1. Melanie reported on livetime (elastic) and helicity-gated livetime (DIS) analysis. Something is not quite right: livetime for elastic or Delta 1-pass runs are higher than 100%. DIS helicity-gated livetime (A1n) fluctuate too much. Will ask Carlos to analyze a few runs and cross check.
2. Junhao started working on asymmetry extraction, suggested following proposal binning.
3. Mike posted run stability checks, see elog95. E/p value shows interesting drifts. Suggested plotting all runs in time order (use color to separate kinematic settings).
4. Xiaochao is in communication with Golak group on calculating finer grids around the ppn breakup threshold, see elog98.
5. We discussed upcoming UGM and Hall A/C meetings.

1. HUGS is ongoing for this and the next two weeks. We may need to adjust meeting time to see if it fits better students' schedule.
2. Murchhana reported that the reason for the delta-dependence of the acceptance calculation was due to the "3-sigma" "good recon" cut. The sigma value of xp(recon)-xp(ini) for example has strong z dependence. If fitting z-depenent sigma then this is gone. However, given that there are less than 0.1% of events with recon outside of the 3-sigma range, it was decided that repeating the same exercise for the recon quality of other variables is not necessary and we can extract the acceptance without the 3-sigma cuts.
3. Junhao reported summary of pressure curve analysis for all runs, see elog94. Nothing in too glorious agreement with the fill density expectation and results vary from cell to cell and from kinematics to kinematics. Will need Mingyu's RC xsec grid for radiative corrections.

1. Melanie presented livetime analysis. Suggestions: (1) check eta definition (it should be 100% for low rate runs), and prescale used; (2) plot livetime asymmetries in addition to livetime; (3) for asymmetry plots need to plot error bars (to see how much the cut reduces the statistics), and better binned in x (or W for 1-pass runs); (4) BCM reading shows slight different charge from Chao's online replay (used the same params), need to check.
2. Murchhana showed updated acceptance study, see elog90. The changes since last time are: now using recon not thrown variables, and added "good recon" cuts - for each variable, fit the residue (thrown-recon) with an Gaussian and apply a +/- 3sigma cut. The problem is the cut seems to have a high delta dependence. After the meeting we found out this is because the "sigma" value vary strongly with delta for two of the 4 variables (and weakly for the other 2).
3. Mingyu continued working on generating RC_external tables for the GE180 glass, see elog89. The discontinuity in RC_external output was traced (on slac after the meeting) to how RESMODD in F1F2IN09 is used for nuclear target and it cuts off at Q²=11 GeV². We asked Eric Christy but was told the F1F2IN21 isn't optimized for 3He yet. Since Mingyu already generated tables for all other material, we will continue using F1F2IN09 and will contact Eric later (end of summer) for an update.
4. Mike reported checking Golak table's smoothness and test out the intepolation algorithm for "in between" kinematics (using both his own and David F.'s code), see elog91. The interpolation looks very good. For "smoothness", it may be necessary to request for 0.1-deg steps (currently 0.25 deg) near the QE edge but we will wait another week before going back to Golak. Suggested: add the "below ppn" points which aren't so many. Then will proceed with last week's to-do list.
5. In-between meeting notes: Xiaochao posted from XEM elog a bump in xpfp reconstruction, see elog84. Though looks like if we apply an abs(delta).lt.8 cut then it should be automatically rejected.

1. Melanie presented livetime calculation updates, see elog86. Looks good (for A1n DIS very low rate) but will check the calculation with the helicity cuts and runs with higher rates. Cautioned that both EDTM pulser and beam ramp rate changed during the run.
2. Mike showed plots for elastic 3He with Golak's QE xsections implemented. Comparison with data looks good for runs where all scintillator bars were on. Will: (1) make plots of xsec vs. theta for fixed Eprime and see if finer theta binning is necessary (finer than 0.25-deg calculation will have to go back as a request to the Golak group); (2) proceed to runs with custom configurations of scintillator bars but will start with ref cell runs with known pressure, then to runs of elastic polarized cell. (3) Also need to code the asymmetries. There were some nice chats and followup code snipits about how to do "4-point interpolation" (colinear? coplanar? Binomial? what's that jargon again for the double linear intepolation?)
3. Junhao showed more pressure curve analysis for Brianna and Austin. Will (1) check N2 subraction in detail as DIS runs do not produce as good results as 1-pass runs. N2 correction plays a much bigger roles for DIS runs; (2) We noticed Dec'19 Brianna HMS runs show nice elastic peak. Might want to consider using those for physics analysis (later, for Mike).
4. Mingyu showed more RC external xsec grid outputs. One of the settings show plotting glitch that needs checked. Still running RC-external for the glass windows because we need 7 grids and then take the mass-average (that's for the 7 elements of the glass composition).

