PVDIS analysis and meeting minutes

01/17/2013 01/24/2013 01/31/2013 02/07/2013

07/10/201207/17/201207/24/201210/08/201212/13/2012 01/10/2013
04/05/2012 04/13/2012 06/12/2012 06/19/2012 06/26/2012 07/03/2012
01/13/201201/20/201203/02/201203/15/201203/22/201203/29/2012
12/02/201112/09/201101/06/201201/09/2012(theory)01/10/2012(theory)01/13/2012(theory)
10/13/201110/21/201110/28/201111/04/201111/11/201111/18/2011
07/25/201109/02/201109/16/201109/23/2011 09/29/2011 10/07/2011
05/31/2011 06/13/2011 06/20/2011 07/05/2011 07/11/2011 07/18/2011
03/21/2011 03/28/2011 04/04/2011 05/02/2011 05/09/2011 05/23/2011
01/31/2011 02/07/2011 02/14/2011 02/28/2011 03/07/2011 03/14/2011
11/19/2010 12/03/2010 12/17/2010 01/03/2011 01/10/2011 01/24/2011
10/08/201010/18/201010/22/201010/29/201011/05/201011/15/2010
08/30/201009/03/201009/10/201009/17/201009/24/201010/01/2010
05/19/201005/26/201006/23/201006/30/201007/14/201007/21/2010
03/29/2010 (M)04/07/2010 (W)04/14/2010 (W) 04/21/2010 (W)04/28/2010 (W)05/05/2010
01/21/201001/12/201012/23/200903/08/2010 (M) 03/12/2010 (F) 03/22/2010 (M)


  • 02/07/2013 Meeting to discuss paper writing:
  1. Remote participants: Robert Michaels, Diancheng Wang, Kai Pan, Xiaochao Zheng
  2. NIM paper is near completion after one week of circulating. Bob will send it through the JLab approval system, then XZ will submit it to arXiv and NIM.
  3. Kai completed the analytic analysis for OR deadtime, see ELOG #154. This combined with Diancheng's deadtime analysis will be used to analyze the uncertainty of the OR fraction of the deadtime.
  4. Diancheng in the updated ELOG #157 (file resonance.pdf):
    1. Except for RES#7b, completed the deadtime analysis, including uncertainties, both at 100uA and the run-by-run values; Also including all group deadtime for resonances.
    2. PID correction for RES#7 requires 2D PID efficiency table to pin-point the missing block.
    3. Due to the same PID problem, the radiative correction for RES#7 is pending.
    4. For other kinematics's radiative corrections, need to evaluate uncertainty. We sort of agreed on a conservatie estimate: when there is only one table available (MG's), we can use a 20% uncertainty since MG claimed his model is okay to 15-20% level.  (HL quoted similar "reliability" for his table). When there are two tables available (MG + HL), we can add 20% in quadruture to the difference between corrections using these two tables.
  5. Here is the updated list_of_things_20130207.
   
  • 01/31/2013 Meeting to discuss paper writing: We had the meeting but no minutes
    • 01/24/2013 Meeting to discuss paper writing:
    1. Remote participants:  Diancheng Wang, Kai Pan, Robert Michaels, Paul Reimer, Xiaochao Zheng
    2. Update during the week:
      1. We received calculation from M.G. on the resonance kinematics. See this directory. Perhaps Diancheng can add some curves to the group asymmetry plot (for both MG's and HL's calculations)?
    3. Kai has completed the contamination analysis, see ELOG #159
    4. Diancheng posted his updated in this PDF file (also attached in ELOG #157). The focus of last week was to get the group-dependent simulation working. Also have results on the group-wise Q2 and W (both data and HAMC, although HAMC needs to be tuned for PID of RES#7).
    5. Updated list_of_things_20130124.
    • 01/17/2013 Meeting to discuss paper writing:
    1. Remote participants:  Diancheng Wang, Kai Pan, Robert Michaels, Xiaochao Zheng
    2. Update during the week:
      1. Diancheng made the plots of asymmetry/Q^2 vs. W, with all preliminary corrections applied. Xiaochao sent these plots to the 3 theory groups on 1/14. Very positive response from W. M., neutral response from the other two.
      2. Toru sent in his A/Q2 vs. W plot from the Delta model, see this directory.
    3. Diancheng posted his updated in this PDF file (also attached in ELOG #157). Xiaochao will implement the completed items in all writeups.
    4. Kai posted ELOG #159 that contained the nearly-complete contamination results for narrow path. Todos:
      1. Need to include systematic uncertainties due to group deadtime uncertainties for f1 and f2.
      2. Need to apply the same method/code to wide path. Will use DIS group deadtime results throughout. (64\pm 6ns, 1xx\pm 8ns?)
      3. Also need to include all DIS contamination results (never did the f1 f2 comparison before).
      4. Then will proceed to other tasks in the previous todo list;
    5. Kai also sent PID tables to Diancheng, for calculation of the run-by-run PID corrections for all resonances.
    6. Bob mentioned Phys. Rev. C 69, 065501 (2004) to be the reference for the HAPPEX code nucleon form factors. This was copied to HAMC for quasi-elastic calculations for PVDIS. Eq.(42) was for quasi-elastic scattering calculations. The beginning of Section VII provided all references for the nucleon electromagnetic form factors.
    7. Updated list_of_things_20130117.
    • 01/10/2013 Meeting to discuss paper writing:
    1. Remote participants:  Paul Reimer, Robert Michales, Diancheng Wang, Kai Pan, Xiaochao Zheng;
    2. Immediate to-do's:
      1. Diancheng will make a plot similar to Fig.11 of resonance.pdf, and Xiaochao will send those plots to theorists to see if it's worth putting the plot in the resonance paper. It will also give the theorists an head-up for requesting curves;
      2. To do: Here is the SOM/NIM/RES analysis todo list, adopted from the 2012/12/13 version but with completed items removed.
    3. Diancheng sent in his analysis updated in 3 PDF files, see today's meeting directory: asymmetries.pdf, questiontodo.pdf, resonance.pdf. Questions and discussions after the first glance of these documents:
      1. Radiative corrections for RES3 is large, about 13% or one sigma. This might be a good news for both DIS and RES paper.
      2. For the group asymmetry results at the resonances, if we plan to publish these results (see Fig.11 on last page of resonance.pdf), what corrections are group-specific and will require more time? Deadtime, pion contamination(?), radiative corrections, transverse asymmetry (require average out-of-plane angle of each group)...?  On the other hand, the smooth appearance of the plot might help to make the statement such as "duality stands for PVES at the 20% level...".
    4. Before the X'mas break we have contacted 3 theory groups to discuss about co-authoring the resonance paper. Summary of responses:
      1. Received results on RES3 from H. Lee: see this directory. Explanation of all files:
        1. f25-v2 : the results using theta and ep calculated by using your eb, w2, and q2
        2. to examine the W-dependence of asymmetry in the region close to your W in RES3 and make sure that the table is used correctly:
              a. f25-v3: the results for W2=[1.55-1.65] at Q2=0.95, ep=4.867
              b.asym-memo.pdf: Toru's old note (sent to you before) on how to use our output to get asymmetry
              c. asym.ps: asymmetry calculated using Eq.(1) for n and p, and Eq.(7) for the deuteron.
      2. positive response from both Wally M. and Misha G, but no calculations yet.
    5. Xiaochao received from J. Erler on 2013/1/4 a review paper on "The Weak Neutral Current" (draft to be circulated by email only), where he "went into some details on the radiative corrections (see in particular Sec. 5.4). In Eqs.~(113)-(115) included simple formulas to extractthe Q^2-dependence". He is asking for comments.
    6. Xiaochao received an invitation to present PVDIS at the APS Topical Group on Hadronic Physics (GHP) workshop, in Denver, from April 10-12, just preceding the APS April meeting. Who should go?
    • 12/13/2012 Meeting to discuss paper writing:
    1. Remote participants:  Paul Reimer, Robert Michales, Diancheng Wang, Kai Pan, Xiaochao Zheng;
    2. Over the Thanksgiving week, contributions from everyone (thanks!) were put together and worked into a single document in the following 2 weeks. V1.0 of the Science SOM has been circulated by email;
    3. A todo list can be found here: list_of_things_20121213
    4. Per request from theorists working on the box diagram correction for Qweak, we agreed to work on a short publication on the resonance data. Bob had a first draft (being circulated by email only), but mostly we need to get the analysis done.
    • 10/08/2012 Meeting to discuss Science SOM (Supplemental Online Material) writing:
    1. Remote participants:  Paul Reimer, Robert Michales, Diancheng Wang, Kai Pan, Xiaochao Zheng;
    2. For our main results we will try Science. It is very important that each contributing author (all listed in the line above) read the whole page of Science contributor FAQ at http://www.sciencemag.org/site/feature/contribinfo/faq/index.xhtml, before attempting to write anything.
    3. The general guideline for how to prepare manuscript the first time can be found online, broken down to several levels:
      1. General guidelines for contributors: http://www.sciencemag.org/site/feature/contribinfo/prep/index.xhtml;
      2. How to prepare Science manuscript the first time: http://www.sciencemag.org/site/feature/contribinfo/prep/prep_init.xhtml;
      3. How to prepare supplemental material: http://www.sciencemag.org/site/feature/contribinfo/prep/prep_online.xhtml;
      4. File size should not exceed 10MB, as described here;
      5. A WORD template is provided, but we will go for Latex, which should follow these Latex instructions.
        1. Note that some latex editing formats, such as eqnarray and tabular, are forbidden.
        2. All filenames should be kept 8 characters or less.
        3. Templates can be downloaded from the bottom of the above page.
        4. Although the initial submission will only need the generated PDF file, we might want to observe these latex editing rules in the 20% chance that we are asked to submit a revised manuscript.
    4. A revised outline for the SOM, as we discussed today, can be found below along with assigned person's name and suggested number of pages. Note that
      1. when two names are assigned to the same section, the first appearing name is the one in charge for that section and the 2nd appearing name is the "helper" who will provide figures and tables;
      2. Bob's version 1 can be downloaded from som_v1.tar.gz, which already have many sections written. If Bob already started a certain section and now a different name is assigned, the new responsible author should build upon Bob's already written version as the starting point.
      3. We should use the Science Latex template, which I have organized as follows:
        1. Each author should organize files only under his/her own directory: PR/, DW/, KP/, RM/, and XZ/. This includes all latex, figure, and citation files;
        2. The main latex file is PVDIS_SOM_v0.tex, which you can compile using "./labps PVDIS_SOM_V0";
        3. With the exception of Bob's own sections, I have copied and pasted sections that Bob has already written to new author's subdirectory.
        4. I have not divided the references. If you are "adopting" a section from Bob, please make sure you pull out the references for that secion from Bob's som_v1.tar.gz.
        5. The tar ball for the whole directory can be downloaded here. This is where everyone should start from.
    5. Our target schedule is to have the first draft among ourselves by Thanksgiving, and a draft for circulate within the collaboration by X'mas.

