Hello,
just a comment about the consistency between polarization results.
Aaron's results were 11.0+-6.0 (L) 10.0+-5.0 (R)
Chi's results were 17.8+-4.5 (L) 17.6+-9.7 (R) and 18.0 +- 4.3 combined
In either case the error is the uncertainty in the fitted Pzz parameter.
With the eep data we could also establish a very small value for a constant
false asymmetry offsets and I believe the binnning is optimal. Note that
the AedT should integrate to zero in case of smooth reconstruction
efficiency. This could to some extent also change day by day.
The polarization results for the two channels seem to be consistent.
However T20 is much larger than Ad(T) in nature. With these data set
the observed asymmetries were in the scale of 2% (eep) and 10 % (T20).
Clearly this shows that the analysis suffers greatly from a low FOM.
It is just so much harder to be more stringently conclusive with a 2%
signal. Our S/N will go with the square of the polarization. Already
40% would be significant and the goal is 70% !!. But I think you all see
the first steps of a consistent analysis. To which we could add the ee'n
channel once the NC are in.
-- ________________________________________________________________________________ Tancredi Botto, phone: +1-617-253-9204 mobile: +1-978-490-4124 research scientist MIT/Bates, 21 Manning Av Middleton MA, 01949 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^On Mon, 28 Jul 2003, richard milner wrote:
> Hi Chi, > Thanks for your complete report on the elastic analysis. Now that we have > results from both elastic and quasielastic, I am interested in > consistency. Does the quasielastic analysis of Aaron also see > significantly less yield as you do? I would like a number for Aaron's > target polarization. I am not sure `about 10%' is consistent with 18 +/- > 4 %. > Thanks, > Richard > > On Sun, 27 Jul 2003, zhangchi wrote: > > > > > Hi all, > > > > First: atatched daily_pol.eps is a plot of fitted polarization day by day. > > here s a brief explaination of what is plotted: > > > > tensor polarization is fitted from ed elastic asymmetry in 3 manners: > > > > 1. fitted from asymmetry in parallel kinematics only, i.e. left deuteron, > > right electron. asymmetry in this sector is large. > > > > 2. fitted from asymmetry in perpendicular kinematics only, left electron, > > right deuteron. Asymmetry on this side is small and more often than the > > other side, it is not distinguishalbe from 0. > > > > 3. fitted from the A_perpendicular - A_parrallel. Because both kinematics > > measure the same polarization, this could be a way to combine statistics > > and cancel out some overall shifts from 0. > > > > The fit is done over an function in form of: pol*(a*x + b*x*x + c*x*x*x) > > where pol is the parameter fitted and x being Q^2. The expression in the > > parenthesis is acquired by fitting asymmetry from Monte Carlo. Also see > > Aaron's earlier message. > > > > I did not include a constant term in the parameterization of asymmetry to > > force the curve to start from 0. > > > > with only two valid data point along Q^2 from 0 to 0.4, only a 1-parameter > > fit is possible. > > > > Now going to the figures: > > > > the x-axis marks the date. the first point(x=13) in fact include most of > > the runs on 13th and 14th. that was when we realize that we were using a > > obsolete target cycle script so most of the runs on 13th are tensor+ and > > then most on 14th are tensor-. > > > > the second point (x=14.5) includes all the runs from 13th to 15th, before > > cell tempreture control was back functional. > > > > starting from the 3rd point, x correspondes to the date in July. > > > > left, top, the red curve is tensor polarization from 1st fit > > method(parallel kinematics). there was a dip on 19th. and it shows an > > interesting trend. on 19th, polarization is minimal and then there seem to > > be a trend to recover. > > > > right top, the blue curve is tensor polarization from 2nd fit method > > (perpendicular kinematics). most of the time polarization is not > > distinguished from 0 and the curve fluctuates more. > > > > left bottum, the purple curve is tensor polarization from 3rd fit > > method(difference between to sector). it is more stable, has smaller fit > > error. a dip on 