Run 08 summary

From RHIC Spin Group
Jump to navigationJump to search

Last modified by Dmitri Smirnov on 2-12-2011

2008 Polarizations for RHIC

Final polarization along with absolute stat. and syst. uncertainties in each physics fill are provided in

Pol2008_blue.dat - Blue

Pol2008_yellow.dat - Yellow

(Unlike in previous years, we provide the same tables for PHENIX and STAR, due to negligible difference in polarization for colliding bunches in PHENIX and STAR IR)

The supporting presentations are

Media:RSC_12sep2008_vipuli.ppt

Media:RSC_12sep2008_hiromi.pdf



All uncertainties below are relative (Delta P/P or Delta A/A). 

Throughout the 2008 run the hydrogen jet target polarimeter (HJet) was used, 
alternating blue and yellow beam on the jet (two beam mode measurements were 
not used for normalization of pC). Generally a measurement with 
one beam extends across many fills (a period). Totally, we obtained 4 periods 
with blue and 3 periods with yellow beams for 100 GeV running. 
The statistical uncertainty (deltaP/P) in Run8 was ~10% per fill 
in HJet measurements. The whole collected Run8 HJet statistics allowed 
to provide pC to HJet normalization with stat. uncertainties 2.7% in blue 
and 3.2% in yellow (which includes stat. uncertainties from pC measurements for 
these fills). 
The systematic uncertainty of HJet measurements 
includes uncertainty on molecular hydrogen contamination in polarized atomic 
hydrogen jet (2% for deltaP/P) and an upper limit for the effect of "other" 
background estimated from the asymmetry in "non-signal" strips, 
1.3% for blue, 2.4% for yellow. Despite that backgound level increased about 
twice compared to previous Years (presumably due to removed collimators), 
it didn't show any asymmetry either relative to target spin pattern or 
beam spin pattern, so eventually didn't affect the polarization measurements, 
which is extracted from the ratio of beam over target asymmetries.

In 2008, the pC polarimeters used a target scan mode, when the measurements 
were performed with targets, stepping, in x (transverse 
horizontal coordinate) or in y (transverse vertical coordinate) across the beam, 
with equal measurement time at each step. In the later part of Run8 we used 
continuous target move instead of step-wise. 
It allowed to measure either horizontal or vertical polarization profile in each 
run separately (though with limited stat. precision). 
About half of all blue fills had measurements of vertical polarization profile, 
another half had horizontal profile. Due to horizontal target problems in yellow 
pC polarimeter, no vertical profile measurements are available in yellow. 

The strategy is to obtain the normalization for pC measurements using absolute 
polarization measurements with HJet in the fills for which HJet measurement is 
available, and after that use the properly normalized pC measurements to define 
the polarization in each physics fill.

On the first step of pC data analysis, two parameters, t0 and dead layer (DL), 
were extracted for each strip in each measurement (run) from the fit of the 
"banana" plot, the recoil Carbon time-of-flight (ToF) vs energy. The DL parameter 
carries the meaning of "effective" dead layer and is used to correct the carbon 
deposited energy to obtain carbon kinetic energy. t0 is a ToF offset.

After quality checks (QA), on the average each fill contained about ** "good" runs. 
The list of QA checks was the same as in Run5/6 pC analysis. It included control 
of the width and position of the carbon (C) mass peak, as well as C mass peak 
position vs its kinetic energy (which detects problems with WFD and/or DAQ and/or 
in the fit of "banana"; only a few strips were masked due to this QA); strip by 
strip variations of the number of events in the "banana" (** runs were removed 
from the analysis); consistency in bunch-by-bunch asymmetry measurements 
(** runs were removed from the analysis). All systematic uncertainties from 
the effects above were estimated to be negligible for the final fill-by-fill 
polarization measurements, except the energy correction effect (described by DL), 
which was defined to be 1.0% in blue and 0.7% in yellow, from fill-by-fill 
variatio of DL.
After QA checks, it was found that blue fills 9942-9948 didn't have any reliable 
measurements, mainly due to thick target (so very high event rate); 
the polarization values for these fill were taken from HJet measurements, 
corrected for average pol. profile, measured in other fills.

To obtain average polarization over the beam intensity distribution in the 
transverse plane, the knowledge on the polarization profile (polarization 
vs x and y in transverse plane) is necessary. The correction due to 
pol. profile  depends on the ratio of width of the beam intensity profile 
(sigma_I) and beam polarization profile (sigma_P). 
Similar to Run6 analysis it was obtained for each fill
from the fit P/Pmax=(I/Imax)^R, where R is (sigma_I/sigma_P)^2, 
Pmax and Imax are polarization and intensity (event rate) at beam maximum intensity.
The fit parmeters Pmax and R were then used to calculate the 
average beam polarization for fixed target mode (when doing normalization to 
HJet measurements) and for colliding beams: <P>=Pmax/sqrt(1+R) and 
<P>=Pmax/sqrt(1+R/2), correspondingly, for one dimensional case. 