    6. Total: 50-60 pages

      I. PVDIS Formalism - P.R.         (3-4)

      II. Apparatus (11-13)

         A. Polarized Electron Beam      - R.M. (1)
         B. Beam Monitoring              - R.M. (1)
         C. Spectrometers and Detectors  - R.M. (2-3)
         D. Data Acquisition System      - X.Z., R.M. (1-2)
         E. Beam Polarimetry             - R.M., D.W. (4)
         F. Target                       - R.M. (2)

      III. DIS Asymmetry Analysis (30-38)

         A. Data Selection               - D.W. (1)
         B. Pedestals and Linearity      - D.W. (2-3)
         C. Systematic Fluctuations and Beam Corrections - R.M. (1)
         D. PID - K.P.  (7-8)
         E. Background Analysis          - X.Z. & D.W., K.P.
             (i) Target EndCap Correction     (1)
             (ii) Pion Background Corrections (1-2)
             (iii) Positron Background Study  (1-2)
             (iv) Transverse Asymmetry        (1-2)
         F. Kinematics (DIS and Resonance)
             (i) Calibration of The HRS Optics  - K.P. (7-8)
             (ii) Reconstruction of $Q^2$ and $x$ (HAMC) - D.W. (3-4)
         G. Radiative Corrections        - D.W., X.Z. (5-6)
         H. Timing Simulation            - D.W., X.Z. (7-8)
        
