PVDIS analysis and meeting minutes

• PVDIS (E08-011) main page
  
• Meeting schedule: We meet every Monday 2-4pm (2011 new schedule!). 
• To receive meeting reminders and call-in instructions, please send an email to xiaochao@jlab.org. 
• Color coding: Typically, todos are in red, followups are in green.
• PVDIS wiki and elog, and analysis flowchart.

• Next meeting: 12/09/2011, Friday, 2-4pm. UVa people please meet in room 120. Off-site people please use: US 866-740-1260; International 303-248-0285; Access code: 8206694.
  

• Here is an overview of various things that should be done, their status and the person in charge

• Electron pile-up analysis - Kai, in progress but not concluded, last reported on 1/31
• Electron efficiencies - Kai, mechanism developed and will need to apply to each run. Error in asymmetry due to error in efficiency still need to be estimated using HAMC log 74.
• Optics calibration - Kai, completed
• Fine tuning of the HAMC apertures - Kai (left arm), Diancheng (right arm), in progress.
• Continue with Compton summary - Diancheng
• Positron contamination - Kai, in progress, last reported on 7/4
• NIM paper - Xiaochao, waiting for Diancheng's plots.

• Unblinding Worksheets have been moved to this page.

• 09/16/2011
Present: Kai Pan, Diancheng Wang, Xiaochao Zheng
• Kai:
• Settling down at UVa
• Working on positron dilution analysis
• Will work on pion dilutions. Xiaochao commented that, should do this for everyday's data using optics (VDC on) runs. The best is to make a table of
• Date
• optics run number
• pi- dilution results of this run.
• Diancheng:
• Checked the RES#3 analysis. This directory contains some of the plots. Beside the slight Q2 mismatch which is sort of expected due to Q2 and Q2 mistuning, do not see big problems here.
• But, during the meeting we noticed the right arm was not sitting at RES#3 for slug 25. And for slug 26, only about half of left arm runs were for RES#3, then it was moved to RES#4. Will double-check the runs listed.
• Xiaochao: At PAVI 11 Dave Armstrong presented the Delta APV results from G0. Both LH2 and LD2 targets were used. Only the LH2 results are compared with Sato/Lee's model and they agree very well. The LD2 results were presented but not compared to Sato/Lee's result. Go to PAVI program page to access his slides. Currently a W&M student is working a Ph.D. thesis on the Delta resonance.
• Todo list:
• Both Diancheng and Kai should do the run list check and the asymmetry analysis
• first, focus on resonance data, make a table for:
• Date
• Parity run# (range)
• kinematics (RES #?)
• slug result and plot (if the slug is for single kinematics)
• asymmetry for each resonance kinematics
• Diancheng will continue with radiative corrections. But with more confidence in Sato/Lee's calculation (and our data), this will hopefully proceed with less difficulty.
• Both Diancheng and Kai will provide final number for the pion and electron wide asymmetries. Also electron narrow (if not already reported in the unblinding page).
• Also work on the transverse asymmetries. Note that we need final numbers companied by elog entries with plots posted.

BREAK, PAVI 2011



• 09/02/2011
Present: Diancheng Wang, Xiaochao Zheng
• Diancheng: 
• focusing on RES#3 analysis, but currently the data don't agree with Lee's calculation so well. Did point checking of Lee's calculation and found no problem;
• Not comfortable to send a conclusion to the theorists yet;
• Xiaochao commented that Lee/Sato's model might not work perfectly, since we used the deuteron. 

• 07/25/2011
Present: Jie Liu, Zhiwen Zhao, Xiaochao Zheng, Paul Reimer (off-site)
• Xiaochao Update on the resonance calculation:
• Received calculation from Sato/Lee on DIS: dis1.gz,   dis2.gz; putting all tables together here: RES#3, 4 and 5 (same table for all 3), and for RES#7
• From Mikhail Gorshteyn: 
• writeup on how the model works;  My understanding of the calculation is
1. Take the non-PV proton structure functions F1 F2, decompose it into the resonance and the background like other phenomenological models;
2. Do isospin rotation to get the neutron, then for the deuteron;
3. Then replace the EM coupling by the appropriate weak couplings, get the interference term F1(gammaZ);
4. Form the asymmetry as (G_F * Q^2)/(4 sqrt(2) pi alpha) * g_A^e * (F1^gammaZ/F_1^gamma).
5. Compared to the exact APV formula, this is a very simplified version: it contains only the a1*Y1 term and implies Rs, Rc=0. The F3^gammaZ term is not included and according to Misha is not calculable.
• plot to show the resonance structure. Here the Cahn-Gilman value is the leading twist APV, which for the deuteron is a constant (9/5-4*s2w). Compared to my APV formula, this is the a1 term if neglecting Rs, Rc=0. However I do not understand why at high W and high Q2 the calculated value deviate from the C-G value.  followup: here is Misha's reply on why the ratio does not converge to LT at high Q2 and W.
• I am still confused with how the calculation is done (mainly, why it does not converge to leading twist at W>2). I am in contact with Mikhail about this. Meantime, the following tables are for record only, do not implement them in HAMC yet.
• the tables: RES#3, #4, #5, #7, DIS #1, #2. The format is:

  They have 9 entries: E(GeV) E'(GeV) theta(deg) W2(GeV2) Q2(GeV2) APV(CG) APV(central value) (APV+delta) (APV-delta) APV(CG) stands for the leading twist result of Cahn-Gilman; APV(central value) - predictions for the APV on the deuteron in the resonance+background model. Only the a1 term is calculated; APV+-delta - upper/lower values of APV in the resonance+background model. Only the a1 term is calculated. If the W value falls outside the region where the model is applicable, APV(central) and APVdelta are set to zero.

  of course the table for #3 should cover #4 and #5 (which it does here), but because Misha used 0.2GeV bins for E and E', the actual E and E' bin values are different in the three tables. This should not cause too large a difference whether one use table #3 for HAMC at #4 and 5, or tables #4 and #5 for HAMC at #4 and #5 (I think).
  

• 07/18/2011
Present: Kai Pan, Jie Liu, Zhiwen Zhao, Xiaochao Zheng
• Kai:
• reported on the positron asymmetries (elog #101). 
1. page 1: Poisson fit shows the average rate during the positron run is 1.94 for each helicity pair (or single?).
2. page 2: Found PAN set A=-1 if both N+ and N- are zero (which gave the -5646 ± 1436 ppm raw positron asymmetry for DIS#2 reported last week. Top plot is done by hand (Kai) to form the asymmetry with the same A=-1 rule, which agrees perfectly with PAN results (middle plot). After correcting the rule to A=0 if N+,N-=0, the results are shown in the bottom panel, giving much smaller asymmetry. 
3. page 3: corrected positron asymmetry, same as bottom plot on page 2. Note that this was done with only one positron run, while the results reported last week was from all four runs.
4. page 4 and 5: same thing but for the left arm (DIS#1). Note again that this is from one run, and the central A value is different from the value reported last week (four runs).
• Xiaochao
• Emphasized that a lot of analysis were nearly complete but never finalized. Must finalize them one by one and fill out the master table.

• 07/11/2011
Present: Kai Pan, Zhiwen Zhao, Jie Liu, Xiaochao Zheng
• Kai:
• reported on the positron and pion corrections and asymmetries. See the table above. The positron asymmetries are
• DIS#1: -180±1229 ppm(raw), -219.9±1438 ppm (pure e+, i.e. after correcting the pion+ contributions)
• DIS#2: -5646 ± 1436 ppm (raw), needs to be studied further.
• Xiaochao
• Update on the Delta calculation from Sato/Lee: 
• the table reported on 5/31/2011 had an angle coverage of only ±3mrad (my mistake, should be ± 30 mrad or ± 2 degrees). We got the fully radiated table with the correct angle range for RES#3, 4 and 5 (same table for all 3), and for RES#7. The bin sizes are: 1 deg for angle, 0.0186 GeV for Ei, 0.07 GeV for Ef. The coverage is illustrated in this plot but the plot has larger bins.

• 07/05/2011
Present: Robert Michaels, Kai Pan, Diancheng Wang, Zhiwen Zhao, Xiaochao Zheng
• Diancheng:
• Presented results for the resonance simulations with radiation on. However, had a mistake of using W^2>2 instead of W>2 in whether the RES or the DIS asymmetries can be used. This should be corrected ASAP. But, some interesting observations:
• RES #3: using deuteron elastic (Xiaochao's code), quasi-elastic, and Sato/Lee's table (radiation off), the result is pretty close to the data. (Data has ~11% error bar)
• RES #4,5,7: using deuteron elastic (Xiaochao's code), quasi-elastic, and two "toy" models for the RES region: One is A(RES) = Xsec(RES)/xsec(DIS) * A(DIS), the other isA(RES) = (2 - Xsec(RES)/xsec(DIS)) * A(DIS). Because of the wrong W cut, about 87%, 97%, and 98% of the events are actually using the DIS formula for APV. But the agreement is remarkable. Does it imply "duality" in PV asymmetries?
• todo: In addition to correcting the W cut, will also try: 
• enlarging the RES effect, such as using A(RES) = [ a*(Xsec(RES)/xsec(DIS)-1) + 1] * A(DIS), where a=2,3,...
• run RES#3 simulation with the toy models and the DIS formula alone (that's 3 simulations)
• Xiaochao will write to Harry Lee for more calculations on the Delta. (followup: Sato will do the fully radiated table for RES#3 first).
• For deuteron elastic, we have 3 codes: Peter's, Xiaochao's (probably out-dated) and the SAMPLE code from E. Beise. Peter's code does not compile with C. Xiaochao will take a look.
• Kai
• Did the transverse running asymmetry analysis. However, did it "unblinded". We must be causious to not compare his with Diancheng's earlier AT results (included in Xiaochao's PVDISlog, Diancheng's 4th year review talk, and other talks by Xiaochao and Bob);
• Worked out all the formula for positron and pion corrections. Did the positron rate analysis. Found the dilution effect of positrons to be at the 1E-3 level, and pions to be at the 1E-5 level. 
• todo: 
• for pion rates, should use (T1&notGC), not the pion trigger from the PVDIS DAQ (pion triggers in our DAQ had too high thresholds in the discriminator 706).
• for positron rates, should use high rate parity runs.
• should do this analysis for both kinematics (both arms during AT running)
• should extract asymmetries of positron runs, (both positron and pion trigger)
• Also, it was brought to attention that we should eventually extract all pion asymmetries from data, regardless of how (large) the error bars are.