To relate HJet measurements to pC measurements, only R parameter in one 
direction (vertical or horizontal) is necessary, because the carbon target 
automatically averages polarization in the other direction. 
The obtained normalization corrections were 1.062 for blue and 1.091 for yellow, 
with stat. uncertainties 2.7% and 3.2% correspondigly. 
The initial normalization was from Run4 results. 
The source of such a shift in normalization from 1 is not yet clear. 
No shift was observed in Run5 (normalization correction ~1.01), and larger shift 
was observed in Run6 (normalization correction ~1.15)
One reason could be an improper energy correction. 
An incorrect energy correction doesn't affect the average measurements in pC, 
due to normalization to HJet, but it may give a fill-to-fill relative 
effect due to drift in energy correction (DL), which was about 8mkg/cm^2 
(6mkg/cm^2) in yellow (blue) from the beginning to the end of Run8. 
So this +/-4mkg/cm^2 (+/-3mkg/cm^2) variation from 
the average over Run8 DL may introduce +/-2.4% (1.8%) 
relative effect on polarization. 
It was taken as an upper limit of global uncertainty due to DL drift.

After normalization for pC measurements is obtained, the last step is to 
provide polarization values for experiments which are averages obtained 
weighting with a product of two beam intensities in both x and y transverse 
dimensions. For the simple case when the transverse size (sigma_I) 
is about the same in yellow and blue beams: <P>=Pmax_2/sqrt(1+R_x/2)/sqrt(1+R_y/2), 
where R_x and R_y are (sigma_I/sigma_P)^2 in horizontal and vertical 
direction respectively, and Pmax_2 - is polarization at the intensity peak 
in two dimensional transverse plane, which is  equal to Pmax*sqrt(1+R), 
where Pmax is polarization at the intensity peak in one dimensional case 
(integrated over the perpendicular direction; this is what we get from pC). 
So the knowledge on pol. profile in both transverse directions is necessary. 
About half of blue fills had horizontal profile measurements, and another half - 
vertical profile. Both profiles showed similar average R and similar variations 
from fill to fill: 0.15+/-0.11. This variation lead to ~3% fill-by-fill 
polarization uncertainty due to that only one profile measurements were 
actually available in each fill. 0.5% pol. uncertainty was assigned to global 
error due to that we effectively propogate measurements of horizontal or 
vertical profile measured in one half of fills to another half. 
Yellow measurements had only horiz. profile measurements, which was considerably 
sharper than in blue: R=0.30+/-0.10 (0.10 is variation from fill to fill). 
Having no clue about the other profile, we assumed it to be any value from 
0 to 0.30+0.10=0.40 => 0.20+/-0.20. This large variation (+/-0.20) 
led to 5% fill-by-fill uncertainty. And the uncertainty on the average R, which 
was also assumed to be +/-0.20, led to 5% global unceratainty. 

Below is a summary of systematic uncertainties for fill-by-fill (non-correlated) 
measurements, discussed above.

From non-measured profile: 3% in blue and 5% in yellow
(Blue fills 9942-9948 have additional uncertainty 3% 
due to assumption for the other profile)
Energy correction: 1.0% in blue and 0.7% in yellow

Below is a summary of global systematic uncertainties 
(considered as correlated from fill to fill), discussed above.

                                          blue       yell  
Jet normalization, stat:                  2.7%       3.2%  
Jet normalization (profile):              1.1%       same
Jet normalization, syst (molecular):      2.0%       same  
Jet normalization, syst (other):          1.3%       2.4%
Pol. profile (vert. for exp):             0.5%       5.0%  
Energy correction:                        1.8%       2.4%  

So the final global uncertainties, deltaP/P, are:

Blue:       4.2%
Yellow:     7.2%

Considering that "Jet normalization, syst." as well as "Energy correction" 
uncertainties are mostly correlated between blue and yellow, 
the final global uncertainties for a product of two beams, 

delta(P_B*P_Y)/(P_B*P_Y): 9.6%, 

which is 
sqrt{2.7^2+3.2^2+1.1^2+1.1^2+(2.0+2.0)^2+(1.3+2.4)^2+0.5^2+5.0^2+(1.8+2.4)^2}

Last modified by Dmitri Smirnov on 2-12-2011