      IV. Extraction of Results                - X.Z. (6)

         A. The formalism (2)
         B. Parton Distribution Functions (1)
         C. Extraction of $C_2$ Coefficients (1)
         D. Discussion of Higher Twist Effects (2)
    • 07/24/2012 General analysis meeting:
    1. Remote participants:  Paul Reimer, Diancheng Wang, Kai Pan, Xiaochao Zheng;
    2. Diancheng:
      1. Updated elog #152 on DAQ deadtime analysis:
        1. Posted new plots for the NIM paper, Figs. 6 and 7.
        2. Updated the uncertainty of "group dt" on page 5 using tagger results. 
        3. Because the OR dt is calculated as 1-(group dt)-(VETO dt), it has large uncertainty due to the uncertainty in VETO dt. The deduced OR dt is (16.5\pm 18.5)% for narrow and (19.1\pm 17.5)%.
        4. Still need to do: uncertainty of VETO dt for left arm using FADC (not easy).
    3. Kai: Posted and updated the OR deadtime analytic results in elog #154; 
      1. The pdf file globalDT.pdf shows that for Left arm we have good agreeement between Diancheng and Kai's numbers; This file shows preliminary results on the right arm, first shown on 7/17 (no minute), but later Kai found a mistake in the calculation, and a 4% discrepancy of his OR output rate simulated using the 8-group input rates from data.
      2. New results for the right arm narrow group are explained in file deadtime2.pdf. He found:
        1. The 4% discrepancy comes from problem in TDC for group4 at high rate (see page 2). Group 4 TDC rate is two times higher than what we expect from low rate data. Fortunately the PVDIS scalers did not show the same problem.
        2. If using low rate TDC runs to get the "K" ratio, obtained a deadtime loss of (4.2 \pm 0.084)E-4. The error comes from statistical fluctuation of  K and a 1.4% systematic uncertainty of K (based on the 98.6% agreement between the simulated OR output rate and the data). 
        3. There is a factor of 3 difference from Diancheng's 0.0014, but Diancheng's has large error bar due to uncertainty in the VETO dt, see above.
        4. To do: same analysis for right arm wide, and summarize also the left arm results.
    4. Discussions about the OR dt:
      1. With the ~100% error on Diancheng's OR dt, the two analysis are consistent, but we can't use the difference as an estimate of the OR dt error;
      2. Kai's analytic method is different from reality because the real 8 group signals do not follow Poisson distribution (minus 20ns deadtime due to trigger width): The high VETO deadtime is universal and should already forbit some signals from happening in <150ns of the previous event. The chance for an electron to be killed in the 150ns range at the VETO level is proportional to the rate of pions (and other junk) which trigger T1 but not GC. Is there a simple way to model this so Kai can use it? The actual OR deadtime could be smaller than Kai's current result because of this mechanism.
      3. OR is relatively simple, perhaps we should use Kai's results as the contribution to the total DT. But there is an added problem: Kai's results show the OR contributes to ~25% and ~6% of the total deadtime for kine#1 and #2, respectively. We know with the VETO the actual OR deadtime is actually smaller than that (see above point B). In this case, if we do not improve Kai's calculation to include the VETO we will have to quote a 100% error on OR dt, which will be ok for kine #2 but larger than desired (>30%) for kine#1.  So if we can find a way to model VETO for Kai's analytic method, we should.
    • 07/17/2012 General analysis meeting:
    • 07/10/2012 General analysis meeting:
      1. Remote participants:  Paul Reimer, Diancheng Wang, Kai Pan, Xiaochao Zheng;
      2. Diancheng:
        1. Posted elog #152 on DAQ deadtime analysis:
          1. Adjustment #1 is to change the discriminator threshold in HATS.
          2. Adjustment #2 is to change the t1 constant in HATS. This affects the relative contribution of count loss vs. pileup, but not the total tagger deadtime. - Xiaochao commented the t1 value can be cross checked with the actual TDC spectrum taken in the tagger runs. Any change from the scope picture taken at the end of the experiment tell us how much time jitter/drift t1 could have.
          3. The new plots on page 4 of the PDF file is very good, but see check of ii above. 
          4. Page 5 shows the breakdown of total deadtime to group, veto and OR. Definition of OR is "total - group - veto". Next step would be to estimate the uncertainty of these numbers:
            1. Uncertainty of group deadtime can be as small as 10% if the thresholds are tweaked to match the tagger data, as shown by the plot on page 4;
            2. Uncertainty of VETO deadtime has to come from FADC data, for which elog #89 showed a 10% agreement between FADC and analytic estimation for DIS#2 (lower rate). Diancheng commented extraction of information from high rate data is much more difficult, but will try.
            3. Uncertainty of the OR deadtime can be done by comparing Diancheng's result with Kai's analytic estimation. Kai will use the rate information from Diancheng's spreadsheet in elog #152 for this calculation. The only missing info is the overlap rate of individual adjacent group pairs, but if the group rate ratio of Diancheng's spreadsheet agree well with Kai's HRS data, can use directly Kai's HRS data on the overlap rate.
        2. Also updated the full depol calculation in elog #150. 
          1. The depol effects on the asymmetry are 0.11% and 0.2% for DIS 1 and 2, respectively. Compared to the depolarization itself (0.22% for DIS1 and 0.4% for DIS2), these are exactly halved.
          2. Xiaochao will confirm the factor of 2 using Diancheng's photon depol for elastic, QE, RES and his numbers for the asymmetry/contribution of each region - done!.
      3. Kai: Posted the first draft on the optical calibration note in elog #149;
        1. Will need to check the kinematics of optics runs 26395-26398, they must be one of the mistuned settings of RES#3, not RES#7. Will update the table on Q2 uncertainty accordingly;
        2. For ytarg uncertainty of RES#3, should we use the max 2mm or the median (perhaps 1.5mm)?
        3. For DIS#1 and #2 on left arm, we did not have sieve data. Should we add an extra 0.5mrad to the angle optimization? Kai: Using the RES#3 sieve results to carbon foil data for these two kinematics show only a universal shift in z-targ, indicating the angle is not changed. For right arm, using HAPPEX database for our carbon foil data does show different ytarg shift for the 5 foils, indicating the angle optimization is different. - Will double check this with Nilanga when he comes back.
    • 07/03/2012 General analysis meeting:
      1. Remote participants:  Robert Michaels, Paul Reimer, Diancheng Wang, Kai Pan, Xiaochao Zheng;
      2. Xiaochao:
        1. The radiative correction note was revised following Jens' comments and is being circulated by emails.
      1. Bob: 
        1. worked further on the depolarization effect, fixed a bug in HAMC and the effect is now 0.3% (click for plot). Found the "full screening" and "complete screening" limits to be very similar.
        2. helped editing the NIM paper (now called v3).
      1. Diancheng:
        1. Registered for GRC;
        2. jobs on HATS simulation submitted to farm but not completed;
        3. Also did the full depol calculation in his lastest DIS version of HAMC, see elog #150. 
          1. The depol effect is 0.22% for DIS1 and 0.4% for DIS2. 
          2. Also found very small difference between no- and complete-screening
          3. Xiaochao commented should plot the asymmetry weighted by depol and x-section, compared to asymmetry weighted by x-section only, to see the effect on the asymmetry.
          4. The depol effect should eventually be quoted as part of the HAMC-based radiative correction.
        4. Posted the beam asymmetries in elog #151. These were requested by Xiaochao for writing the PRL.
        5. Power in the apartment was out for 3 days due to the storm but the office was only out for half a day.
      2. Kai: Still working on the optical calibration note.
      3. General discussions: We should write a PRC short report on the resonance results.
    • 06/26/2012 General analysis meeting:
      1. Remote participants:  Robert Michaels, Diancheng Wang, Kai Pan, Xiaochao Zheng;
      2. Xiaochao:
        1. A technical note on radiative corrections and C2 extractions is posted on the main website under "Notes and draft papers" .
        2. Using Diancheng's plots, the NIM draft paper is also posted under "Notes and draft papers" .
        3. Will add to the tech note a section on the depolarization formula, see Bob's input below -- done.
      3. Bob: 
        1. worked on the depolarization effect of Bremstralung based on the paper of Olsen&Maximom, Phys. Rev. 114  (1959) 887
        2. A preliminary estimate for PVDIS DIS#1 is shown under this directory (copied from Bob's home dir), including the source code. The average depolarization is in the order of 1.4%, but should be investigated in a full updated simulation.
        3. Note: Polrad2.0 did not describe the depolarization algorithm in their paper.
      4. Diancheng:
        1. Need to register for GRC. 
        2. For the NIM paper, we prioritize the work to be done as follows:
          1. understand quantitatively the contributions to the total deadtime: how much (in %) is from group, veto, and OR? What are the error bars we should assign to them?
          2. Investigate if tweaking the narrow path threshold can improve the agreement between tagger narrow count loss data and simulation.
          3. Check TDC spectrum (I1 integral as in Fig.5 of NIM draft) and see how much uncertainty we can have there compared to the pictures of t1 measurement done at the end of the experiment.
          4. Investigate if a 5-ns adjustment (decrease) to t_1 can improve the agreement between tagger deadtime data and simulation (relative size of count loss vs. pileup);
        3. Coding the depolarization into HAMC (priority slightly lower than NIM paper but perhaps can be started while waiting for the timing simulation):
          1. Read the paper of Olsen&Maximom (as much as possible);
          2. Read Xiaochao's note on the depol formula (not yet written), cross check with OM paper;
          3. Code the depol factor into HAMC, cross check with Bob's code (link posted above);
          4. Proposed plots to show (all weighted by cross section and cut on the accepted events): DeltaE from external, internal brem and ionization, Depolarization from (sum of) external and internal brem, asymmetry weighted by both cross section and depolarization compared with asymmetry weighted only by the cross section.
        4. Other general to do: Will try to pick up and finalize resonance asymmetry results (main part missing: radiative corrections);
      1. Kai:
        1. Starting writing the optics calibration note. Now has finished bpm and raster corrections part (see elog 149), and is working on the optics.
    • 06/19/2012 General analysis meeting:
      1. Remote participants:  Diancheng Wang, Kai Pan, Xiaochao Zheng;
      2. Diancheng Wang:
        1. Applied for GRC but hasn't registered;
        2. Working on plots for the NIM paper, see his directory, still one more plot to do;
        3. Other general to do: Will try to pick up and finalize resonance asymmetry results (main part missing: radiative corrections);
      1. Xiaochao Zheng:
        1. Working on a note to summarize the radiative correction discussions;
        2. After the RC note will work on the PRL and the NIM drafts;
        3. Will go to GRC to give a talk.
    • 06/12/2012 General analysis meeting:
      1. Remote participants:  Diancheng Wang, Kai Pan, Xiaochao Zheng;
      2. Diancheng Wang:
        1. Participated in CIPNAP and gave a talk;
        2. Doing a lot of reading to prepare for thesis writing;
        3. Will apply for Gordon Conference;
        4. Will work on the NIM paper plots;
        5. Will try to pick up and finalize resonance asymmetry results (main part missing: radiative corrections);
      3. Xiaochao Zheng:
        1. No response from Andrei Afanasev on the gamma-Z box correction yet;
        2. Contacted John Arrington and Peter Blunden on the gamma-Z box correction and a general examination of our radiative corrections.
      4. Kai Pan:
        1. For the thesis will focus on optics calibration and DAQ PID work, may have a short section on PID deadtime (analytic work)
    • 04/05/2012 Phone meeting to discuss unblinding:
      1. I think it suffices to say that we unblinded the asymmetry at this meeting. See log #144.
    • 3/29/2012 Phone meeting to discuss unblinding:
      1. Local participants: Kent Paschke, Diancheng Wang, Xiaochao Zheng;
      2. Remote participants: Kai Pan, Krishna Kumar;
      3. Diancheng gave an update on the radiative corrections and the transverse asymmetry. These are posted on (the updated version of) his March 15 slides:
        1. Radiative corrections update:
          1. Slide 5 shows the updated uncertainty estimate for the radiative corrections, using the scheme we agreed on the 3/22 meeting. The error on radiative correction is 2% for kine #1 and 0.43% on kine #2.
          2. Simulations with Misha's table are shown on page 33 for the resonance data and page 35 for the DIS data. They have the same agreement with resonance data as Lee&Tao+Toy model, but has  a 1% difference from Lee&Tao+Toy model for the DIS corrections (compare slides 34 with 35). This indicates that the radiative correction error on kine #1 is large, as expected.
          3. We discussed different schemes of applying the radiative correction and agreed on these iteration procedure (PDF) for unblinding.
        2. Transverse asymmetries have been updated, see slide #36-37:
          1. Measured AT are shown. At this stage there is still discrepancy between Kai and Diancheng's analysis, but they will straighten this out.
          2. The error due to AT should be estimated using the 24.15ppm (for kine #1) and 44.91 ppm (for kine #2).
          3. The average out-of-plane angle of our DAQ is very small (slides 38 and 39). Therefore the first term in "correction to Ad" on slide #36 can be neglected.
      4. Xiaochao reported on the 2C2u-C2d vs beta_HT correlations, see elog #141
        1. The actual kinematics of our measurement is summarized below. These are now used in the calculation shown here;