• 06/20/2011
Present: Kai Pan, Diancheng Wang, Zhiwen Zhao, Xiaochao Zheng
(suggested agenda and topics to be covered:)
• Diancheng:
• Discussed in details the structure of radiative corrections in HAMC.
• Xiaochao commented that perhaps we should put the following 2 tasks at high priority:
• Extract the asymmetry (unblinded) from RES#3 and compare to HAMC (radiation off) using Lee/Sato's table;
• Extract the asymmetries (unblinded) from all RES settings, and list them along with the corresponding W and Q2.
• Xiaochao
• follow-up on eD elastic scattering: 
• for formalism, see this review paper by M.J.R-M and T.W. Donnelly, Phys.Rept.239:1-178,1994 (Sciencedirect link, downloaded PDF), or S.J.Pollock PRD 42 (1990) 3010 (link, downloaded PDF) that contains some actual calculations.
• but some much more experimentalist-friendly information is in this slide from E. Beise. She also provided two codes: SAMPLE_deut_elastic.F and SAMPLE_prot_elastic.F (form factors in the deuteron code are using the K.Kelly parameterization of the nucleon form factors, which worked better for the backward angle measurement of the SAMPLE experiment. We can test the calculation with other form factor fits.). As a cross check, at Eb=6.0674 and theta=12.9 and 20.0 degrees, I got: 

  	Q2 (GeV2)	xsec(nb/sr)	APV (ppm) using J.Kelly's nucleon form factors	APV (ppm) using f.f. from Friedrich+Walcher  hep/ph/0303054
  kine 1	1.7180	9.412E-6	142.896+4.244*GMs	142.896+4.066*GMs
  kine 2	3.7154	8.0213E-9	309.035+295.678*GMs	309.035+294.259*GMs

• 06/13/20101
Present: Kai Pan, Diancheng Wang, Paul Reimer (off-site), Xiaochao Zheng
• Diancheng:
• working on the resonance and QE/elastic in HAMC and the master macro. Not sure what PV asymmetry can be used for eD elastic scattering, Xiaochao will find some theorists about this. 
• Kai:
• started asymmetry analysis. Mostly, spent lot of time on sorting out the runlist
• posted entry 97
• Fig.1: extracted pedestals. However, found 51 pedastal runs instead of 27 that Diancheng used on his 5/19/2011 entry. Diancheng commented that one should be very careful in selecting only clean pedestal runs (no pulsed beam, etc.). For example Kai's plot showed a lot of events at 3 and 4 (2 orders higher than 1 and 2), indicating some of the runs used may have had beam.
• BCM calibration. Here the plot is BCM1 scaler vs. UNSER scaler.
  • Diancheng commented that using ADC BCMs (HAPPEX crate copies) are probably much better than using our scalers (less discontinuity -> higher accuracy).
  • Diancheng also commented that the "bcm1r" copy in HAPPEX crate is the actual "bcm1" plus some "DAC noise" that were also recorded. HAPPEX analysis used a pedestal of ~5000 for bcm1r, while Diancheng used ~4200 from bcm1 which he believes should be the correct one to use. He also checked the consistency of the two and found bcm1 = bcm1r - DACnoise (signals are consistent).
  • Xiaochao commented that she is always curious about which BCM to use. Perhaps Kai can extract the asymmetry of a couple of runs (only), using scaler and ADC BCMs, respectively, and see how large the effect is.
• Xiaochao
• Posted entry 97 on the exact vs. approximate PVDIS asymmetry formula. Also writing a "mini" macro to extract C2's from the measured asymmetries.
• About the beam energy: The "Hall A momentum (Tiefenbach) (HALLA:p)" recorded in the End-of-Run entries are the ones to use (info: Doug H.), as long as the "Hall A momentum orbit correction (dpp)" is not zero (you don't need to apply this correction, a non-zero value means things are working properly). Doug suggested spot-check every 50 runs or use a script to extract this from all End-of-Run entries.

• 05/31/20101
Present:  Kai Pan, Diancheng Wang, Zhiwen Zhao, Robert Michaels, Paul Reimer (off-site), Xiaochao Zheng
• Diancheng:
• worked on the resonance models, but no result to show yet.
• for next week, will work on Sato/Lee's table for RES kine #3. We need to address two questions from Lee:
  • Is the table formatted correctly for our use?
  • Does the prediction agree with our data? They need to know this before proceeding to DIS kinematics.
• Kai
• will start the asymmetry analysis. Goal is to provide independent results from Diancheng's asymmetries
• Completed optics analysis for left arm 20 degress
• Xiaochao
• On the resonance PV asymmetries:
• Delta PV theory:
• Sato-Lee model: see Phys.Rev.C72:025204,2005 or nucl-th/0504051. We also received their calculation in the tabulated form for our RES kine #3: table, and a brief memo on how to use the table.
• Shi-Lin Zhu, et al.: see hep-ph/0107076. and others.
• Pollock, Mukhopadhyay, M.J. Ramsey-Musolf et al., see nucl-th/9806079, (less results than nucl-th/9801025)
• Mukhopadhyay, M.J. Ramsey-Musolf et al., see Nucl.Phys.A633:481-518,1998 or nucl-th/9801025, (Table III is for APV for beam energies 1,2,3,4GeV and various scattering angles. Can possibly take this table and do a rough estimation of the Delta asymmetry to compare to our RES kine #3).
• Nath, Schilcher and Kretzschmar, Phys.Rev.D25:2300,1982. (graphical predictions of A are given for E=2 and 10 GeV).

• 05/23/20101
Present:  Diancheng Wang, Zhiwen Zhao, Robert Michaels, Paul Reimer (off-site), Xiaochao Zheng
• Diancheng: showed more results on the asymmetry analysis, see PVDIS log#95
• Fig.1-3: asymmetries from the two kinematics, raw, and after dithering and after regression corrections. Each figure has 4 plots: kine 1 (left arm only), kine 2 from left arm, kine 2 from right arm, and difference between left and right arm for kine 2. The small tails will be explained using Fig. 5 below. Diancheng mentioned that he did threw out one event which is one pair at -400000ppm (that's -0.4)!
• Fig.4: during the week we discussed how the asymmetry width varies with beam current, the answer is width propto sqrt(current). This figure intend to show that this is indeed the case, since width decreases with BCM readings. Xiaochao commented that one can add a sqrt() fit to guide the eye. Diancheng also explained that most of the low current runs were from slugs 4 and 5 where we had beam problems.
• Fig.5: same as Fig.1 but with slugs 4 and 5 removed. So the "tails" in Fig.1-3 were from lower current runs.
• Fig.6-8: Asymmetries for the 2 DIS kinematics. We tried to look for problems in these plots, but did not see anything obviously wrong.
• Fig.10-15: pull plots from data;
• Fig.16:
• Fig.17: Beam charge asymmetry. Noticeably one setting has -3ppm asymmetry. Bob commented that since the BCM linearity is about 2%, this would cause a 3ppm*2% systematic error on the measured asymmetry. Still this is very small (0.06ppm), and should affect only HWP in data.
• Fig.18-24: Similarly to Fig.16, these are diagnostic plots to optimize cuts.
• At the end, Bob commented these are "good progress", and "means we are close to unblinding". But we should still contact Kent et al. to find out what their suggestions are for getting rid of the outliers in Fig.16 and 18-24 -- just for the record.
• Bob: we should contact some theorists for suggestions on resonance PV models.
• Xiaochao: worked a bit more on the NIM draft paper.

• 05/09/2011
• Diancheng: Discussed asymmetry analysis, see PVDIS log#94
• page 1: Asymmetries for the two kinematics, from about 3/4 of data. Slugs with obvious problems have been removed. The distribution agree very well with Gaussian. (This may suggest that the effect of uneven PID efficiency is very small)
• page 2: corrections dithering and regression are very small;
• page 3: pedastals - appears to be 0.0005 for disrenar. Xiaochao commented that although this pedestal is very small, it may still have a non-negligible effect on the calculated asymmetry, since the average count per pair could be as small as 10^3 counts. Then a 5E-4 pedestal would give a fraction of ppm "dilution" if not corrected properly.
• page 4-5: looked more at BCM calibration in the hope of making the rate scan data useful, including using lumi reading to "extract" the target density change. But it does not look promising.
• Just for the record, we did ask Kent about the Q2 meaning in the final publication. The answer is that it is <Q²_apparent> simulated using the beam energy minus the average energy loss, and acceptance and the central setting of the spectrometer.  From the measured asymmetry to here, then, corrections must be applied in multiple steps: 
• unique for PVDIS: from <A(Q²_{vertex with an uneven/real PID})> to <A(Q²_{vertex with an even PID})>
• from <A(Q²_vertex)> to <A(Q²_apparent)> (radiative corrections)
• from <A(Q²_apparent)> to A(<Q²_apparent>) (acceptance corrections)

• 05/02/20101
Present:  Diancheng Wang, Zhiwen Zhao, Xiaochao Zheng
• Diancheng:
• Discussed asymmetry analysis, adapting codes from HAPPEX; A few questions arose here:
• for beam, using bcm1 which is a HAPPEX crate copy. Will check on HALOG if this is consistent with our crate timing;
• how does beam position work between PAN and HAMC?
• Typically, one should use the real beam position from each run for HAMC simulation of that particular run (given that we need different PID input in the simulation, it's not extra work to also vary the beam). However, if dithering or regression corrections are applied, what beam position we should use in HAMC?
• For HAPPEX, some "double asymmetries" were used, such as A(bpm4x)-A(bpm4y) from each run. When making this type of plots, we look for "outliers" that are outside the main Gaussian, use this as a criteria to ID problematic runs, and reversely look at variables (such as bpms) of those runs to determine the cut on bpm. Once the cut is optimized, there should not be any outlier except for some low-stat runs which should have wider std. deviation than the main Gaussian (who width is determined by "normal-statistics" runs). These low-stat runs should be kept in the analysis. 
• Is there any other cut(s) we should use in addition to HAPPEX cuts?  Probably not.
  
• waiting for Kai's database to start on HAMC aperture study and resonance analysis;
• Also working on master macro.
  