1. Melanie finished pion PID and contamination analysis for d2n, summarized all (both A1n and d2n) in a nice slide, see elog78 (the details are on slide 26 of this PDF or 2nd attachment of the elog entry). Previously we talked about dividing the data (where spectrometer momentum is very close to NGC threshold) below and above NGC threshold but given the pion contamination is so low, Brad said that's not necessary. Same is probably true for the NGC efficiency "dip" other experiments are seeing. Next will focus on livetime calculation with and w/o beam cuts. So far w/o cut results agree with report file.
2. Murchhana showed acceptance study for SHMS 18-deg and 30-deg and HMS 20-deg and 30-deg, see elog79. Look all reasonable. SHMS delta is asymmetric but that could be just how the magnets bend the particles. For A1n 30 deg settings may want to reproduce theta/phi plots for |vz|.lt.17 or 18cm just so to make sure the cuts are still fine for the smaller vz range. Mingyu or Melanie will use these plots to determine acceptance cuts in the asymmetry analysis.
3. Mingyu showed window thickness measurements with more points, see elog82. Because the previous 3-point measurement and the new one do not place on the same curve (the 2nd point is 10% thicker than target center) though they are consistent within error bars, the new average up-stream window thickness (now 1.014) is less than previous value (1.08). Suggest using the curve for now and use the variation for uncertainty study. Target length variation has been found negligible.

1. Replay is done but we need to check if any variables are missing.
2. Mingyu showed RC external calculations for 6 GeV d2n kinematics, see elog75. Look reasonable. Also showed target cell window thickness measurement using a cut-out from Todd. Measured only two points, will measure more before returning it to Todd as it can be useful for the upcoming GEn-2 running.
3. Melanie continued working on livetime corrections. A relevant elog is at elog74 but we did not discuss as it's not final.

1. Replay is almost done!!!
2. Melanie: posted skim script at elog73. Now everyone can develop own analysis script!
3. Melanie also pointed out the hodo ADCTDCtimediff on SHMS are different among channels. Mark said previously this was not important but with hcana's recent update it's safer to align them to zero. Fortunately it was already done prior to the latest replay. Phew!
4. Mingyu:
1. showed a few plots of RC_external's output cross sections. Elastic and QE tails are negligible. Questions remain why xsec_RC is lower than xsec_Born. We decided Mingyu should double-check the code with established data (hint hint: David Flay's thesis).
2. showed calculation of window thickness with raster on. The difference is at 1E-4 level when the beam is centered. Even for 1.5mm of off-centered beam the effect is tiny.
3. Regarding window non-uniformity, Chris does not have this info. Todd commented he has a window for GEn now and can do a multi-point measurement. (Similar questions were raised for GEn as well.)
5. Murchhaha
1. showed: for SHMS with rectangular (xptar,yptar) cuts, wide vz cut and plotted acceptance vs. delta. Results look promising because the 3 tightest cuts show nearly flat delta acceptance. Jianping suggested using forward-MC (with some model like F1F2) and extract the cross section using these acceptance grids. Any dependence on the (xptar,yptar) cut used can be treated as the relative uncertainty of the extracted acceptance as well as the method to extract xsections. Mark reminded that when doing so, we need to apply bin-centering corretions.
2. Still working on HMS and 30-deg settings (rootfiles disappeared over the weekend and have to be regenerated)
6. Mike: concluded the carbon excited states are quite important. We made some suggestions for how to improve the plot. One thing we noted is the MC elastic peak seems to be offset from data. Need to plot W(12C) or equivalently excitation energy (not delta, which has the angle in it) and see if the offset is consistent with the expected knowledge of Ebeam and Eprime (both are at 5E-4 level??).
7. Jixie reported on making the acceptance code working with multi-thread method.
8. Post meeting chat: we exchanged "2nd shot experiences"...