          		Kine #1
          		 Kine #2 (L, R combined)
          		 Kine #2 (Left arm only)
          		 Kine #2 (Right arm only)
          Central
          		 Q2=1.121, xbj=0.248
          		 Q2=1.925, xbj=0.298

          Actual
          		 Q2=1.085, xbj=0.241
          		 Q2=1.901, xbj=0.2947
          		 Q2=1.886, xbj=0.2926
          		 Q2-1.907, xbj=0.2956

        2. Correlation ellipses are shown for different blinding factors and are explained in this log entry;
        3. Scroll down to Attachment 15 (the spreadsheet), one column of APV based on "no nuclear structure, no PDF" is shown. For example, "on the sheet named "SFfits-1.901", column N is using the PDG Eq.(10.21). The results of column N are to be compared with column H (CJ fit, our best fit) and T (MSTW2008 NLO F2 based with Christy R, the one we will be using for unblinding). The very small difference shows that we are not affected at all by the structure functions (and their uncertainties). A 1% uncertainty due to structure functions can be quoted and is very generous.
        4. Attachements 16 and 17 shows some plots I got from Wally on the resonance APV model and how well our data can constrain them. This is a work in progress by Wally. The goal is to contrain the background (box diagram) correction to Qweak. It's remarkable how much our Delta result disagree with the model, and how well our other resonance data agree with it.
      5. Kai reported his study on the systematic uncertainties, in elog #142. A similar table is shown on slide 7 of Diancheng's March 15 talk.
      6. To dos in preparation for unblinding:
        1. Iteration scheme for radiative corrections - Diancheng, Xiaochao;
        2. Xiaochao will get C1,2 for the actual Q2 from J. Erler, although the values we have now (central Q2) are probably good enough.
        3. finalizing transverse and pion asymmetries, and their corrections to our measurement - Diancheng, Kai;
        4. finalize corrections due to pion asymmetry and/or dilution - Diancheng, Kai.
        5. Summarize all corrections and their uncertainties in a pretty table - Diancheng, Kai.
      7. To dos that are needed for the publication, but can be done after the unblinding (partial):
        1. Radiative and PID corrections to resonance asymmetry results (check all other corrections if not final);
    • 3/22/2012 Phone meeting to discuss unblinding:
      1. Local participants: Kent Paschke, Diancheng Wang, Nilanga Liyanage, Xiaochao Zheng;
      2. Remote participants: Kai Pan, Krishna Kumar, Dave Armstrong, Paul Reimer, Robert Michaels, Zhiwen Zhao
      3. Xiaochao's report (slides, spreadsheet on PDF calculation, and spreadsheet on extracted C2q as posted elog #138), comments and to-dos:
        1. KK: we see a 0.5% difference between different PDF fits. What is the size of the structure function effect? And how does this 0.5% compare to the s.f. effect?  - Xiaochao will come up with some sort of "no-structure expression" for Ad and answer this question;
        2. KK: What is the error on beta_HT if we use the Q2=1.1 point alone? XZ's quick answer is +/-0.014, using 0.03 times (1-x)^3*Q^2 where 0.03 is the statistical error of the Q2=1.1 point. But we will check this.
        3. We concluded at the meeting that we will use the "1-parameter" fit to extract C2q from the Q2=1.9 GeV2 point only. We will publish a correlation plot (but not numerical values) for the extracted C2q vs. beta_HT from the 2-parameter fit. Xiaochao will work on such plot.
        4. We concluded at the meeting that using neutrino data's HT results on our calculated F3gammaZ term is informational, but we will not do it because there is no reason to a) trust the neutrino data or its application to our case; and b) apply a HT term extracted from the structure function to the ratio of structure functions (asymmetries).
        5. Will redo all these practice at the real Q2. All results shown were using the central Q2 and x.
      4. Kai's report (slides as posted elog #139), comments and to-dos:
        1. PID correction for day 21 and 34 have been re-worked (slide 2);
        2. So far only applied PID corrections to DIS production runs. We need to do this for all production data (resonance and transverse). See general to-do below
        3. Agree with Diancheng on the run list;
          1. Short runs (as few as 800 pairs) should be included as long as they are good runs;
        4. Found a sizable difference in kine #1 asymmetry using different BCM pedestal value (slide #1). We concluded at the meeting that this can only be caused by the beam current cut being too high/tight. We should apply different current cuts to different run periods, depending on the nominal beam current used during that period. Kine #1 for example had quite some runs taken at 70 uA rather than 100uA. Once we use a lower current cut, we should not see the bcm pedestal to influence the asymmetry so much.
        5. Compton tune runs are divided into "beam tune" and "laser tune" runs. We concluded "beam tune" should be rejected but "laser tune" should be included in the final analysis (already applied to the results shown on slide #1).
        6. To do for Kai: continue with master macro and work out all systematic corrections and uncertainties.
      1. Diancheng: does not have slides today but have gone through A-C from the 3/15 minutes. We discussed about radiative correction (was not clear on 3/15) and concluded:
        1. We do not scale the model in the Delta resonance and quote a 2-sigma error on the Delta model;
        2. RES#5 and #7 were centered around W=1.8 GeV(but different Q2) and  data agree with model almost perfectly. We do not scale the model. We will combine the statistical error of RES#5 and #7 data (take the relative error and add the inverse in quadrature?) and used the combined statistical error as our error on the toy model for W~1.8 GeV.
        3. RES#4 was centered around W=1.5 GeV and data had a ~1.1 sigma discrepancy with the toy model. We will not scale the toy model, and will quote a 1.1 sigma error on the toy model in this W region. One reason for not scaling the model is because we do not scale RES#3, 5 and 7 and we do not feel comfortable with scaling the model only in the 2nd resonance range (and not the 1st or the 3rd).
        4. We will apply different error bars to the resonance model as described above. Diancheng will check the W spectrum of RES#3 through #7 and come up with the W range that should be used for different error bars.
      2. We discussed about transverse asymmetry at the end of the meeting. In the equation shown on slide #34 of Diancheng's talk on 3/15, AT refers to the "absolute" beam transverse asymmetry. But our measured transverse asymmetry already has the k(cross)k' in and thus already have the sin(theta0) factor. So this whole equation should be divided by sin(theta0) if AT refers to the measured transverse asymmetry. To estimate the final uncertainty due to AT:
        1. Diancheng will use HAMC to calculate the acceptance/cross section-averaged theta_tr; To estimate the uncertainty, one can cut off 10% of the acceptance on the edge and see how that affect the extracted theta_tr;
        2. Nilanga will also help to estimate the error on theta_tr;
        3. AT analysis should be finalized, but see general to-do list below.
      3. Other general to do:
        1. all corrections we have applied to DIS data should be applied to Resonance data as well. This includes PID corrections.
        2. Kai and Diancheng will work on the asymmetry analysis of all other kinematics and triggers, including: Transverse asymmetry, resonances, and pion asymmetries. We should have AT and resonance results ready by the time of unblinding. Pion asymmetries are not part of the unblinding but will be published, so they should be finalized also around the time of unblinding.
    • 3/15/2012 Phone meeting to discuss unblinding:
      1. Local participants: Kent Paschke, Diancheng Wang, Nilanga Liyanage, Xiaochao Zheng;
      2. Remote participants: Kai Pan, Krishna Kumar, Dave Armstrong, Paul Reimer, Robert Michaels, Zhiwen Zhao
      3. Kai's report (slides as posted elog #137) and comments
        1. PID correction for day 21 should be re-worked: Is day 21 RES#7? 
        2. PID correction for day 34 should be re-worked: We did not have VDC on runs for day 34, but should not use interpolation of PID efficiencies between days 32 and 37 (since day #37 was 4-pass resonance runs). If day 34 was the end of kine #2, should use the average of the previous couple of days.