• 04/04/20101
Present: Diancheng Wang, Kai Pan, Vince Sulkosky, Zhiwen Zhao, Robert Michaels, Xiaochao Zheng 
• Diancheng reported on his VETO deadtime work using FADCs, see log 89:
• Wrote a code to recognize pulses and extract their timing from FADC data. Here, call them "TDC"s, which are truly "extracted timing from FADC";
• Fig.1: GCtdc shows an afterpulsing peak around 500ns after the main peak. This afterpulsing peak corresponds to a multiplicity of 2 or more. People questioned whether this timing is too long, and Diancheng reminded us that Ramesh did some measurement of T1 afterpulsing during the experiment, see halog 297168. The pictures showed the afterpulsing is indeed about 500ns after the main signal. Although these pictures were taken from T1, GC should be similar since they are due to PMT afterpulsing.
• Fig.2: Difference between T1 and GC timing. The main peak at 0 are events with both signals. The two side peaks are either "T1 no GC" or "GC no T1". 
• Using these 3 peaks can in principle give us the fraction of physics signals that are "have both T1 and GC", "T1 no GC", or "GC no T1", which are needed to provide input models for the trigger simulation. But Diancheng commented that the ratios here might have complications from prescaling factors. Instead, he used a different run to extract these ratios, see Fig.5.
• Applying the red and the blue cuts gives:
• red: (NT1 == 1) && (NGC == 1) && (NVeto == 0). These are signals with valid T1 and GC but no electron VETO
• blue: (NT1 == 1) && (NGC == 1) && (NVeto == 0) && (NPS == 1) && (NTS == 1) same as red but are definitely electrons
• The small peak on the left side of the main peak (at 0) are due to VETO circuit deadtime. In principle, the ratio of this red peak over the main black peak at 0 gives the VETO deadtime loss. This is the another direct evidence of VETO deadtime, in addition to the tagger test and the simulation results. This is very important as it might be the only evidence after we adjusted the T1 input width to the VETO.
• Fig.3: zoom-in of Fig.2. The width of the VETO deadtime peak should be (T1input-T1output) (~70ns) plus the timing jitter, so the observed 100ns width here is reasonable;
• Fig.4: similar to Fig.2, but applying different red and blue cuts. Extracted from this plot: VETO Deadtime loss is 0.35% (552/158570). Total loss is 2.7% (154285/158570). Total loss also include PID and timing efficiencies.
• Fig.5: similar to Fig.2 but using a different run to extract the ratio of "T1 no GC". (the "GC no T1" peak vanished here?). Using this ratio as input to theory, we get 0.42% VETO deadtime, compared to the 0.35% VETO deadtime from data. The 20% agreement is not bad.
• Overall, great work on using the FADC data!
• Xiaochao commented that we still want to see more comparison of HATS vs. tagger, as suggested in 3/28 meeting.
• Otherwise, the HATS work is near complete. Suggest Diancheng writing a technical note on this.
• Diancheng also worked on asymmetry analysis, but nothing to report yet.
• Kai reported on optics calibration, see log 90
• Fig.1-3 are left arm mis-tuned sieve slit runs, after using multi-dimensional cuts for angle optimization. 
• Fig.4 is right arm ytarg optimization for data taken at 20 degrees. Do we have this for both angles?
• We concluded that Kai's optics calibration is (near) complete. And that Kai should write a technical note.
• Will give a dry run this Thursday for next week's DIS2011.
  
• Bob:
• commented that both Kai and Diancheng have "blossomed", although he also said this might not be the right word for Chinese young men.
• Xiaochao
• reported on Q2 error from optics based on information from Nilanga, see log 88. Bob commented that the error bars quoted are "believable".
• talked about how to put together all the analysis we have done so far and proposed both Kai and Diancheng start writing (or think about writing) a "master macro" for the final analysis and unblinding.

• 03/28/20101
Present:  Diancheng Wang, Kai Pan, Vince Sulkosky, Zhiwen Zhao, Robert Michaels, Xiaochao Zheng
• Diancheng reported on VETO deadtime from trigger simulation, see log 87
• page 2: brief overview of the deadtime
• page 3: path deadtime from HATS (bottom of the page shows a mistake that was causing earlier errors for 4 of the 8 groups);
• page 4: review of the tagger system and a diagram drawn during the experiment to show the large T1/VETO deadtime of tagger;
• page 5: details of the VETO deadtime cause, now implemented into HATS (bottom right plot);
• page 6: now start on physics signals, how it still suffer the large VETO/T1 deadtime even after our correction which made the VETO deadtime to disappear for tagger ONLY;
• page 7: evidence from FACD data. Here the T1 is caused by an event ~30ns before the electron event, and no VETO (electron allow) is generated due to the non-updating feature of T1's 755;
• page 8: This large deadtime loss is proportional to the rate of "T1 no GC" plus rate of "GC no T1". Here trying to model these rates in HATS, but not sure if there is any correlation within?
• page 9-12: showing good agreement between theory and HATS;
• page 13: again, are there any correlation between different background rates?
• page 14: History of VETO circuit during the experiment; 
• Overall, theory agrees with HATS on VETO deadtime loss to 20% relative. Xiaochao suggested presenting more detailed simulation results. For example, instead of one comparison of the total deadtime as on page 14, should present comparison of data vs. HATS for all tagger data (8 groups, narrow and wide, thus 16 sets of comparisons).
• Will use FADC data to model the rates of "GC no T1", "T1 no GC", and possible correlations. - see followup report on 4/4
• Kai reported on optics calibration, see log 86
• page 3: angle optimization for the good-tune sieve runs;
• page 4: zoom-in of dummy target sieve holes;
• page 5: same as page 4 but before angle calibration (page 3 vs. 4 shows the clear improvements on the edges);
• page 6: after angle opt, ran another iteration for ytarg;
• page 7: mis-tuned sieve runs, before calibration. reactZ does not show any foil and the sieve holes are completely messed up;
• page 8: review of angle optimization for good tune runs. The starting point is to use the reactZ 2D plot (right) to select a particular carbon foil, using such cuts will give better sieve patterns, allowing angle opt.
• page 9: for this mistuned sieve run, checked all possible 2D combinations, there is no clear pattern available for foil selection;
• page 10: tried using target_y vs. focal plane phi to select foils (left). The right plot shows the same 2D plot after 7 iterations.
• page 11: Sieve hole pattern for dummy target, after 7 angle opt iterations. Now they agree with the expected positions very well!
• page 12-14: quality check for angle opt. 
• bottom of page 14: several ways to proceed. We agreed that Kai can try approach #2. Approach #1 is NOT necessary anymore since Kai already had good calibration results;
• page 15: Some formula for estimating the Q2 error.
• page 16: other to-do's, including right arm optics - see followup report on 4/4
• Bob commented that this work is "heroic".
• Xiaochao will have a meeting with Nilanga to discuss the error on Q2 from optics - see followup report on 4/4.

• 03/21/20101
Present:  Diancheng Wang, Kai Pan, Zhiwen Zhao, Robert Michaels, Xiaochao Zheng
• Diancheng:
• Was thinking about the VETO simulation and realized an important fact: During the experiment, for the first part of running we saw large deadtime from tagger test (the deadtime is in the order of 100ns times the T1 rate and is the same for all groups, but it could actually be proportional to rate of (T1 no GC) or (GC no T1), not the actual T1 rate). At that time we concluded that the reason is because the T1 input to the VETO circuit is 100ns wide. So we reduced T1 to 20ns then input it to the VETO circuit. After this the tagger test no longer showed large deadtime. However, because T1 itself has 100ns deadtime (upstream of our PVDIS electronics), this 100ns deadtime is still present for physics signals. Diancheng will try to estimate the deadtime loss in theory and use the simulation to prove the results. FADC data will be used to establish the input model of T1 and GC signals (including rate, double-pulsing possibility, and pulse shape).
  
• Kai reported on optics calibration, see log 84,
• page 2: reminder of last meeting, conclusion was that phi_targ reconstruction was good, but y_targ is NOT;
• page 3: overview of how optics optimization is done. Target variables are determined from focal plane variables using the optics matrix;
• page 4: before y_targ calibration. Here the ± numbers are width (RMS) of the peak. The database used has Kai's VDC and BPM/raster calibration, but otherwise is idential to Kiad's (HAPPEX-III optics);
• page 5: after y_targ calibration. The resolution (indicated by the RMS) is much narrower and the peak position are more consistent with zero. THis is after one y_targ iteration. More iterations can be done easily but we do not think it's necessary.
• The results shown on page 4 and 5 were from the carbon multi foil runs at the first 12.9 deg setting of the LHRS (for this we had the HRS angle survey). Then LHRS was moved to 20 deg (also had angle survey), then to 12.9 deg again (no angle survey). Using exactly the same optimized database, now the foils have a consistent 3mm offset in reactZ, as shown on page 6 and 7. The difference between page 6 and 7 is that, the dummy target "invaded" the multi-foil during this period. In order to get good Gaussian peak of the multi foils, the dummy target was cut out on page 7 to allow better fits of the central reactZ.
  
• page 8: Kai suggested this 3mm offset is completely due to spectrometer mispointing and can be used to figure out the "D" for the 2nd 12.9 setting. The "D" is usually obtained through angle survey but we did not have survey for the 2nd 12.9 deg setting. He used the central foil's offset 2.5mm to estimate the "D".
  
• With the estimated "D", the reactZ for multi foils is shown on page 9.
• page 10: with the optimized y_targ optics database, now the strong dependence of reactZ on tgt_phi is GONE!
• page 11: sieve hole before y_targ optimization. Left is Kai's variables (which use only reconstructed phi_targ, not y_targ), and Right is Kiad's variables (using both reconstructed phi_targ and y_targ). Kai's plot shows better agreement with the expected positions than Kiad's. Last time this was the evidence that phi_targ recon is good but y_targ recon is bad, hence the reason for Kai to do y_targ optimization last week. Certain offsets were applied, which are actually different from the sieve position survey results from the HAPPEX-III period. (Kai what are the offsets here??)
• page 12: same plots as page 11, but using the y_targ optimized database. Just another proof that the y_targ optimization worked, since now both show good agreements.
  
• Plan:
• Kai is planning to work on the angle optimization, before moving on to calibrate the mis-tuned sieve slit data. Xiaochao emphasized the angle optimization should be treated as a practice, since the results are already pretty good (page 12 of Kai's slide) and we should focus on the mis-tuned settings.
• Xiaochao will discuss with Nilanga on details of how to estimate the Q2 error,including:
• It's important to understand the uncertainties due to the fact that we did not have any survey for sieve position and target. Right now what we are doing is sort of mixing together HAPPEX optics database and our own (vertex). So how much uncertainty should we quote on all contributions to Q2?
• Bob mentioned again the importance of starting the asymmetry analysis now. We concluded that Diancheng will start looking into this, focus on establishing all cuts and general criteria about what (cuts, corrections, etc) are good enough and what should be worked on, in parallel to his HATS work. Kai will catch up on this later, hopefully RIGHT after his qualify exam (end of April). The goal is we should have two independent analyses and should produce exactly the same results.
  

