Deuterium/show_deep_asym_V.C



This archive was generated by hypermail 2.1.2 : Mon Feb 24 2014 - 14:07:31 EST