        3. Error on the deadtime correction does not seem right;
        4. will continue working on the master macro.
      4. Diancheng's report: (slides as posted elog #136)
        1. slide #2: for uncertainties related to BCM asymmetry, we know from HAPPEX-III that nonlinearity is controlled to 2%. We can use 3%, and the uncertainty would be 3% of the BCM asymmetry (1.32ppm). Also need to:
          1. work out this number separately for kine #1 and #2;
          2. check if there were any run with significantly larger BCM asymmetry, such as >100ppm?
        2. slide #2: for uncertainties related to beam modulation, need to use the measured slope (5ppb/nm) times the grand average (40nm), giving 200pb for the uncertainty on our measurement. But these two numbers were the overall performance of kine #1 and #2. Need to
          1. work out these numbers separately for kine #1 and #2;
        3. slide #3: for beam polarization, KK suggested the following exercises:
          1. take the statistical-weighted average of Moller (=local correction), compare to an overall average of Moller (overall = straight line fit);
          2. take the statistical-weighted average of Compton (=local correction), compare to an overall average of Compton (overall = straight line fit);
          3. construct a polynomial fit to Compton and do local correction, figure out the statistical-weighted average;
          4. Compare all above. The difference should be much smaller than the 2% systematic uncertainty;
          5. For month #2, since we have both Compton and Moller, will combine these two. The overal systematic uncertainty would be 2% and 1.9% divided down (based on the central limit theorem), giving 1.4%.
        4. On radiative corrections (slides #5, #31-33):
          1. D.A. and K.K. commented that maybe we do not want to scale the Delta model since Delta is well known. We can quote a 2-sigma error on the Delta model;
          2. We can scale the toy model or Misha's calculation based on our RES#4 result, which shows a slightly >1 sigma discrepancy from the model. Are we going to scale the model for W~1.5 GeV???
          3. kine#1 has about 26% events from the resonance region. Using a 9-10% error (based on our RES#4, 5 and 7 data), this will cause a 2.6% error on our kine#1 asymmetry.
          4. kine#2 has only 3.6% events from the resonance region so the error due to rad corr will be much smaller (good!).
      5. Other general to do: Kai and Diancheng still need to finalize a common run list and cross check asymmetries;
      6. Xiaochao will figure out how to use the kine#1 data: simultaneous fit of C2 and HT using both kine #1 and #2, or else?
    • 3/02/2012 (Xiaochao) I am very behind on logging the meeting minutes so below is just a summary of what happened in the past month, what was reported today, a to do list for the next 2 weeks, and a brief plan for the phone meeting on March 15.
      1. We will have a phone meeting at 9-11am on Thursday, March 15, to discuss where we are on the PVDIS unblinding. The plan is:
        1. Diancheng and Kai will each provide slides to summarize the procedure of the mastermacro and summary of all systematic corrections that go into the macro. The presentations will be much like what we had at the Jan. 6th collaboration meeting;
        2. Diancheng and Kai will each show their own version of the master macro. We will have screen-sharing to show how the macros work and run;
        3. We will raise any unanswered question about the analysis and answer any question people may have;
        4. In the slight chance that we do not spot serious problems, we will talk about unblinding.
      2. To do for the next two weeks:
        1. Diancheng and Kai will work on the runlist, reach an agreement and provide final asymmetry results;
        2. Kai will work on his own version of master macro. (Diancheng already has his own version but will work further on all final corrections);
        3. Kai will understand HAMC and cross-check Diancheng's work;
        4. Diancheng will use RES#3 results to figure out an overall scaling correction factor to Harry Lee's table (presumably the correction is the same for Misha's table as well). Apply this correction to Harry's Lee's table , or to Misha's table but on the Delta (1232) only, then run radiative correction again and post updated results;
        5. Diancheng will send a tar-ball of his HAMC code to Bob for a cross-checking;
        6. Diancheng will solidify the concept of different Q2's. Compare the simulated Q2 to the measured
          value. 
        7. Diancheng will also email Xiaochao (as early as possible) the Q2 and x values at which we want to publish the data. Xiaochao will use these (instead of the central settings) to calculate C2q.
        8. For beam polarization correction we will adopt PREX and HAPPEXIII's method (Bob): 
          1. for the 1st month when Compton was not available we will use Moller for a time-based run by run correction;
          2. for the 2nd month when both Compton and Moller were available we will take the average of the two (weighted by the total uncertainty of each polarimetry) for atime-based run by run correction. The error of the correction will be the larger of Compton and Moller.
      1. Today we focused on the error of HAMC simulation, see log 133 and log 134. The main method is to run HAMC with a certain number of trials several times but with different random seeds. The spread in the resulting asymmetry should give us an idea of how big the "statistical" uncertainty of HAMC is.
        1. The ratio spread (RMS)/mean in log 134 is about 0.145% for DIS#1, 20k trials per run, 50 runs.
        2. The ratio spread (RMS)/mean in log 133 is about 0.089% for DIS#2, 1M trials per run, 100 runs, and about 10% fall within the acceptance.
        3. Xiaochao tries to relate the spread of the histogram to the number of trials but do not find a clear relation. Perhaps there is no math model behind the two and we can only say the more trials we do, the less spread it is.
        4. Using the spread as the error of HAMC we think 1M is good enough. I'd like to see a spread of much below 0.1%, but since HAMC is used only to apply relative corrections, the size of the stat. spread is not that important as long as we run HAMC with fixed random seeds when calculating radiative, acceptance and PID corrections (this way any "stat" fluctuation of the simulation might just "cancel out").
      1. Run-by-run PID correction factors are now near complete. The format of the PID input to HAMC and some results are posted in log 131 and log 132. Diancheng has corresponding tables posted at his "mastermacro directory". These entries are dated and discussed on 2/24:
        1. We discussed why the correction is mostly greater than 1.00000 (this means PID corrections changes the asymmetry only towards one side), but perhaps this is just the way it is.
        2. It's not clear if 200k trials is enough for HAMC simulation. This has been studied and reported on 3/2 (see above).
      2. Change to physics models in HAMC:
        1. As pointed outby Diancheng, there is a discontinuity (a jump) in A/Q2 using Xiaochao's MSTW2008NNLO code for DIS#2. Xiaochao contacted MSTW group and found out this is a known "problem" of their NNLO PDF fits: The PDFs have a discontinuity at the charm quark mass squared. Some plots calculated by us are posted in log 130. Suggestions were provided by Graeme Watt, see his email.
        2. Now we have changed to using the fixed 3-flavor MSTW fit and in fact this has changed the sign of acceptance correction for Q2=1.9.  For the central Q2 setting, however, the 3-flavor PDFs are identical to 5-flavor values because our central Q2 (1.925) is below the charm mass sq of MSTW2008.
      3. Xiaochao continued the PDF and structure function study. Updates are posted in log 97. A snapshot is shown below:
        1. Using running value of alpha(EM);
        2. MSTW2008 routine does provide R at LO, NLO and N2LO level. R values are summarized in the spreadsheet of this log;
        3. The physics model in HAMC is using a running value of s2w (0.235) and tree level definition of C1,2q. Xiaochao got the updated EW-corrected C1,2 values and the latest s2w at Z-pole from J.Erler. In the spreadsheet a simple scaling of 2C1u-C1d and 2C2u-C2d is done to the F1gz and F3gz contributions of the asymmetry.