• 03/14/20101
Present:  Diancheng Wang, Kai Pan, Zhiwen Zhao, Bob Michaels, Xiaochao Zheng
• Diancheng updated on the trigger simulation, named HATS! see log 83, it included some general description of the simulation software. Some explanations for the slides:
• page 6-7: FADC signals were used to determine time constants for the lead glass signals. On the right arm, preshower time constant tau is much smaller than the shower tau. This agrees with our expections (since we know the PMTs are different). The "tau" histogram on the right hand side of page 7 shows the time constant (1=5ns)
• page 8: used data, found the shower cluster is only 1x1, not 3x3 as previously expected. This will not affect the simulation but is interesting to know.
• page 9:
• cables and intrinsic delays of the modules are treated the same way (using a time buffer inside the module class). Bob commented it's better to treat them differently, maybe by creating a "cable" class.
• what are still missing are falling and rising time of the digital outputs, and some characteristics such as double-pulsing resolution of the module (etc.).
• page 11: Input here is the total rate of the focal plane, and used data to get the lead glass signal (from ADC) and position of each event as inputs to HATS. The agreement of group rates and double-counting rates with data is a promising indication that the timing simulation is working. (although since the data are used as input, the absolute agreement is somewhat expected).
• Main progress this week is on tagger simulation, see page 12. There are some problems:
• both narrow and wide groups have smaller deadtimes than observed;
• wide path of groups 5-8 are smaller than 100ns, and narrow groups have negative tagger loss. Xiaochao does not think we observed negative tagger loss. This should be studied further. Perhaps adding more realistic digital signal shape and detailed module characteristics will help.
• page 14: A reminder of the T1/veto circuit and what happened during the 1st period of running (wrong T1 input causing large deadtime). This is to be added to HATS.
• Kai reported on the optics calibration, see log 82:
• page 2 is a nice plot to show the coordinate systems used and how Zreact is calculated;
• page 3-7: raster x and y are swapped in the databased since 2003, however many people are still using the wrong map. This has been corrected;
• page 8 and 9: reactZ spectrum of multi-carbon target, Kai's (page 8) vs. Kiad's (page 9). Kai's has the updated VDC, and raster/BPM calibration, and the results have better resolution. Both have problem for the side foils.
• page 10: Kiad's sieve plot and the variable used. Kai pointed out it's important to figure out where the problem is coming from, whether it's phi_targ or y_targ, that is causing the disagreement in the horizontal direction.
• page 11: Zreact has strong dependence on phi_targ! Kai used some cuts (red boxes) to select particular foils
• page 12-13: Kai used different variables here to make the sieve plots, that are only sensitive to phi_targ but not to y_targ. page 12 is for the foil at 0cm and page 13 is for the invaded dummy at -12cm. The same arbitrary offsets (move the data to the right and down by 1mm each) were used for both foils. The agreement with expected hole position look pretty good. This indicates the problem is in y_targ reconstruction, not phi_targ.
• Summary and plan:
• will try y_targ optimization and see if it produces good results for each foil. (Xiaochao commented that should not spend more than 2-3 days on this).
• If y_targ optimization is successful, proceed with it. If not, back off to HAPPEX optics database. Either way, need to come up with a table of systematic uncertainties that we can use to estimate the error on Q2. Here, the error will be dominated by the main offset of the holes (angles) and foils (Zreact). The effect of offsets of particular holes on the edge (while if all other holes coincide with the red grid) will be small. Bob mentioned we can estimate this by: error = (error of the wrong hole) weighted by (# of wrong hole)/(total # of holes).
• Should contact Nilanga to see what data we can use (surveys? optics?) from HAPPEX and the uncertainties of it.
• Should proceed to optics calibration of the mistuned settings ASAP.

• 03/07/20101
Present:  Zhiwen Zhao, Diancheng Wang, Kai Pan, Xiaochao Zheng
• Diancheng:
• Update on trigger simulation, see log 81: (right arm)
• What are included?
• T1/veto circuit not included yet
• originally, the plan is to use HAMC to generate the preshower/shower position of events and feed them to the trigger simulation, but is not easy to do because HAMC uses a uniform event generator and use the cross sections as weights. Now, instead is using data as an event generator for both the shower/preshower position and the lead glass signal. Here the lead glass signal is simulated using  the real ADC spectrum of data with some assumption on the time constants of the signal.
• got the deadtime loss for the global triggers for the first time!
• Also simulated the pileup spectrum of taggers and the results look very similar to data. (although no absolute comparison with the data were presetned because the simulation used much higher tagger and event rates to save time).
• Xiaochao suggested adding the T1/veto circuit which are very important for the first period of running (when the T1 input to the veto circuit was 100ns and caused the deadtime loss of each group to go with the total trigger rate instead of the group rate).
• Kai
• still learning optics calibration, see log 80.
• focused on VDC and BPM calibrations
• For survey results, not sure if the results "HRS is missing the target center by xxx-mm upstream" is measured in the hall coordinates (Jin/Vince/Bob) or spectrometer coordinates (Kiad/Nilanga/Chris). (Note: later found out it is indeed in the spectrometer coordinates). Once this is clarified, can start doing the calibration! Answer: it's in the spectrometer coordinates.
  
• Zhiwen
• analyzed asymmetries from resonance kinematics. 
• Xiaochao suggested plotting the asymmetry of each group vs. W or Q2.

• 02/28/20101
Present:  Diancheng Wang, Kai Pan, Xiaochao Zheng, Zhiwen Zhao
• Diancheng:
• still working on the trigger simulation
• Kai
• learning how to do optics calibration, including raster/bpm calibrations.

• 02/14/20101
Present:  Kai Pan, Xiaochao Zheng, Paul Reimer (off-site)

• not much to report because not many people were at the meeting and we had problems with the computer in the meeting room.

• 02/07/20101
Present:  Diancheng Wang, Kai Pan, Xiaochao Zheng, Zhiwen Zhao
• Diancheng:
• reported progress on HAMC fine tuning, see log 76.
• what was done? previous problem is tails on focal plane x (bottom right of fig 10 in log 66) and target dpp (3rd row, middle plot of fig 10 in log 66). Cutting on this tail reveals they are coming from top edge of focal plane (last plot of fig2 in log 76, red are tail events, black are all simulated events). Cutting on focal plane x improves the tail significantly, see Fig.1 in log 76, focal plane x is 3rd row right plot in Fig.1, target dpp is 2nd row left plot in Fig.1.
• Also tried to change the dipole exit cut because Fig.4 of log 66 showed it could be improved (bottom and left side). New cuts are shown in Fig.3 of log 76. However, the trapozoid should be symmetric and the left side cut seems to create problems for focal plane phi (3rd row, middle plot of Fig.1 in log 76)
• Suggestions: Noticed beam profile does not agree with data. Perhaps should use the real bpm x or y shape (the one we always see in the counting house with spot++) instead of square functions.
• Suggestions: change dipole exit trapozoid left side back to be symmetric with the right side
• Then, wait for the final optics database before doing any fine tuning.
• reported progress on trigger simulation, see log 77. It looks great! We can simulate any electronics with this! To-dos are as follows:
• Timing of input pulses should follow Poisson, not uniform distribution
• Magnitude of input pulses should follow distribution of electron energy deposit in lead glass
• Add OR of 6 (or 8) groups
• Combine with HAMC (or sth. else) to allow overlapping
• Add VETO circuit
• Kai:
• used Gaussian fits and area method to extract pileups, results are not too different, did not post the numbers.
• report on pion contamination, see log 78
• Xiaochao commented that pion contamination should contain two parts: One is from intrinsic efficiencies of cherenkov and lead glass, which is what Kai did before the experiment. 
• Suggestions: Kai should run the intrinsic analysis for runs throughout the experiement running period to make sure there is no significant changes of detector performance
• What Kai did in log 78 is pion contamination due to pion-electron coincidences. The way Kai extracted this ratio in log 78 also include those caused by detector intrinsic effiencies. 
• Suggestions:
• For the N1/N0 ratio extracted in log 78, should think about all possibilities/cases, then
• calculate cases caused by detector intrinsic efficiency (using intrinsice efficiency analysis)
• estimate possibility of cases caused by timing coincidence, identify whether they affect PVDIS DAQ triggers (our main production DAQ)
• Xiaochao
• At the SOLID meeting it came to my attention that we have not done any work on internal radiative correction yet. A talk by Lorenzo on the SOLID rad corr can be found here. Paul will talk to John A. about this and see how to proceed.

• 01/31/20101
Present:  Diancheng Wang, Kai Pan, Xiaochao Zheng, Zhiwen Zhao, Paul Reimer (off-site)
• Diancheng: Still working on the trigger simulation software, will report later.
• Kai: worked on electron pileup fraction, see log 75. Comments:
• instead of using a sharp cut on PS+SH, can fit the two peaks (main electron peak and pileup peak) by Gaussians and use the are of the Gaussian to extract the pileup fraction.
• Should not use 1D Preshower cut.
• We discussed about optics calibration (last report on this was 10/22/2010, see minutes below). Although the error on Q2 should not be large, we are at the stage of fine-tuning the HAMC acceptance cut and any mis-calibration in optics  could be relevant here.  Kai has agreed to learn optics calibration from Jin and Vince and will take over this task from Kiad.
• Xiaochao
• At the SOLID meeting it came to my attention that we should double-check our radiative correction. A talk by Lorenzo on the SOLID rad corr can be found here. Paul will talk to John A. about this.
  

• 01/24/20101
Present:  Diancheng Wang, Kai Pan, Xiaochao Zheng, Paul Reimer (off-site)
• Diancheng: working on a full trigger-simulation package. Everyone is curious how he manage the storage vs. speed problem for the timing simulation, but sounds like it's not a problem at all in his new approach.
• Kai: worked on putting the electron efficiency into HAMC. See log 74 for preliminary results.
• It's a good thing that the asymmetry does not change much from low to high rate.
• The error here are NOT the error from HAMC, though. Diancheng will add the proper error to HAMC.
• Suggest Kai adding the error on asymmetry due to uncertainties in electron efficiency, using error propagation.
• Followup to Kai's report at the last meeting: Kai got some updated numbers on the high rate electron efficiency. With the TDC cut the numbers are much more reasonable (no log here). The high rate runs' efficiency are actually not too different from low rate runs.

• 01/10/20101
Present:  
• Diancheng: thinking about implemeting radiative corrections. See log #72. Basic idea is there. Need to implement more "toy models" for the resonance asymmetry.
• Kai: worked on high rate runs' electron efficiency. See log #71. Xiaochao commented that a TDC cut should be applied to the main TDC peak. Missing this cut might explain why the high rate runs have double the efficiency, since a lot of grass events are included and they are not proportional to the rate. (Rather, they are rate squared terms and the ratio of N1/N0 does not give the efficiency.
• Xiaochao

• 01/03/2011
Present:  Diancheng Wang, Kai Pan, Zhiwen Zhao, Robert Michaels, Xiaochao Zheng, Paul Reimer (off-site)
• Kai
  1. went to New York over the holidays
  2. presented using printouts, change of electron PID efficiency from low rate to high rate runs. Found high rate runs have lower efficiency, opposite to what we expected.
  3. Xiaochao suggested:
  1. We expected high rate runs to have higher efficiency because the ADC spectrum shift to larger values due to baseline drifts from high rates. So the first thing to do is to make the ADC spectrum for low vs. high rate runs. Conditions are:
  • plot ADC sum of 8 blocks, PRL1 vs. PRL2, for each group, with Cerenkov cut >1000. COmpare this spectrum for low vs. high rate runs.
  2. Cherenkov spectrum itself could have shifted due to the same reason, causing more pion contamination in N0 (# of electron samples). Should make the Cherenkov spectrum for low vs. high rate runs.  In principle, both N1 (numerator) and N0 (demoninator) are mixtures of electrons and pions.
• Xiaochao (late entry): commented on the importance of keeping the meeting minutes up-to-date. In a recent article by a UVa faculty member, he said if he organized everything of his daughter's wedding, but was hit by a truck in the morning of the wedding day, someone can pick up his notebook and run the wedding for him. -- Meeting minutes should be like this.