1. Murchhana reported on acceptance study (round 1, no cut), see elog70. Suggest:
1. For round 2, apply 2D theta vs. phi cut (probably multiple), delta between -10% and 10% (for HMS) and an equally wide range for SHMS, and make all acceptance 1D delta plots.
2. Generate round 1 plots for 30 deg MC files. Need to pay attention that the z cut should be very wide (perhaps +/-24cm is good).
3. Need to automize and run all these on the farm.
2. Melanie showed the Calo E/p plot for d2n settings. Observed a 3-4% shift from A1n. May need to rerun the calibration, see elog72.
3. Mingyu showed the first round of RC_external xsec calculation, see elog71. There were questions on whether xsec_RC is lower than xsec_Born, so will check/plot the elastic or QE tails. Also need to consider that with the raster and/or the beam offset from (0,0), the upstream glass window is thicker than the window thickness itself.

1. Melanie reported checking ECal calibration for d2n data. Some settings not consistent with A1n observation and need re-calibrated.
2. Melanie also reported PID analysis for d2n. The problem is that at least two of the d2n setting had the SHMS central momentum close to the Cherenkov threshold pion momentum, and it's best to divide the data into "above" and "below" threshold. The exact value of the threshold cut (supposedly using delta) should be determined from data. For -7.5 GeV/c runs, no need to apply NGC cut because the momentum is too much higher than the threshold the NGC does not contribute to PID at all.
3. Junhao reported on pressure study using 3He reference cells, see elog65. General suggestions are that in addition to look into pressure gauge accuracy and calibration, should look into the low-prescale (high deadtime) ref N2 run 11474 as it will be a main source of the systematic uncertainty of this study.
4. Friday 4/16 discussion: focused on dry runs for APS April meeting. Plus some discussions on the acceptance and RC_external usage.

1. Mike on elastic/QE: raised the possiblity that 12C excited states may be the reason for simulation missing data in the el/QE gap region. Will ask CSR for excited state information.
2. General discussions: kinematic binning for window dilution will be "naturally applied" when analyzing data.
3. Friday 4/9 discussion: Junhao went over asymmetry analysis roadmap; We discussed RC_external, and Jixie showed how to submit (old system) farm jobs as we will need to generate xsec tables.