    • 1/20/2012, Present: Kent, Xiaochao, Diancheng, Bob (phone), Paul (phone)
      1. Xiaochao updated on how to calculate the expected APV:
        1. Updated R using Hall C (Christy's) fit. References are 1201.0576, 0712.3731, nucl-ex/0410027 (has formula for F2,R->F1), and nucl-ex/0611023. The value is between R1998 and R.Petti's calculations (so we are good);
        2. Discussed the runnings of constants with Jens Erler:
          1. Only sin2theta_W and alpha(EM) run, GFermi does not.Updated slides on Running of constants: PDF;
          2. EW corrections to Ciq: should use the Ciq definition as in Table 10.2 of PDG2011 with the s2w value in the MSbar scheme at the Z-pole. In other words, no need to run s2w if we apply EW corrections to Ciq. Jens provided the values for kappas and lambdas at Q2=1.9.
          3. Jens pointed out there is a 1% difference between the alpha (on-shell or effective) and the alpha(MSbar). For consistency we should use the MSbar scheme throughout. For example, E158 used the on shell alpha(Q2) but then applied an additional QED correction factor of 1.01+/-0.01.
        3. Kent pointed out we need to be careful with the scheme used in PDF and struct. func fits as well.
        4. Todo: 
          1. Discuss with Zein-Eddine on radiative corrections
          2. Test MSTW structure function routine and see if it provides R (code is here)
          3. Use F2's in CJ fit instead of F1's, combine with the best R to get F2 (CJ fit does not provide reliable values for R - no fit to R data!);
          4. Goal: Look for a "generic" code to calculate struct functions from PDFs in the NLO or higher order, modify Ciq in the code to study how to extract results from our APV data.
      2. Diancheng reported he is working on the master macro and discussed the following two topics:
        1. How to combine the systematic error of run-by-run corrections (such as Compton)?
          1. Should combine adjacent Compton runs anyway so each "time window" has a reasonable Compton stat error. If treated this way, no need to "combine" Compton stat error on a run-to-run basis.
          2. Systematic error of each Compton "window" should be weighted by the error of the parity (asymmetry) measurements from the same time window.
        2. Will try all 4 methods in the asymmetry analysis but how to check the results?
          1. (Xiaochao): Can derive the difference between methods on paper;
          2. Can apply a tight current cut and all 4 methods should give identical results;
          3. (Kent): Can weight the asymmetry of each pair based on the width of that pair. If can do this numerically or using ROOT, should yield the same results as the run-wise method.
    • 01/13/2012 Present: Xiaochao, Diancheng (phone), Kai, Bob (phone)
      • Xiaochao 
        • looked into the Q2 running of all constants and couplings in the APV formula
        • see below discussions on 1/9, 1/10 and 1/13(theory)
      • Diancheng: updated radiative correction slides.
    • 1/13/2012 (theory followup):
      1. XZ compared CJ (NLO) with CTEQ6L+QPM and MSTW2008LO+QPM and found comparable results on the F1 vs. F3 contributions to APV (see spreadsheet in log 97);
      2. Question: If we use NLO PDFs in constructing the structure functions F1gz and F3gz, what R to use in Y1 and Y3 to calculate APV? Anwer from WM:
        1. Since CTEQ fits did not include any data on FL (R), should use the phenomenology value (R1998 or Christy's fit).
        2. MSTW does include cross sections but don't know if they provide results on the corresponding QCD orders of R.  
      3. Since CJ fit does not provide the correct value of R, should use F2g and F2gz, then use phenomenology R to get the F1g and F1gz. This will not affect the F1gz/F1g contribution to APV, but will affect the F3gz/F1g term.
      4. For sin2W, should use the value at our Q2 (1.925GeV2). The CJ code has the running subroutine that we can "borrow".
      5. WM: Should try the ABKM PDF fits (published 2010) since they include FL.
    • 01/10/2012: Xiaochao had a phone meeting with H. Lee and M. Gorschtyen:
      1. Preparation: 
      2. H.L.:
        • Recap of the calculation:
          1. The model includes both W1,W3 terms. All parameters used in the model were determined from previous work on e-pi productions or neutrino pion productions [neutrino analysis published in PRC67,065201(2003)];
          2. Model includes Delta(1232) as well as background. At some places the background may contribute more than the Delta itself;
          3. Should not be used above W=1.4 GeV. Above 1.4 GeV the calculation only gives background, not higher resonances.
        • Need to see proton results from G0 to confirm the proton model works well (PAVI11 G0 slides sent at 1pm - Xiaochao);
        • Has been waiting for this data for 10 years.
        • The 2 sigma discrepancy may motivate a full deuteron calculation. Expect the full calculation shows smaller asymmetry than simply adding p+n. Full calculation will take min 1, max 2 years. Will talk to collaborators about this.
      3. M.G.:
        • Recap of the model:
          1. The model includes all 7 resonances from Delta to F37 (1915). Above W=2 is all background extension. Inputs are from PDG resonance data (proton and neutron/deuteron separate). HL commented the PDG resonance strengths are not well determined, for all resonances.
          2. The model is completely fixed at Q2=0. Certain uncertainties (~20%?) arises at low Q2, and more at higher Q2. For the particular calculations sent to PVDIS, used LT AD calculation to fix the high Q2 limit.
          3. The model is based on vector meson dominance, where the virtual photon fluctuates into vector mesons, which interact with the proton. This is used to determine the isospin structure of the backgrounds, which are not calculable otherwise.
          4. Can in principle include the W3 (gammaZ) term, but question how big the contribution is.
          5. The table provides W>2 calculations, but can't say whether it is more reliable than direct DIS calculation.
        • The 0 asymmetries in the table should only come from kinematically impossible regions, so do not understand why only covers 44% of the RES#3 simulation.
      4. Later addition by Diancheng: still looking into the table coverage to trace down the Misha table problem.
    • 01/09/2012: Xiaochao met Wally Melnitchouk and Alberto Accardi.
      1. Inquire what causes the big difference between Rgamma calculated by Wally and those by R. Petti (see slides 11-12 of Xiaochao's talk on 1/6) 
        • WM's is NLO(+TMC), no fitting to data. RP's is NNLO with fit to data. Could ask RP the relative NLO vs. NNLO composition of his results for a more direct comparison (followup: email sent to RP on 1/10). 
        • WM: do not plan to do NNLO for R anytime soon because it's too complicated. "What we are working on now" is a global analysis (jet+DIS data) of FL.
        • Since RP's includes fitting to data while WM's calculation miss data by a factor of 4, will use RP's vs. R1998 to study the effect on APV.
      2. What's the difference between CTEQ and CTEQ-JLab (CJ) fits:
        • Answer (slide 13 of AA's talk on 1/6): 1) different data input (CTEQ dominated by high energy, CJ includes low Q2 and low W); 2) CJ includes higher twists and TMC (that's why CTEQ had to cut out low Q2/low W data); 
        • Latest CJ fit was published in PRD84:014008,2011 or arXiv:1102.3686[hep-ph]. Direct comparison of the CJ and the corresponding CTEQ fit results can be found in "the first paper": PRD 81, 034016 (2010) or arXiv:0911.2254[hep-ph].
        • Concerning the deuteron: nCTEQ (NNLO) used A-dependence in their fitting functions, and used heavy nuclear data as inputs (from proton to iron); CJ used only proton and deuteron data and used nuclear models for the deuteron nuclear correction.
      3. How much will our 12% measurement on the Delta APV help Qweak? 
        • WM: don't know the answer right now.
        • followup from WM: will have a student in mid Feb to start looking into this. A 10% measurement could be helpful.
      4. Followup from AA: this directory contains the CJ fit code.
    • 01/06/2012: We had our first collaboration meeting. Click here for talks and meeting minutes.