• 12/17/2010 
Present:  Diancheng Wang, Kai Pan, Xiaochao Zheng, Zhiwen Zhao
• Diancheng:
  • posted the talk at the Hall A collaboration meeting at: PVDIS log 69
  • made a master-plan for the asymmetry analysis, see PVDIS log 70. Will split this job with Kai (one for half of the parity runs, with small overlap between the two for cross checking).
• Kai
  • Reported analysis on the high rate PID in PVDIS log 71
  • Xiaochao commented that the cut on TDC (PVDIS group electorn triggers) should be on the main peak only, while Kai used a very wide range. This is because the "grass" area is due to accidental coincidences and are proportional to rate^2, not the rate itself. Since N1 has the cut and N0 does not, including the grass in N1 will cause the efficiency N1/N0 to increase at high rates due to higher ratio of accidental coincidences to true event rates.

• 12/03/2010 
Present: Diancheng Wang, Robert Michaels, Zhiwen Zhao, Xiaochao Zheng, Kai Pan (off-site) 
• Diancheng and Kai are working on the Hall A annual report as well as the oral report for the collaboration meeting (Dec. 10th 9:30am).
  
• Kai
  • reported on the structure of TDC spectrum PVDIS log 67:
    
  • slide 4: will calculate cnt2 and see if the rate agrees with the grass rate in the spectrum
  • slide 7: are T3, T4 and T5 analog signals or digital? (Ask Ramesh)
  • Kai thinks the problem is in the structure, Xiaochao thinks the problem is in the grass. Should do some quantitative work to see.
  • reported on PID shift and pileup from low to high rate data PVDIS log 68:
    
    • slide 4: Xiaochao commented the definition of N1 is not appropriate here, since the software cut automatically includes some efficiency already. To reject background (junk) events from the electron sample, can try to put a very low cut on PS+SH, or on the total ADC sum of all blocks for PS+SH, not the grouped ones.
    • slide 5: Xiaochao commented that since there is almost no difference in spectrum shape from slide 4 to here, it indicates that the software PS+SH cut (800) is higher than the hardware discriminator threshold, which is another problem.

• 11/19/2010 
Present: 
• Diancheng reported more on HAMC simulation PVDIS log 66:
  • Noticed the acceptance range are different between data vs. HAMC, starting from the collimator through all intermediate planes. Xiaochao asked John Lerose after the meeting, and John said if we used the HRS transport functions to plot the data at intermediate planes, it does not guarantee the correct distribution. This is a known problem of HRS transport functions. One way to solve this is to use data to determine the acceptance cut for simulation, for example by observing where the data drop to 50% (or even 90%) of maximum. - still need to be done
• Kai
  • 
    
• Xiaochao agenda:
• Plan for asymmetry analysis?
  
• Diancheng's Compton "wrap-up"?
• Diancheng's HAMC:
•  acceptance tweaking?
• Will discuss how to process with radiative corrections
• Resonance data/mistuned magnets: add John's functions to HAMC. waiting for Kaid/Nilanga for optics.
  
• Kai's trigger analysis:
  
• Effect of including the 4 other peaks in TDC counting?
• We already saw that including the "grass" of TDC in counting does not give correct results. But do we understand why?
• Effect of ADC pileupe on PID efficiency, how much did it change compared to low rate?
  
• Other to dos: effect of PID shift <-> HAMC asymmetry?
• Have we incorporated tagger deadtime with trigger simulation?
  
• Should come up with a plan about deadtime/pileup correction for global electron triggers.
  

• 11/15/2010 (Monday)
Present: Diancheng Wang, Kai Pan (off-site), Zhiwen Zhao, Xiaochao Zheng, Robert Michaels, Paul Reimer (off-site). 
• Diancheng
  • Posted PVDIS log #63 on HAMC vs. data with collimator included. The agreement is much better. Bob commented that arbitrary shifts could be added to the fp_x and fp_y in the simulation to match data.
  • Was planning to use HAMC to simulate the overlapping ratio, but this requires incorporating lead glass responses and simulating the showering process, as well as timing alignment. We agreed that this is too complicated to do and is also not necessary for the analysis purpose.
  • Will continue checking the acceptance to see if the agreement between HAMC and data can be better.
    
• Kai
  • Posted PVDIS log#64 on trigger simulation:
  • The main progress this time is on how to use high rate TDC data to obtain the "K" ratios. Now instead of counting all entries in the TDC spectrum, count only entries within the "main peak", see page 3 and 4.
  • Some structure in the TDC spectrum is visible on page 3. Diancheng has a note from last year to explain what they are. He also commented that all structures should be included in the counting. This should be studied as part of the systematic uncertainties of Kai's work. see followup on 12/2 - still need to be done
  • Using the main-peak-only counting, the K ratios obtained from TDCs give very good agreement with parity scaler rates, see page 5.
  • From last week's minutes, we concluded that even if high-rate TDC give consistent results with parity runs, it is still necessary to study how the ADC spectrum shift from low to high rate data in order to study pileup effects and shift in PID efficiency. Page 6 is intended to start this but we will discuss this in details next time. - still need to be done
• Bob commented that since HAPPEX-III and PREX are both doing asymmetry analysis now, it's a good time for us to go through all parity runs and extract asymmetries. THis include dithering analysis (ready) and cut study (to be studied). This can be done in parallel with HAMC and trigger simulation and both Diancheng and Kai should be involved. - still need to be done
• Xiaochao

• 11/05/2010 
Present: Diancheng Wang, Zhiwen Zhao, Kai Pan, Xiaochao Zheng, Bob Michaels (off-site), Paul Reimer (off-site) 
• Diancheng:
  • Posted PVDIS log #62: Used the older Fortran-based simulation but there is little difference between Fortran- and HAMC. However, by comparing phi_target distribution for different y-target slices, found there are obvious cuts in data. Going back in halog found the left arm probably had the collimator in, while HAMC currently has open collimator. Will run HAMC with collimator in and report next time for possible progress.
  • Was suggested to "wrap up" Compton results. The missing piece is sources of different systematic uncertainties and their sizes. - still need to be done
• Kai
  • Posted PVDIS log#65 (late entry) on trigger simulation:
  • Page 2: missing block plots for 3 typical runs: low rate, moderate rate Al dummy, low rate after zeroing-offset;
  • Page 3: How missing blocks show up in entry # of ADC spectrum between low rate 25886 and mod rate 25620 (Al dummy). the numbers show entries of block3:4 and block 7:8 (block 4 and 8 are obvious missing from page1)
  • Page 4: Similar plot but between low rate 25886 and high rate 25887. Here, because cannot use position cuts to select block, simply cut on ADC value and compare entry #. Same conclusion as page 2: that entry ratio of block 3:4 and 7:8 changed from low to high rate, although other conditions were identical since these two runs were back to back. It is obvious that this is due to baseline drift at high rate.
  • Page 5: How the problem shows up as PID holes
  • Page 6: ADC sum of all blocks of PRL1 vs. PRL2, for low rate (red) and high rate (blue) runs. From low rate to high rate, two factors are clearly seen which can change the PID efficiency: a) Pileups, and b) baseline shift. The shift in ADC is about 100 channels.
  • Page 7: If we use the TDC count from high rate HRS runs and scaler count from parity runs, in principle we should not have a problem even though there is baseline shift and pileup etc. On this page, black(red) numbers are results from using low (high) rate HRS runs. So high rate HRS runs are used correctly.
    
  • Page 8: Kai had the question why we see a 5us-window in TDCs even though they were programmed to have only 1.5us windows (1.5us from Bob checking the CRL code). - still to be studied
  • Suggestions:
  • try to use ADC spectrum to get the PID shift due to pileupe and baseline. Plots to use are similar to page 6, but only plot ADC sum of the 8 blocks of a particular group - still need to be done
    
  • use HAMC, study how much the simulated asymmetry change from the best to the worst PID efficiency (holes) - still need to be done
  • Continue study with how to use high rate HRS/TDC data -in progress 11/15
  • Note (after the meeting): Even if we can get the K ratios to work for high rate HRS vs. parity runs, we still need to know how much PID changes from low to high rate runs - still need to be done
• Sasha: the Moller summary for PVDIS is ready and posted on the Moller web page:
    http://www.jlab.org/~moller/e08-011.html
  The Moller systematic error is posted here:
    http://www.jlab.org/~moller/plots/moller_systematics.pdf
• John Lerose produced HRS transfer functions for the mis-tuned magnet settings: http://hallaweb.jlab.org/news/minutes/tranferfuncs.html Right arm results will be posted soon.
  

• 10/29/2010 
Present: Diancheng Wang, Zhiwen Zhao, Kai Pan, Xiaochao Zheng, Paul Reimer (off-site) 
• Diancheng
  • We decided to let Zhiwen and Kai working with Diancheng on HAMC
    
• Kai
  • Click here for directory of this week's report. Files are explained as follows:
  • 1029trigger.txt: Previously, found that if using low rate HRS and the high rate parity runs together to get the "K" ratios, the results are not good for gr 2 and 6 (left arm), and suspected deadtime loss to be the problem. But  this time, tried to use low rate 26051 and high rate parity 14480, the "K" ratios are actually stable for all 6 groups. The overall rate also agree within sub-%. So the reason of previous bad "K" ratios were not due to deadtime loss;
  • The problem is probably in the "missing block" problem that occurse late Nov-early Dec last year. The following plots show the shower block for 3 different runs:  Al dummy 25620.pdf (moderate missing block problem), missingblock 25886.pdf (severe missing block problem), and production run 26051.pdf (new kinematics, missing blocks fixed by zero-ing the sum8 output offsets).
  • When Sum8 has a positive output offset, it is more difficult for ADC signals to pass the discriminator threshold, hence cause "missing blocks" in the final trigger.
    
  • The problem is less severe at higher rate because higher background raised the baseline of the ADC signal of these blocks, and make them easier to pass the discriminator threshold. THis was concluded from the fact that Al dummy runs (VDC on, moderate rate) shows less problem than low rate HRS VDC on runs taken at similar time.
    