18th. > > > > right bottum, the bluish curve is yield with date. the yield is expressed > > in 10^16 atom/sec equivalent flow assuming 100% detection efficiency. > > > > so on average the yield is consistant with 0.65e16 atoms/sec target gas > > flow. if we believe our detectors have an overall efficiency of 50%, then > > ABS flow intensity can be deduced as ~1.3e16 atoms/sec. Just for a > > reference: 0.1sccm atomic gas flow or 0.05sccm molecular gas flow is > > equivalent to ~4.48e16 atoms/sec. > > > > Averaged over all days for polarization: > > parallel kinematics: 17.8+-4.5 % > > perpendicular kine: 17.6+-9.7 % > > combined: 18.0+-4.3 % > > > > atatched ed_asym_1490_1523.eps is 17th data. left top: data in parallel > > kinematics: please ignore the 1st and the last data points. red is theory > > curve with fitted polarization. green dashed is the 100% theory curve. > > right top: data in perpendicular kinematics. left bottum: data from both > > sides, red are theoretical curves scaled by polarization fitted from > > difference between two sectors. right bottum: > > Asym_perpendicular - Asym_parrallel, and a fit on it. > > > > if you look carefully at left bottum, you ll see two black curves behind > > the red. those are theory curves with polarization fitted from single > > sector data. looks like everything agrees with everything. that is why I > > call it thrilling. > > > > ************************************************************************ > > Second: analysis on unpol runs: > > flase asymmetry in acquired in form of equivalent target polarization. > > Yield is acquired by comparing total counts with prediction of Monte > > Carlo and is expressed in equivalent gas flow assuming 100% detection > > efficiency. > > > > 1.recent 0.1sccm runs: > > false asymmetey equivalent to target polarization: > > parallel kinematics: 1.9+-4.2 % > > perpendicular kine : 2.3+-10.1% > > combined: 2.0+-4.2 % > > very much all 0. > > > > Yield: equivalent gas flow: 7.67 e16 atoms/sec. 0.1sccm molecular flow > > equals to 8.96 atoms/sec. so the yield for ed-elastic is a little too > > high(85% of everything perfect). However, I d like to remind you that a > > 1sccm flow controler with 1% max flow error measures 0.1sccm flow with 10% > > error even if we ignore the fact that the 1% is calibrated with N2 gas. So > > we may have been flowing more gas in than we thought. This has to be > > checked with e'p channel yield. > > > > 2. earlier 0.05sccm runs: > > false asymmetry > > parallel kinematics: 5.3+- 2.0 % > > perpendicular kine : 13.8+-32.9 % > > combined: 4.3+-17.5 % > > no false asymmetry we may say. > > > > Yield: equivalent gas flow: 1.5 e16 atoms/sec. compare to 4.48 e16 > > atoms/sec which is what 0.05 sccm really should be. > > > > atatched ed_xs_1642_1713.eps shows cross section data with MonteCarlo(back > > curve). right sector forward(red cross markers) has efficiency problem. > > and it is easily verifiable in an nsed session that is the inefficiency > > lies in software: all chambers hit and produce segments but no valid fit > > merge because of bad fit quality. > > > > ********************************************************************* > > Third. I have a few things to remind data analyzors: > > > > somewhere between run 1323 and run 1353. left tof 0 delay was changed. it > > result in a shift of tdc_right - tdc_left spectra to the right by about > > 300 channels. This change probably fixed left 0 strobe problem and got rid > > of the double peak there. However any timing cut must be checked and > > adjusted accordingly. I suspect this change will change ep elastic timing > > cuts as it changed ed elastic timing cuts. > > > > I am still concerned about the absolute timing. In reconstruction, we must > > correct for time of flight of heavy particles. However, we need have a > > somewhat resonable absolute timing between left and right sectors inorder > > to compute proton time of flight in one sector from electron time of > > flight in the other sector. And anommalies in tof tdc spectra should be > > monitored and reported to experts timely. > > > > Well, this is a long email now. see you at Bates tomorrow. > > > > Chi > > > > >
This archive was generated by hypermail 2.1.2 : Mon Feb 24 2014 - 14:07:29 EST