    We had meetings on 12/09, 12/16 and 12/22 but no minutes is available
    • 12/02/2011
      • Present: Kai Pan, Diancheng Wang, Xiaochao Zheng, Paul Reimer (phone)
      • Diancheng:
        1. posted an entry on the first result of radiative corrections: entry #120. Below are comments and suggestions:
          1. The numbers on each histogram (page 2-4) are fractional events that come from the corresponding model (quasi-elastic, elastic, resonance table from theorists, DIS, or toy models).
          2. To better demonstrate the simulation, should add a Q2 plot, and color coding and decomposition of both the Q2 and the W histograms (red for Lee/Sato table, green for Misha's table, blue for DIS, magenta for elastic, something else for quasi-elastic, etc..)
          3. Misha's table covers less than Lee/Sato for the Delta(1232), but covers the majority of the rest of the resonance AND extend all the way to DIS.
          4. page 2 (resonance kine #3, centered on Delta 1232): 
            1. The toy model does not seem to work for the Delta region, as can be seen from the summary table.
            2. Lee/Sato's table seems to be very close to Misha's table;
            3. Our measured asymmetry is 2 sigma away from the simulation.
          5. page 3 (DIS kine #1):
            1. need to understand why there is a 4ppm difference between the two tables. Adding a row of "using DIS only" would help. Diancheng will do event-by-event comparison in order to trace down the difference.
          6. page 4 (DIS kine #2):
            1. now the difference between the two tables is only 0.6ppm. Will analyze this the same way as page 2 (see above);
            2. There is also a 3ppm difference between the A_<Q2> (supposed to be the point calculation at the apparent Q2), and the radiation-OFF <Asym>. We do not understand where this is coming from, and why we do now see this difference on page 3. 
              1. An additional row of "using DIS only" in the summary table would help;
              2. A simple analysis of whether the Q2-dependence of the asymmetry can cause this difference will help (here, can  plot a histogram of Q2, and a Q2-weighted histogram of Q2. The relative change in the MEAN value would cause a difference in A_<Q2apparent> and <Asym_radOFF>.
            3. From the summary tables on page 3-4, the radiative correction seems to be 2-3ppm for our production data. This is smaller than Xiaochao expected, but perhaps a good news for the analysis.
        2. Out last discussion on how to perform the acceptance and radiative corrections was on 05/09/2011. As pointed out above, the acceptance correction for our experiment is not as straightforward as HAPPEX, since as soon as we do full acceptance calculation, some resonance events come in and we have to use certain resonance models (not DIS). Thus we have to separate whether the correction is from the acceptance, or resonance models, or the Q2-dependence of the asymmetry.
        3. Still working on summarizing the timing simulation and working on the tagger+deadtime results.
      • Kai: 
        1. Met Kent last Thursday to discuss about the asymmetry analysis. Kent suggested additional plots to study dithering/regression and Kai is working on those.
        2. posted entry #121 on electron contamination of the pion triggers.
          1. Instead of including some "grass" in the fbTDC of the PVDIS pion trigger, now only cut on the main peak. However, cutting on the main peak seems to reduce the electron contamination by a factor of 10, which would not be sufficient to explain the observed pion asymmetry.
          2. This raised the question of how to interpret the grass in the TDC. Since the main peak cut was used in the pion contamination of electron trigger analysis, if we find we should somehow include the grass, then we have to do so also for the pion contamination analysis.
          3. We will review previous minutes and get in touch next Tuesday afternoon.
          4. Studied various fit to find the electron "tail". Poisson did not work, a 9-th order polynomial and a partial Gaussian seems to work, but did not know which one to use. Xiaochao's answer: As long as the fit works and does not behave abnormally when extended beyond the fit range, can use either one  and the difference would can be used as the error on the tail.
      • Xiaochao: Calculated the effect on asymmetry due to higher twist in R_gamma, using R calculations from Roberto Petti <Roberto.Petti@cern.ch> email dated 10/24/2011. Will present results next time.
    • 11/18/2011
      • Present: Kai Pan, Diancheng Wang, Xiaochao Zheng, Robert Michaels (phone), Zhiwen Zhao (phone)
      • Kai: 
        1. updated final pion asymmetries: entry #115
          1. concerning the discriminator threshold drifting, still need to analyze PS/SH spectrum to confirm this (see last week's minutes).
          2. The pion rates between the left and the right arm is different by a factor of ~5.
        2. updated final electron asymmetries: entry #116.
          1. now include slug 18 (and thinks Diancheng's electron results do not include this slug).
        3. Still need to meet Kent to go through these asymmetry results;
        4. Did a first estimate for the effect of electron contamination in the pion measured asymmetry, see entry #118.
          1. effect seems to explain the observed pion asymmetry (order of mag).
          2. need to improve the estimate of the e/pi ratio.
        5. Should learn about radiative corrections within the next month (since Kai is leaving UVa at the end of Dec).
      • Diancheng: 
        1. Posted entry #117 , which is the first overall deadtime correction to the left arm kine #1. 
          1. the deadtime correction is about 1.6-1.7%, smaller than what we thought before;
          2. The fits give non-zero p0, might be due to the low statistics at low currents. Will run longer at 20uA, maybe even add a point near 0 uA.
        2. Right arm is still ongoing.
        3. Preparing a final writeup on the deadtime correction/simulation.
        4. Meanwhile, also picked up a little radiative corrections.
      • Xiaochao: Will update the unblinding page. to reflect recent analysis updates.
    • 11/11/2011
      • Present: Kai Pan, Diancheng Wang, Xiaochao Zheng, Paul Reimer (phone)
      • Kai: 
        1. posted final pion asymmetries: entry #115
          1. the rate drifting (dislpnar/bcm1) is due to discriminator (or Sum8) threshold drifting.
          2. For runs before 14120, left arm pion trigger had mis-aligned timing (total shower low threshold 706 output was too early for the pion VETO). Around run 14120 a 16ns delay was added between the 706 output to the logic module input, correcting this misalignment. This change could result in a chance in the electron contamination of the pion trigger
          3. Still need to estimate the correction to the pion asymmetry from electrons.
        2. doing the final electron asymmetries: entry #116.
          1. the run list is a subset of Diancheng's
          2. has not done the detailed diagnostics. Will do this, then meet with Kent to discuss both electron and pion results.
      • Diancheng: running the high statistics timing simulation and preparing for a summary of this work;
    • 11/04/2011
      • Present: Kai Pan, Diancheng Wang, Xiaochao Zheng, Kent Paschke, Nilanga Liyanage, Paul Reimer (phone), Robert Michaels (phone)
      • Kai: reported on the pion asymmetry work (entry #114)
      • Diancheng: Produced the global deadtime result. Low rate simulations seem to have large statistical uncertainty and should be improved by running longer simulations. Thus simulations for kine #1 is much easier than for kine #2. In general, 1 hour of simulation provides 10msec of simulated data.
      • Xiaochao: posted entry #113 for the effect of electron asymmetry on the pion measured asymmetry.
    • 10/28/2011
      • Present: (?) Kai Pan, Diancheng Wang, Xiaochao Zheng
      • Kai:
      • Diancheng:
      • Xiaochao:
    • 10/21/2011
      • Present: Kai Pan, Diancheng Wang, Kent Paschke, Robert Michaels, Paul Reimer (phone), Xiaochao Zheng
      • Thanks to Bob who drove all the way from JLab for this meeting!
      • Kai: Working on the pull plot and the sigma vs. bcm plots suggested by Kent last week.
      • Diancheng: 
        • has been working on HATS simulation for the final deadtime corrections; 
        • the plots in entry #95 are not the final ones, will need to check this at some point.
      • Xiaochao: commented that for the paper we need to know the asymmetry errors before and after Nov. 28 so we know what numbers to quote for the deadtime correction. But if we quote both values then we may not need a single number for the overall correction
    • 10/13/2011
      • Present: Kai Pan, Diancheng Wang, Kent Paschke, Xiaochao Zheng
      • Kai: was sick most of the week, but updated log entry 109 on pion asymmetries (last two figures only). Now all pion asymmetries were not blinded. 
        • Wierd thing is the kine#2 pion asymmetry is non zero (15ppm level, about 2.5 sigma) and do not flip sign with HWP in/out. Although they average to 0, we should investigate this further.
        • Will update other figures of this entry to reflect the latest change.
        • Need to understand the non-Gaussian tail in left arm kine #1, although we suspect it's simply because of lower beam currents. Will use pull plots to diagnose.
      • Kent: explained the beam energy asymmetry is obtained from Delta x of bpm12 divided by 4m (this is dE/E).
      • Although we are eager to see Diancheng's asymmetry results on pions and also some results from the rad cor, Xiaochao set the priority of this week for Diancheng to be: 
        • generating the final tagger (deadtime) data, complete the corresponding table on this page.
        • generate the revised plots for the NIM paper.
      • Xiaochao: will send the kinematics to Andrei Afanasev, who promised to do internal radiative corrections (point calculation), and generate a 3rd set of tables for our external radiative corrections.
    • 10/07/2011
      • Present: Kai Pan, Diancheng Wang, Robert Michaels (off-site), P. Reimer (off-site) Xiaochao Zheng
      • Checked the blinding part of PAN and Diancheng found that the pion asymmetries were blinded on the right arm (narrow & wide), but not on the left arm (narrow & wide). We removed all pion triggers from both PAN code of Diancheng and Kai after the meeting and will try to be consistent this time: ONLY DIS electron triggers (left and right, narrow and wide) should be blinded in the analysis.