• Xiaochao

• 10/22/2010 
Present: Diancheng Wang, Zhiwen Zhao, Kai Pan, Robert Michaels, Xiaochao Zheng, Paul Reimer (off-site) 
• Diancheng
  • updated last PVDIS entry to correct a typo in the plotting script. The correct transport x vs. y distribution for events rejected by the Q3 out cut is shown now in Fig.9. Momentum distribution for the same events are plotted in Fig.10.
  • Xiaochao commented the Q3 out cut in Fig.8 still does not look correct.
  • Will post plots for target_y and dp as well.
    
  • Posted an entry for the dithering/regression analysis. Generally, everything looks as expected and the effect on the asymmetry width after correction (vs. before) is small. The only question left is how to take into account the correlation between the 2 different BPMs to avoid over-correction.
    
• Kai
  • will report next week
    
• Xiaochao reported for Kiad and Nilanga on their optics work. Kiad posted a HAPPEX entry. Overall, using HAPPEX database:
  
• sieve slit pattern looks reasonable for the "good tune" runs (top figure on page 2). Some holes have ~1mm of shift and can be made better. This 1mm shift in horizontal direction correspond to about 1mrad error in angle (horizontal spacing of slive holes is 1.25cm and vertical is 2.5cm, the 1mm shift is estimated from the plot, distance between sieve slits and target center is 1.18m and (1mm/1.18m)~1mrad). The effect on Q2 is <1% (similar derivation for mispointing is given below).
  
• will do calibration also for the two "mis-tune" settings (figure on page 1, and bottom figure on page 2).
• Vertex reconstruction:
  
• looks good for LHRS (bottom figure on page 3). Nilanga said what's important here is the central foil since it indicate mis-pointing of the HRS;
• For RHRS there is a systematic shift towards the downstread side (bottom figure on page 4), corresponding to about 1mm of mispointing. Divide it by the distance from the HRS to the target center (1.18m) this corresponds to about 1mrad in angle. Compared to 20 degrees this is a 0.29% effect. On Q2 the effect would be therefore 0.58%.
  
• We should ask John LeRose for transport function models for the two mis-tuned settings, to use in HAMC.
  

• 10/18/2010 
Present: Ramesh Subedi, Diancheng Wang, Zhiwen Zhao, Xiaochao Zheng, Paul Reimer (off-site), Kai Pan (off-site) 
• Diancheng
  • working on dithering and regression analysis, will report on Friday
  • working on HAMC. Comparison of MC vs. data for theta_tg and phi_tg slices can be found in this entry. Similar work for PREX by Bob can be found here. Unlike PREX where the acceptance is problematic on the edge, we seem to have problem in the center of phi_tg.
  • In the same entry, found (in figures 7 and 8) that the Q3 out cut might be doing the wrong thing, although it's not clear how it is wrong in HAMC or if it's an artifact caused by the plotting script.
  • Xiaochao suggested making the reverse plot of Figures 7 and 8 to show what events "pass thru these cuts" instead of being rejected. Also can make a plot similar to Fig.6, but using x_trans and y_trans, to see if cut "Q3 out" rejects events in the center and how much it rejects.
    
• Kai (late report for 10/8 meeting) reported on trigger simulation. See this entry.
  • included TDC multihit in analyzer;
  • with the multihit correction, used low rate HRS run to get the overlapping ratio and analyzed the rate of the corresponding low rate run. The agreement between (rate total from data) and (from MC) is 99.74% which is within error bar of 100%;
  • Did similar analysis, using low rate HRS overlapping ratio (same run as above) to analyze the rate of a high rate run. The K values now vary greatly between the 6 groups shown. The agreement between data and MC on the total rate is 94.8% (bad);
  • Suggested that group 6 could give us a hint on what's going on: Group 6 is the only one where overlapping counts with the adjacent group (overlap between gr5 and gr6 here) are much higher than the count of that group alone (Group 6 only). Could this cause the loss of the group to deviate from the normal deadtime we expect?
  • More suggestions: should use corresponding high HRS run to see how the ratio changes compared to the low rate HRS run. Complication from TDC multi-hit could occur here. Must find a way to correct the complication.

• 10/08/2010 
Present: Ramesh Subedi, Diancheng Wang, Robert Michaels, Zhiwen Zhao, Xiaochao Zheng, Paul Reimer (off-site)
• Diancheng
  • still could not find a solution to the Compton problem. 
  • We discussed the vacuum endcap alignment. Since the photon angular distribution is in the order of <400micro-radians from the electron beam and the distance from the CIP to the endcap is about 6m, the photon spot size at the endcap should only be a couple of mm in radius. Going from HAPPEX energy to PVDIS energy changes this size but won't make it larger than 3mm in radius. Since the window (0.5mm steel) on endcap (up to 7mm steel) is 5mm in radius, misalignment of the photon w.r.t. to this window is unlikely.
  • Will work on dithering to take a break from Compton;
  • Will also work on acceptance in HAMC using Bob's method, see below.
• Bob:
  • Acceptance problem in HAMC for PREX was solved by taking slices of phi_tgt and compare theta_tgt distribution between HAMC and data for each slice, or taking slices of theta_tgt and compare phi_tgt. Found the problem is solely on the edges, while for central region the absolute rates from HAMC agree with data. Now using an "emperical acceptance" to solve this problem.
• Suggested Diancheng to write a MENU10 proceedings.

• 10/01/2010 
Present: Ramesh Subedi, Robert Michaels, Diancheng Wang, Zhiwen Zhao, Kai Pan (off-site), Paul Reimer (off-site), Xiaochao Zheng (off-site).
• Ramesh is working on MENU10 contribution.
• Diancheng is working on reproducing HAPPEX results, will post results on PVDISLOG.
• Kai
  • trigger simulation report.
  • suggestions: list the "ratios" used for the low and high rate runs on page 6;
  • comments: 
  • deadtime from tagger runs on page 3 was from "after T1 correction". If use Q2=1.1 runs as shown on page 6, should use tagger runs before T1 correction. "T1 correction" happened between parity runs 14236 and 14237.
  • Do we understand the results on page 6?
  • Overall it's going in the correct direction. Next step will be VDC multihit analysis as shown on the last slide.
  • reported scaler has "grass". Will post a plot on ELOG so we can discuss from there. - still need to be done
• 
  

• 9/24/2010 
Present: ...
• Diancheng
  • still working on Compton. thickness of the lead shield was 0.3cm for PVDIS (comfirmed by Megan). Vacuum endcap is 0.8mm steel. The simulation still shows roughly a 0.4cm lead shielding and the inconsistency has to be found elsewhere.
  • Megan reported at the Compton meeting a modified algorithm for pickup correction, but this does not seem to affect ours much.
  • Bob suggested Diancheng run the simulation for HAPPEX and see if can produce Megan's result;
  • Xiaochao suggested adding misalignment of the photon beam and the vacuum endcap window, since the steel endcap thickness increases dramatically outside the small window.
• Kai
  • working on trigger simulation, will have results by Monday.

• 9/17/2010 
Present: ...
• Diancheng reported more on the Compton simulation, click here for report.
  • measured thickness of the lead shield (0.2cm) and size of the collimator hole (0.5cm) with a caliper. Note that from simulation the lead shield looks like 0.4cm. Xiaochao commented should quote more sig. figures in the measurement;
  • also found a vacuum cap in front of the lead shield which looks like 1-cm thick steel.
  • best match with data is obtained using 0.2-cm lead shield, 0.5-cm collimator hole, and a 0.8-cm steel vacuum cap.
  • Bob suggested looking for engineering drawing of Compton for the vacuum cap thickness.
  • Diancheng suggested might be able to measure the cap thickness if the Compton vacuum is broken for other reasons.
• Kai reported on his trigger simulation, click here for report.
• Xiaochao is working on the NIM draft.

• 9/10/2010 
Present: Kai Pan (off-site), Robert Michaels, Zhiwen Zhao, Diancheng Wang, Xiaochao Zheng, Paul Reimer (off-site)
• Diancheng continued his analysis on Compton. Click here for his elog. He found that:
  • adjusting the lead (plate) shielding thickness from 3 to 5mm affects significantly the simulated Compton spectrum. A 4mm thickness could agree with data very well.
  • The size of the collimator (radius), on the other hand, does not affect the spectrum as much.
  • The GEANT simulation stops at the ..., it does not include the light collection efficiency or Gaussian smearing of the PMTs.
  • Suggestions:
    • Track down the lead shielding used during the experiment and measure it to better precision;
    • Try 4mm shielding (or some other value), adjust the smearing as well, to find the best match with data;
    • Bob suggests if it's possible to add the light collection efficiency in the simulation;
    • Xiaochao still suggest adding a non-symmetric smearing by hand for estimating the systematic uncertainty. The exact reason for non-symmetric smearing can be studied in details later;
• Kai:
  • reported on his bull's eye scan results; Unfortunately during that time the MCC could not steer to beam to negative x. In principle, since the fits are linear, if the equipment acts linearly then the calibration should work as well for -x.
  • Nevertheless Kai is suggested to look at the HAPPEX-III bull's eye scan results.
  • Also started to think about trigger simulation. Click here for his plan.
• Xiaochao is working on Ramesh's draft NIM paper.

• 9/3/2010
Present: Kai Pan, Zhiwen Zhao, Diancheng Wang, Xiaochao Zheng, Paul Reimer (off-site)
• Diancheng continued his analysis on Compton. He found that adjusting pileups only change slightly the simulated spectrum and there is still significant difference between data and simulation. Smearing also does not solve the problem. Xiaochao suggested 1) changing the shielding; 2) Add a non-symmetric smearing just to see how it changes the simulated spectrum.
• Diancheng will also try to look at the "beam off, laser on" background to see if it affects the data.
• A HAPLOG on Compton analysis with smearing added to the simulation can be found here.


• 8/30/2010
Present: Kai Pan, Zhiwen Zhao, Diancheng Wang, Xiaochao Zheng
• We discussed how to divide major tasks between Kai and Diancheng:
• Kai will take over trigger simulation;
• Kai will learn HAMC from Diancheng. Since we still have problem with this simulation, it's important that both master this software and look for a solution.
• Kai will start bull's eye scan analysis as inputs for Kiad/Nilanga's optics analysis.