1. Junhao presented pressure curve of 3He elastic using September SHMS data, see . Plenty of discussions and to-do's:
1. Overall looks like already reached 5%. Need to improve to reach 2% (this will go into xsec uncertainties)
2. can try cutting on W(14N) to reject N2 elastic, but there will still be N2 QE contamination. A better way may be to use N2 ref cell data and scale down using N2 fill pressure of the production cell and see how well it works.
3. Can't ignore the vacuum run and the 382Torr ("vacuum") run. These should locate directly on the line if the analysis is done correctly.
4. Currently the line does not cross zero (or the empty target point). It could be that there is window contamination even with the tight z cut. Can try to use the vacucum run to subtract this "vacuum window contamination".
5. z cut seems to be too much on the positive side where the acceptance is dropping. SHould shift z but to more symmetry and avoid the sharply-dropping region (use 3He pol cell run to judge).
6. need to add error bar to horizontal axis: about +/- 1 psi for pressure for each run.
7. If everything is done correctly, the "3He pressure" vs. cut should be a flat line rather than continously decreasing. -> this should be the goal.
8. Instead of using elastic peak, another method is to integrate the full elastic+QE spectra. This should work provided all pressure curve runs were taken under exact same conditions. Can plot yield from all runs and compare.
9. Someone needs to help checking the TC pressure equation -> will leave this to target meeting.
2. Murchhana showed plots on the simulated acceptance using 30-deg MC files. We discussed the following:
1. MC used 6mm-dia raster, will cut on 4.5mm when extracting the acceptance;
2. Do we expect 100% acceptance for some region? -- maybe not. These spectrometers are not like HRS where accp~100% for a wide range.
3. Currently we only have 30-deg MC files made by Jixie. Jixie will show Murchhana how to submit jobs and she will run high-stat MC for all d2n angles.
4. plan to add the acceptance event-by-event using the skim script -- nobody seems to object it so we will do this.
5. After the meeting, Mark and Jixie discussed whether we need to add xtar (or vertical position of raster) to the acceptance and make it 5D. The answer currently is no, because acceptance for a certain (xp,yp,ztar,delta) bin should be quite fixed and not vary much within that bin (for different raster position). Murchhana could run a setting of MC with the raster off and see how big the difference is in the acceptance output.
6. Discussed how to "verify" the acceptance using data: (JP) normalize data by sigma_Mott and the result should be F1,F2 which should be smooth functions of (x,Q2). But we will need a skim script for doing this so the script can be used on pass1 replayed data.
3. General discussions: To add radiated xsec to MC, we decided to use rc_external code, see https://github.com/JeffersonLab/rc-externals. Dave Gaskell commented (last Friday) that this is the state-of-art code for radiated xsections. David Flay looked at the code briefly and confirmed the code does what it's supposed to do and is a good one to use. We may need to run this code and generate huge tables like F2 did (since running this event-by-event will make the MC too slow).
4. Discussions delayed to next week: (1) Do we need window dilution analysis for d2n? (2) focal plane mismatch -> later; (3) PID progress for d2n?
5. Friday 4/2 discussions: Bill showed us how to run RC_external, see recorded video here. We will need to add pol3He target info first (Mingyu).

1. Murchhaha showed plots on the focal plane comparison of SHMS MC vs. data, see elog61. There is a corner missing for the downstream window in the xfp vs. yfp plot and is reflected in other variables too. This is not visible in A1n data (A1n doesn't see downstream window at all). This is more obvious at the SHMS -7.5GeV/c setting.
2. Mingyu showed preliminary window dilution analysis using simulated windows only, see elog58. Overall good results. Next step is to move onto using empty target data as the first-order correction and use MC to correct at the second order. To do so, MC will need to incorporate radiated cross sections.
3. We spent the rest of the meeting to discuss preparation for the collaboration meeting.

1. Our collaboration meeting is next Thursday. Please register here and the event page with agenda and Bluejeans connection information is at event/434. We discussed theory groups to contact for future talks/work.
2. Presentation on radiative corrections by David Flay, see elog55 for slides and bluejean recording. One possible project remains to do is to convert POLRAD to C++. Will check with Sylvester if a C++ package already exists. If not, this can be a possible task for students.
3. Mark looked into MC over the week and found a few things:
1. Related to the suppression of HMS MC in phi angle, see elog51. This is a one-word fix in mc-single-arm. The shms side was fixed long time ago but hms was only fixed last week.
2. Related to the HMS optics problem in Junhao's report two weeks ago, see elog52, basically zreact needs to be calculated twice. This needs modification in hcana (or replay) and thus followup from Sylvester+Brad. THis effect is larger at small angles. As for short target (4-10cm), depends on whether a cut is applied to z.
3. When reading through (our version of) the code, had the following comments: It is probably a minor point but it looks like you are still including the scattering chamber window when calculating the multiple scattering. You should probably remove the scattering chamber window and put more air (N2?) between the cell and the HMS or SHMS entrance. Also it looks like you are using the cyrotarget design for multiple scattering instead of the actual target with the glass side windows. ANyway these are minor points, but probably worth fixing. -> need followup from the sim crew.
4. (this was by email): suggest to not cut too tight on theta or phi. Suggest abs(theta).le.0.08 and abs(phi).le.0.06 for SHMS, as an example. Wider cuts are needed for extended targets.
5. Emphasized our knowledge of P0 is at 0.2% level for HMS and SHMS. Knowledge on SHMS acceptance is still not well understood especially for our long target.
6. In addition to the two updates from last week, talked about the importance of correcting "DC offsets", see elog60. This correction seems to have improved Murchhana's d2n simulation/data comparison greatly for the windows. Also suggested looking into focal plane checks (x,y,xp,yp). Our data are hitting the DC edge. The survey of DC chamber edge should be checked with data.
4. Murchhana presented the d2n simulation update. The 3 fixes improved a lot the agreement between MC and data on the windows, see elog57. Next step is to compare focal plane distribution between MC and data, develop acceptance cuts, and to generate acceptance files using Jixie's high-stat simulation outputs.
5. Mingyu is working in the target lab today so missed the meeting. He will report on the window dilution study next week.
6. Junhao presented progress on N2 pressure curve, see elog54. Details below:
1. N2 elastic peak is very small for HMS 1-pass runs (again, SHMS runs without the target collimator are useless and we did not take N2 pressure curve runs during d2n 1-pass running when the collimator was in). Will try to get the uncertainty in the N2 elastic peak and then wrap up this topic.
2. Will move on to N2 ref. runs for DIS. For 3He pressure runs, will check both DIS and 1-pass d2n runs on the SHMS with the target collimator.
7. Other updates:
1. Survey results for book-keeping see elog53
2. Todd sent us the Reference cell window info, see elog59
3. Bill pointed us to Aruni's talk on focal plane study: f2_optics.pdf