    • 09/29/2011
      • Present: Kai Pan, Diancheng Wang, Robert Michaels (off-site), Xiaochao Zheng
      • Kai:
        • updated log entry 102 to post the (near-)final electron results:
          • Shows about 1-2 ppm difference in both kinematics from Diancheng's numbers (the error bars are 3.x and 5.x ppm, respectively). This is far from ideal (we expect the final results to differ only in 1/100 of the error bar).
          • Todo: will work with Diancheng to cross-check the run lists and the cuts.
        • posted new log entry 109 on pion asymmetries:
          • now discard slug #1 completely and extract run list for all other slugs. Most of slug plot looks very Gaussian, only 1 or 2 slugs are not (did not post plot).
          • kine #1 combined result shows non-Gaussian tail and is far away from zero (-45.79 +/- 7.84)ppm. Diancheng's asymmetry analysis (to do) might help to diagnose whether this is from a bad run, or is due to change in the trigger timing (which effectively change the pion rate or signal contamination) - which would be "normal".
          • kine #2 combined result is almost a perfect Gaussian and exactly zero: (0.53 +/- 4.07)ppm. This triggered the question whether pion asymmetries are blinded. Will check this in the configuration file (and a few of us have the impression that the blinding factor may not apply to all pion triggers at all).
          • The question then is why the left arm/kine #1 shows large asymmetry. This may very well be due to electron contamination: the electrons are only rejected by Gas Cherenkov, then fed to pion triggers. The lead glass do not play a role for pion triggers at all (except to reject low energy background). Also, since we tried to have very clean electron triggers, the pion trigger must had very high electron contamination. For kine #1, the pi/e ratio is low (both calculated and measured), so the effect of electrons is much larger than for kine #2.
          • Todo: simple estimate of electron contamination and correction;
          • Other todo: add wide pion trigger results(?)
        • updated log entry 107 on positron asymmetries:
          • added the run list.
          • only left arm results are included. Right arm Q3 had wrong polarity during this time and then was set to 0 (see this halog).
          • Todo: We should still extract right arm positron asymmetries because the Q3 setting does not affect the main kinematic settings much (only the detailed acceptance shape is modified, see Jin Huang's hall A logbook entry listed in this elog). Also, we are only confirming the asymmetry of the positron being zero, so the acceptance won't matter much here.
          • Todo: should do Gaussian fit and compare with the MEAN value.
        • have not updated log entry #106 yet (see report on log 106 from last week)
      • Diancheng:
        • have been working on coding Misha's new tables. The new tables are huge and adds difficulty to the programming.
      • Xiaochao:
        • Once Kai is on track for the asymmetry analysis, he can get into the work on radiative corrections.
        • Suggested Diancheng to finish the plots needed for the NIM paper (which can be done along with producing off-line tagger results).
    • 09/23/2011
      • Present: Kai Pan, Diancheng Wang, Paul Reimer (off-site), Xiaochao Zheng
      • Kai: posted results on:
        • log 106: transverse asymmetries (unblinded), final.
          • These are narrow triggers. Will post wide trigger results later
          • The combined plot gives different values from the chi2 fit. The chi2 fit results are now posted on the unblinding page. For the combined asymmetry plot, should do Gaussian fits instead of using the MEAN value from root.
        • log 107: positron asymmetries (unblinded), preliminary.
          • The positron results here are unblinded because in order to avoid (0,0) count pairs (which PAN automatically set to -1), the asymmetry is now calculated from the R tree with 10 helicity windows combined. When doing so the blinding factor (applied to the P tree) is no longer valid. The comments at the beginning of the entry ("run#: blinded or unblinded") refer to the status of the P-tree which is irrelevant for the results posted here.
          • need to double check run list and add info on which arm this is.
        • log 108: pi- asymmetries (blinded -- but see 9/29 report, the pion is probablly unblinded), preliminary
          • Found slugs 1-4 all have non Gaussian shape. Starting slug #5 the data quality is much better. The main problem is early data did not have good pion trigger quality.
          • Diancheng suggested that for pion run list, can start from the electron run list and exclude runs that have obvious problem (from halog) or non-obvious ones (by checking the plot of each individual run).
          • Xiaochao suggested while going through all runs, can combine this work with "picking the run list for VDC on runs", etc.
      • Diancheng: still working on radiative corrections.

    LONG BREAK, SEVERAL MEMBERS ARE IN CHINA

             Present: Diancheng Wang, Xiaoyan Deng, Paul Reimer (off-site), Xiaochao Zheng          Present: Diancheng Wang, Xiaoyan Deng, Ramesh Subedi, Paul Reimer (off-site), Xiaochao Zheng