• 7/21/2010
  Present: Ramesh Subedi, Diancheng Wang, Xiaochao Zheng
• Diancheng reported on his Compton analysis (very preliminary):
1. page1: 
1. in the simulation Compton events are generated according to the cross section as function of the scattered photon energy (top left plot);
2. Compton asymmetry is calculated as a function of photon energy (top right plot shows asymmetry vs. Ephoton);
3. PMT signal spectrum (middle middle plot) is simulated using the generated Ephoton (middle left plot) and the PMT response (middle right plot). The PMT response is an input to the simulation and could be from PMT bench-tests;
4. energy loss of the photon can be simulated using corresponding shielding conditions of the actual run.
5. overall the integration method should have an asymmetry of 0.0493 for PVDIS, as shown in the bottom of this page.
2. page 2: if using 0.0493 as the integration asymmetry, can extract the beam polarization from accumulator 0 data. The right plot shows that Pb=90.6%.
3. page 3: the simulation can also be compared to the trigger data. Here a comparison of the photon energy spectrum is possible. The Compton photon spectrum from the trigger data is shown as blue in the two plots on the left, as well as the asymmetries (bottom left plot). The Compton spectrum and asymmetries from the simulation are shown on the right.
1. in order to compare with data, a pileup effect is added to the simulated Compton spectrum. The size of the pileup is estimated using snapshot data (but with limited statistics).
2. some visible difference can be noted in the Compton spectrum. The simulation does not have a threshold at the low energy. And the shape of the peak is slightly different.
4. page 4: by fitting the Compton simulated spectrum to the data and use this to correct the absolute scale of the spectrum, can obtain the beam polarization from the asymmetries (right plot). Results are shown in page 5.
5. page 5: The beam polarization from the trigger data is 81.85%, in significant disagreement from the integration method.
6. Suggested plan on this analysis:
1. This is already very good progress, but effort should be taken to make the two analysis agree.
2. on the integration method:
1. the simulation is done using a single PMT HV + shielding configuration, while both were changed several times during the experiment. Need to run the simulation according to the real config and compare with the corresponding subsets of data.
2. can use snapshots data to obtain a Compton spectrum of the integration data and compare it with simulation. This is a sanity check.
3. on the trigger method:
1. simulation can be done with higher statistics to obtain a "smooth" Compton spectrum;
2. need to use snapshots data to study the effect of pileup (idealy should have two Compton spectrums, with and without pileups, to show this effect); 
3. need to understand why the Compton spectrum is different in simulation vs. data, can this be explained by tweaking the pileups? can this be explained by PMT non-linearity?
• No report on the resonance hamc simulation except the correction to the transformation formula. With the updated formula there is no visible change to the phi/theta spectrum.

• 7/14/2010
  Present: Robert Michales, Paul Reimer (call-in), Ramesh Subedi, Diancheng Wang, Xiaochao Zheng
• Paul is reading Ramesh's NIM draft paper and will have comments.
• Diancheng continues his resonance simulations, click here for his weekly report. The most obvious disagreement right now is in theta_tg vs. phi_tg (page 3). Data shows a near-rectangle while hamc shows a trapezoid.
  • Bob suggested the following:
    • Look at sieve slits data and see if the slits line-up. If they show a slope, it could explain the disagreement;
    • Assuming the disagreement is due to problem with hamc, can tweak the acceptance such that it agrees more with data, and should estimate the uncertainty in asymmetries due to tweaking.
  • Xiaochao suggested moving on to setup the asymmetry model for resonance regions. Also need to read published data on PV asymmetries in resonances to see what contraints we should have on the model tweaking.
  • Diancheng pointed out a mistake he found in hamc for the transformation of Hall<->HRS(transport) coordinates. Click here for two emails explaining this.
• For Compton analysis:
  • Diancheng has setup the simulation code and will use it to calculate the expected asymmetry.
  • Bob and Diancheng explained that the fact we don't have accumulator 3 and 5 won't affect much the analysis, since the background in accumulator 0 can always be corrected using laser off data, and the asymmetry of the background cancel out when we combine laser left and right data.

• 7/07/2010
  
• no power in F-wing, meeting cancelled.

• 6/30/2010
  Present: Ramesh Subedi, Diancheng Wang, Xiaochao Zheng
• Diancheng started simulation for the resonance kinematics. click here for his weekly report on kine #5.
• suggestions: use zcol for the 2-dimensional plots. The disagreement between hamc and data might be due to only plotting in b/w.
• otherwise, the asymmetries don't look bad.

• 6/23/2010
  Present: Paul Reimer (call-in), Ramesh Subedi, Diancheng Wang, Xiaochao Zheng (call-in)
• Diancheng continued his Compton analysis, click here for his weekly report. It contains explanation for different accumulators, two analysis method (pair-wise vs. laser-wise), and preliminary asymmetries from each accumulator during PVDIS.
• We can only use accumulator zero (0) because of the incorrect threshold setting during PVDIS; For example there is no data for accumulator 3 and 5 (high threshold);
• Will start Compton simulation.

• no meeting on 6/2 (MENU10) and 6/9 (User's group meeting).

• 5/26/2010
  Present: Ramesh Subedi, Diancheng Wang, Xiaochao Zheng (call-in)
• Diancheng continues his rate scan analysis:
• For the rate scan itself, will focus on getting the error bars of p1,p0,b and the deadtime. The results could have so large error bar that they are not useful. But at least we will know if they contradict the expected deadtime or not;
• For the narrow-wide rate scan data, will include efficiencies of the narrow and wide group from fbTDC data, like using these plots from the online expert check.
• Diancheng is also starting trigger simulation. Xiaochao suggested talking to Bob or Paul King for general suggestions.
• Diancheng will give the update talk at the Hall A collaboration meeting.
• Ramesh and Diancheng will circulate their talk for MENU10 by this weekend. MENU10 will start next Wednesday.

• 5/19/2010
  Present: Ramesh Subedi, Diancheng Wang
• click here for minutes.
  • highlight: after talking to Paul King, we realized the large slope in rate scan data might be due to LD2 target density change. Suggestions from Paul including: finding other proof of density drop (such as using lumi), and check solid target data to get a rate scan "without boiling".
• suggestions from Xiaochao (emailed to Diancheng on 5/24, for reference to plots see minutes):
  • for the rate scan:
    • talk to some lumi experts, such as Luis or Bob, and see what's the best way to get "b". I suggested using the average of all 8 lumis, but it could be not the best method. They could also tell you how lumi linearity could come into play;
    • In table 3ratescan.xhtml, give error bars on everything, including "b";
    • You showed me what you can get by taking the difference between en and ew, and I see one plot, but Ramesh's minutes doesn't include it. Could you remind me what you expect to get from this analysis and post the results?
    • The aluminum data shows positive p1, any idea what can cause it? (I will also send an email to Bob about this).
  • Besides rate scan, I think you can use some of your time to do other tasks, such as COmpton. Another thing you can do is:setup simulation for the resonance settings, and run it with whatever resonsance model you have now (could mean using the DIS formula for asymmetry if you don't yet have a resonance model for the asymmetries). I'd like to see how the asymmetries compare with data for these settings. Of course to compare with data you need to find out which slug each setting corresponds to.

• 5/05/2010
  Present: Ramesh Subedi, Diancheng Wang, Xiaochao Zheng
• not much progress on the rate scan, suggestions from last week's meeting did not work.

• 4/28/2010
  Present: Ramesh Subedi, Xiaoyan Deng, Xiaochao Zheng, Robert Michaels
• Ramesh has a first draft for the NIM paper;
• Using the preliminary result of OLO2 vs. Unser calibration the rate scan results can be found in this directory, which use the RMS/sqrt(N) for the error bar and contains:
• T1 and GC rates vs. I
• comment: T1 seems to have reasonable deadtime (~100ns)
• GC appears to have much higher deadtime loss but this might be due to the high double-pulsing rate (we saw on the scope that the GC signals are 70ns wide and about 60% have double pulsing). At higher current more double-pulses will not be counted, which may result in higher loss.
• right arm electron narrow rates vs. I
• right arm electron wide rates vs. I
• all right arm groups appear to have several micro-sec of deadtime (-p1/p0^2);
• Bob suggested also plot the left arm results
• Xiaochao commented that one may try plotting PS or SH rates, instead of the narrow and wide rates. Although for right arm this may not cause a big difference.
• 4/21/2010
  Present: Diancheng Wang, Xiaoyan Deng, Xiaochao Zheng
• BCM calibration:
1. the BCM logging software had problem last fall and failed to log the information. (Though I was told it should have been fixed now).  As a result we need to pull out information on OLO2 and Unser from HRS data. - done.
2. Now the procedure will be: 
1. calibrate our BCM scalers with our UNSER scaler (2nd calibration run);
2. compare our UNSER scalers with HRS UNSER scalers and make sure they had *exactly* the same readings during the calibration run;
3. calibrate HRS UNSER with HRS OLO2 data and get the offset/linearity;
4. Eventually, will relate our BCM scalers with the OLO2 current for rate scan.
5. Alternatively, one could try to calibrate the HRS BCMs with the HRS UNSER, and compare HRS BCMs with our BCM scaler to make sure they are the same.
• Deadtime/tagger analysis: 
1. left arm (by Diancheng, from last week): group 3 (with pileups and tagger loss plotted separately); typical pileup TDC spectrum for narrow paths from a particular tagger loss run; and deadtime loss of all groups.
2. right arm (by Xiaoyan): group 4 (with pileups and tagger loss plotted separately); typical pilup TDC spectrum for narrow and wide from a particular tagger loss run; and deadtime loss of all groups. 
3. Comments: 
1. overall the tagger loss measurements are consistent with expectation that wide path should have ~100ns deadtime and the narrow path's deadtime is dominated by the input signal width, typically 60-60ns. 
2. Three groups -- group 6 on left and groups 1 and 8 on the right -- do not agree with expectation. It looks like the input signal to these 3 groups are very wide, ~120ns, which dominates the deadtime of both narrow and wide paths. One could in principle check 1) the fbTDC spectrum of these groups and see if the timing spread is wider; 2) PMT signal widths in the hall and see if the blocks in these groups are wider than others.
3. We should in principle be able to understand the pileup result alone by looking at the timing structure of the tagger circuit (pictures for those were taken from the scope during the X'mas shutdown).
• Disk space (Diancheng): we may shrink the disk space needed by 2/3, but the required amount is still significant (4-5TB for HRS runs and 1.5TB for parity runs). Another solution is to replay only 200k of each HRS run if the whole purpose is to monitor the PID.
• Compton analysis: Diancheng is still looking into the analysis using "accumulator 0".
• NIM paper on DAQ: Ramesh will have a draft by next meeting.