1. Today's meeting will be short because of the SoLID DOE Science Review.
2. Brad: replay is still ongoing, mostly checking all updates and working with Sylvester on including the helcity scalers.
3. Melanie: also still checking on the skim script.
4. Mike on elastic/QE:
1. used DC residue to add smearing: hdc_sigma increased from 0.03 to 0.06 (in cm) to match z width in data for the central foil. Junhao mentioned he did not update the residue params after DC calibration -> Mike will work with Junhao on this.
2. need to check what gas is in NGC in mc_single_arm, as the 600um seems large.
3. So far did not use radiated cross sections. - will need to radiate the cross section - consult David Flay?
5. Jixie showed last friday how to use XSTree code to extract acceptance from simulated rootfiles. We need very high statistics. jixie will hand over a script to Brad (or vice versa) for running the simulation on farm. But need to wait until some mc-single-arm fixes are done, see Mark's item below.
6. Murchhana showed the first simulation for d2n runs, see elog. Discussions below:
1. Data windows are much lower than simulation. Not sure why. Mark mentioned we are still understanding SHMS (but the windows for d2n are at smaller y so should be better understood than A1n??)
2. We need the window thickness data for Tommy. Mingyu will write to Chris.
3. Please check if the cuts on theta and phi are switched (there appear to be a cut on theta at 0.05 on slide 20, while the texts above the plot shows 0.06). Then, make sure the cuts do not cut into data or MC so we have a fair comparison.
7. Mingyu found a mistake in calculating the number density for GE180. After correcting this the agreement of data vs. MC is much better for the windows for A1n, see 2nd PDF on elog49. Otherwise still working on the window dilution analysis.
8. Junhao worked on compiling all special runs, otherwise still working on the pressure curve analysis.