• 4/7/2010

• General: 
• Xiaochao: will figure out with the computer center for how to purchase a ~20TB disk for our own analysis, the costs, etc.
• Xiaoyan:
1. Continue working on BCM calibration; still need to do the chi-sq and residue plot;
2. Working on the script for the rate scan analysis. Some results are posted here; Note: 
1. we found some typos in the code and these plots are currently being updated;
2. The "I" in the plots are not the real beam current yet, but rather the expected UNSER value calculated from various BCM data based on the BCM calibration;
3. the updated T1/I or GC/I plots show a possible un-corrected offset in BCM calibration. Need to investigate further.
• Diancheng:
1. Worked on Compton analysis, some plots can be found here. The plots show:
1. asym_hist.gif: Compton asymmetry vs. run numbers. Black is the average and green lines show a ±5% range.  todo: find out and mark on the plot all major "events" during this time, such as spot move or any obvious beam polarization change, and change to the Compton shielding.
2. asym_pol.gif: Compton asymmetry vs. cavity polarization. Suggest doing a fit and see if the Compton asymmetry is proportional to cavity polarization. 
3. laser_pol.gif: Cavity polarization from epics for 3 runs (separated by the vertical lines). Most of time epics was dead (flat-lines near 95% and 20%). Suggest plotting the good epics value vs. time and see how stable it is.
4. pulls.gif: pull plot of asym_hist.gif.
5. In general, need to have a complete history of spot move, Moller results, and Compton shielding changes, and mark them on all Compton analysis plots.
6. Now we only have the simulation to do before extracting the beam polarization from Compton data.

• 3/29/2010

• General: 
• work on the database is ongoing, with Xiaoyan focusing on HRS runs and Diancheng on parity runs.
• Both Diancheng and Xiaoyan will have their 3rd-year review at UVa on Apr 13th, Tuesday. Will focus on their talk this week. Will discuss about the slides later this week. We will have dry runs next week, probably during the meeting.
• Ramesh suggested a reading: G.F. Sacco's Ph.D. thesis "Hadronic Effects in PVES".
• Xiaoyan:
1. Here is the directory for BCM calibration that include 2nd-order fits and error bars (errors too small to be visible)
1. To do: suggest adding the chi-sq result and a plot of the residues
2. Working the rate scan code. (though some on-line code exists from Xiaochao, will write from scratch).
3. Will progress to tagger/pileup code later. (what's missing from these online codes are the pileup uncertainties).
   
• Diancheng:
1. Reading the Compton analysis code from Megan.
2. Probably ready to replay some of the runs now. Will start from testing batch farm scripts.
   
3. Asking what else needs to be done in parallel. Suggestions:
1. We had only DIS simulations, should run RES simulations for the real kinematics at which we took data; The first step is without modifying the asymmetry or structure function formula, i.e. using DIS formula whenever possible; (Ideally, the difference between the measured and the simulated asymmetries can only come from: the blinding factor, radiative corrections, and any resonance structure in the structure functions or asymmetries.)
   
2. Should also work on the rate simulation. We seem to have problem here (do not agree with older simulations and with data);
   
3. Then will proceed to adding RES models for the PV asymmetry.

• 3/22/2010
         Present: Diancheng Wang, Xiaoyan Deng, Ramesh Subedi, Paul Reimer (off-site), Xiaochao Zheng
• Xiaoyan:
1. Here is the directory for the initial BCM calibration. Note: 
1. We had two calibrations, 1st was run 14175 with the unser plugged into HAPPEX ADCs; 2nd had runs 14642 and 14643 (the two runs are mostly identical), where the unser was plugged into PVDIS scalers;
2. For each run, performed linear fits of bcmu1, d1, u3, d3 and happex bcm vs. unser;
To do: 
1. add error bars;
2. add 2nd-order fits to find out the non-linearity.
• Diancheng:
1. Copied all "pvdis" codes from adaq machines to ifarm (directory /w/halla/parity/pvdis/), and compiled all codes. This includes both pvdis/ under adaq and apar, but not copy each person's individual home directory.  The same codes are saved also to silo under /mss/home/dwang/pvdis_online/; Others who want to use the code can contact Diancheng for instructions.
2. Setup run database for the experiment. The web interface is https://jmysql.jlab.org/phpMyAdmin/ (on-site only) username is hapvdis, contact us by phone or in person for passwd. A basic "runlist" table is in place. Is working on compiling information from the paper runsheets, halog, and other records for all runs. Note: the "runlist" table is far from complete, but more info can be added easily using mysql (such as VDC, PID, FADC etc). Instructions on how to access the database using perl- or C-based codes will follow;
3. Talked to Megan about Compton analysis. Megan will also help replaying all root files using her mechinery. Main difficulties (different from HAPPEX-III) are:
1. Photon polarization during PVDIS was unstable, varies from run to run and up to 2%. Note: unstability can be dealt with, the main error on P_gamma should be from the photon polarimetry as long as we have enough data on P_gamma in EPICS;
2. Theoretical expectation of the Compton asymmetry: Simulation code to estimate this exists (from CMU, Greg Franklin), will learn from Megan. But the simulation depends on shielding and we changed shieldings very often during PVDIS. May need to run many simulations but should not be a major problem.
• Xiaochao:
1. Copied some online codes from adaq machines to ifarm (directory /w/halla/parity/pvdis_online_homedirs/, with a simple README file listing what is saved). This is complementary to Diancheng's backup described above. The same codes will be saved to silo under /mss/home/xiaochao/pvdis_online/
• Others:
1. Still need to talk to Rupesh on the dithering code;
2. Xiaochao suggested everyone setup a "weekly report" webpage, post plots online before each meeting. It's easier to organize in the long term and off-site people can follow at each meeting without too much difficulty.

• 3/8/2010
         Present: Robert Michaels, Diancheng Wang, Xiaoyan Deng, Ramesh Subedi, Paul Reimer (off-site), Xiaochao Zheng
• Ramesh:
• Both DAQs have been dis-assembled. All modules are in the test lab managed by Albert.
• Here are all the serial numbers of UVa-bought modules and NIM crates. We need to put lables on all UVa properties.
• BNL cables, should we return them to Mahbub? - yes, all labeled and returned;
• work on NIM paper is still underway, is taking much longer than expected. Will need deadtime results from others.
• PREX shift signup - Bob suggested emergency fill-in only. Xiaoyan and Diancheng will serve like "expert on-call" but are not expected to be the lead people;
• Qweak shift signup - we have committed 30 shifts/person, or 90 shifts total over the 2-year period;
• Analysis plan:
• Tasks that we can start immediately:
• copy all adaq and apar codes out from adaq machines. Xiaochao suggested using the home directory on jlabl1 to store all source code, and use the work disk to store in/output files and plots.
• organize runlist, establish a database for all runs - Diancheng will consult Alexandre for how to start. We should discuss what information to be included in the database later;
• Xiaoyan can start the work on BCM calibration, deadtime extraction from rate scan, and deadtime extraction from the tagger data;
• Xiaoyan and Diancheng will arrange a meeting with Rupesh to get his dithering code;
• Other:
• MENU10: Xiaoyan and Diancheng have submitted abstracts; Ramesh will give a "lead" invited talk.

• 1/21/2010
        Present: Robert Michaels, Diancheng Wang, Xiaoyan Deng, Ramesh Subedi, Paul Reimer (off-site), Xiaochao Zheng
• New meeting time. possible slots: Monday morning and early afternoon (must end before 4:30pm), Wednesday afternoons, Friday after 1pm;
• Diancheng/Xiaoyan: Progress on DAQ work, scanned mapping results, schedule, etc:
• Left arm DAQ mapping scanned results can be found here (copied 1/22/2010);
• Right arm DAQ mapping scanned results can be found here (copied 1/22/2010);
• All cable length, module delays have been measured, except right arm "white cables" between the patch panel and back rack;
• Racks will be taken out tomorrow, Jan. 22nd.
• Ramesh: Runsheets copies almost done, checklist (including expert checklist) still need to be picked up from the counting house.
• Re-visit of the task assignments:
  • Compton analysis: Diancheng with help from Megan;
  • Moller analysis: Sasha will do it in April, after analyzing the quasi-family and HAPPEX-III data;
  • BCM calibration: Xiaoyan;
  • Deadtime from rate scan data: Xiaoyan;
  • Dithering analysis. Xiaoyan and Diancheng should meet Rupesh in March to become familiar with the code. Then will have a discussion with Kent to know the physics background.
  • FADC data: Diancheng and Xiaoyan will first identify all FADC on runs, and analyze them later depending on what we need from them -- to be identified
  • Extract asymmetries: divide the whole data to 2 sets, each student develops his/her own code and can cross check each other;
  • Optics data: Nilanga and Kiad will analyze these data, expect results by end of Feb.
  • Deadtime (tagger) data: Xiaoyan;
  • Simulation of deadtime/electronics: to be identified;
  • Physics simulation: to be identified.
• Conferences: see this link
• Meeting schedule from now to March 1st: no meeting.  Our next meeting will be Monday, March 8th. 

• 01/12/2010
       Present: Ramesh Subedi, Kai Pan, Xiaochao Zheng, Robert Michaels, Diancheng Wang, Xiaoyan Deng

• Draft analysis flow chart: PDF, PNG
• Survey results from 11/03/2009-11/04/2009, and from 12/17/2009 (left spectrometer pointing only). A full list of Hall A survey results can be found here.
• All run sheets have been sent to copy center for making 7 copies. Ramesh will pick up shift checklists from the counting house this week.
• DAQ work:
• Kai is working on the left arm DAQ full mapping, and Xiaoyan and Diancheng are working on the right arm.
• Remaining work include: need to measure cable length for those without explicit delay labels (mostly customized cables)
• For cables coming from upstairs (leadglass), will measure 1-2 of them now, then double-check others once detectors are de-installed.
• Left pion narrow and wide for each group is 45ns, not 30 (or 20) ns. 
• will scan all mapping and post them later.
• This will be the focus of next two weeks, everything should be done before Jan. 25th.
• Divide up analysis tasks
• Compton analysis: Diancheng with help from Megan;
• Moller analysis: Xiaochao will talk to Sasha on how to proceed;
• BCM calibration: Xiaoyan;
• Deadtime from rate scan data: Kai;
• Dithering analysis. Ref: Kent, Rupesh;
• FADC data: Kai: identify all FADC on runs, and analyze them later depending on what we need from them;
• Extract asymmetries: divide the whole data to 3 sets, each student develops his/her own code and can cross check each other;
• Optics data: Xiaochao will follow up with Nilanga and Kiad.
• 
  
• Deadtime (tagger) data: to be identified;
• Simulation of deadtime/electronics: to be identified;
• Physics simulation: to be identified.



• 12/23/2009
      Present: 

• Took some pictures of the DAQ on both arms. Click here for a map.
• Left arm tagger measurement (JPG files, from top to bottom of the module):  1n, 1w, 2n, 2w, 3n, 3w, 4n, 4w, 5n 5w, 6n, 6w. (Note: the scaler map is wrong here. From top to bottom should be: 1n, 2n, 3n, 4n, 5n, 6n, 1w, 2w, 3w, 4w, 5w, 6w. So the name of these plots should be corrected.)
• Right arm tagger measurement (JPG files): 1n, 1w, 2n, 2w, 3n, 3w, 4n, 4w, 5n, 5w, 6n, 6w, 7n, 7w, 8n, 8w.