1. Brad: replay is done for about 1/3 of runs, but encounters some farm problem. Still sorting it out. Suggested posting location of completed jobs on wiki or elog.
2. Melanie presented a first draft of the skim script. see elog47. Details:
1. Beam trip cuts can be customized using 3 variables (examples): min current (4uA) and time before (10s) and time after (5s - 10s in elog). For the given example values, the events between 10s before the current drops below 4uA, and 5s after the current rises above 4uA, will be cut out and assigned a "false" flag.
2. The helicity-gated scaler group (Carlos' work) is not implemented yet (same is true for pass1 replay). Currently using "standard scalers". Brad asked if standard scalers are gated and if so, should be able to cut out the beam trip too. Melanie will look into this.
3. Junhao presented first round of work for the N2 pressure curve, see elog46. Details:
1. Surprisingly, the N2 elastic peak is clearly visible in the HMS run (no target collimator) from Dec 2019. Even got the first N2 pressure curve! Some suggestions below
2. check optics: z_react should has no dependence on any of the other variables;
3. after checking optics, see if the dp vs. th correlation remain. Note that th is the "out of plane" or "vertical" angle should have no physical correlation with dp for elastic or QE.
4. tighten z and see if can get rid of that N2 elastic peak for the empty ref cell run;
5. if can get rid of the N2 elastic peak from empty cell data, choose that z cut and repeat for all N2 runs; redo the curve;
6. before we give up on getting a "precision" result from HMS runs, suggest give it a try to separate the N2 elastic peak from 3He peak for the 3He run. We might get a large uncertainty due to overlapping 3He and N2 elastic peaks, as expected.
7. After all above is done, repeat the analysis for September d2n 1-pass N2 pressure curve runs and a 3He run from the same period. The z cut may not be necessary because of the collimators, but needs check.
4. Brad reminded us the special run list should be posted at https://hallcweb.jlab.org/wiki/index.php/Pol_He-3_Analysis_Resources#Special_Runs. Xiaochao posted the A1n list there, only thing missing might be the positron runs, can't recall if we did the positron run for d2n or A1n.
5. Mingyu presented latest simulation for A1n, see elog49. Comments and observations:
   1. The non-physical events within the empty vacuum ref cell run were rejected by PID cuts, not ibcm1 cut, so likely due to the good track requirement in etracknorm;
   2. changing beam position in MC does not cause a visible effect. The empty target HMS phi MC is still narrower than data. Looking at this carefully, could be the acceptance effect: the windows might get cut in theta or phi as they are far away from z=0. Perhaps applying acceptance cuts will help.
   3. Updated MC to BigBrother window thickness: Upstream is 138um which is thicker than the previous 125um. Also moved the cell by 1cm upstream (survey of TOmmy shows it was 1cm downstream, so 1cm off is not atypical, still waiting for BB survey info). Now the window height of data still above MC (both SHMS upstread and HMS both windows). Could be nonuniform glass thickness? Wall scraping makes the beam scraping the (thicker) edge of the window -- not sure.
   4. Discussed the next step on A1n MC: Since we have not decided if cross section analysis is necessary, should wrap up MC work with the goal of extracting and finalizing the glass window cut and the window dilution factor.
   5. Another possible output of MC is the acceptance. But for asymmetry extraction, some "eyeballing" might be good enough, i.e. just choose region where acceptance is reasonable (not dropping sharply).
6. Mike presented an update on 12C foil analysis: added QFS and found it working better than F1F2QE; data is wider than MC; and data elastic peak is lower and wider than MC. Will: (1) add smearing; (2) check if all radiation effects are counted for in MC for elastic; then (3) move on to 3He.
7. SoLID Science Review is next Mon-Tue. We will have a short meeting on Tuesday 3/9.
8. We had planned to hold the first collaboration meeting on 3/17 or 3/18. Show time, everyone!

1. Mingyu presented latest simulation for A1n. (Note that slides shown last week had a scaling factor applied to match data to simulation at z=0, and we spent the week to figure out the agreement after removing scaling, including realizing the sim spectra were missing the cut istop==0). See elog45. They still match (without scaling)!! Comments and observations:
   1. The abs(z_react).lt.30 cut removed the hour-glass shape in the raster replay plot (these are non-physical events);
   2. checked and EDTM events are not plotted (as desired). Also adding ibcm1.gt.1uA cut to clean up data following Bill's suggestion last week.
   3. The BigBrother data windows still look quite upstream by a few mm (both HMS and SHMS data), though the deficit seen last week in the SHMS z=(10,20)cm region is gone. No likely no problem in optics and sim matches data well there now!
   4. HMS data shows downstream window but simulation do not. -- not sure how to explain this but we can just cut it out?
   5. For empty target, last week we saw "grassy backgroun" for abs(z).lt.20cm which can't exist for vaccum. After applying abs(z).lt.30cm, ibcm1.gt.1uA, and PID cuts, now these events are gone. It's not clear which cut cleaned up these events though -- likely ibcm1, but PID cuts also possible if it requires a clean track (as expected for cutting on etottracknorm).
   6. For empty target data, the windows match well simulation (indicating ref cell is mounted at exactly where it should be), though the winow height is a bit off. Also the sim phi distribution is narrower than data. We thought maybe we didn't illuminate the simulation enough but this difference remain after Mingyu widened the initial phi range this week. Thus we need to look further whether it is coming from, for example, beam positions.
   7. At this precision, Mingyu will do: (1) study which cut removed the vaccum events, PID or ibcm1? (2) implement beam position to see if it improves phi agreement for empty target; (3) implement exact window thickness for BigBrother and ref cell -- though keep in mind if we see scraping then the beam is not hitting the center or thinnest part of the window, so we have some uncertainty here; (4) Could consider now mispointing and shifting the cell position, but we are not at this stage yet.
2. With simulation at this precision, we started to argue whether it's worthwhile to also extract large-x cross section data for A1n -- stay tuned!
3. Junhao will discuss with Xiaochao on how to start the pressure curve analysis after the meeting. -- done, we will see how N2 peak looks next week!
4. Melanie emailed an update that she has requested merging of the helicity-gated scalers.
5. Murchhaha emailed an update that she is returning to the US this Friday.
6. Brad emailed an update that he is going to miss the meeting and will be back next week.

1. Brad: still working on the kinematic file for replay launch (HMS is completed but SHMS still needs work, some varialbes such as Ebeam and target mass are totally incorrect). Will work with students to update this using run lists.
2. Melanie: working on the "skim" script. Beam trips are done and the cut is reliable, still need to add helicities.
3. Mike on SHMS simulation: added cherenkov flag
1. the carbon foil ztar is still wider than simulation.(We thought the foil is thick at the meeting but then I recall this is small angle scattering so it's possible the foil looks thick in z);
2. Data shows events beyond z=20 and rising, could be secondary events from the collimators get reconstructed at random places. This is very hard to simulate (so maybe find a way to get rid of the effect);
3. fp-y distribution: MC seems to have a "flat top", should check what fp-y=0 events show up in other variables and maybe they get cut off some where?
4. More suggestions: for elastic spectra, plot W(12C) to check if the elastic peak is at the right place. Also always plot the two foils' simulation in two colors; Do the same comparison study for HMS (to have more tools to help with spotting problems); Then, if we don't see any major problem, next step is to figure out extra smearing (hard to do - DC smearing? optics recon smearing? unknown smearing??).
4. Mingyu on DIS simulation comparison with data, see elog 42. Some comments and suggestions:
1. For replayed data need to reject non-physics events (such as EDTM) and but a cut on current above zero (Bill mentioned the replay may put zero-current events at some fixed/or random places). This should (we hope) explain the raster y vx. x "fan shape" events, and the events in the middle of the empty cell run (vacuum? dark matter?).
2. Note: all pol3he data are from cell BigBrother, some tiny bit of scraping is seen.
3. Initial theta and phi ranges are a bit too narrow (simulation is narrower than data for SHMS theta, HMS theta, and a bit on SHMS phi).
4. Need to add electron cut. Something like (E/p.gt.0.8).and.(E/p.lt.2.0).and.(Npe of Cherenkov.gt.1) would be good.
5. For raster x and y, as well as focal plane x,y,xp,yp, plot 1D histogram in addition to 2D.
6. Target windows are not at the same ztar as simulation, could be a combination of target mount (couple of mm?), spectrometer mispointing (not yet corrected in replay).
7. Suspect the acceptance is not well understood for positive phi side (z=+10 to +20cm at 30 deg), if this remains we will ask Mark for advice.
8. After the meeting we realized some normalization problems that need corrected.
9. Some fine-tuning for the long term: raster shape, window position fine-tuning, resolution smearing...
10. Ugh! This is a set-back: a scaling factor was applied in all slides to match data at z=0 to MC. Need to remove this factor and attempt absolution comparisons.
5. Junhao showed some slides to discuss pressure curve analysis. May be hard to do because of the poorer resolution of Hall C spectrometers. Simulation to correct for radiative effects can be for later.

1. Brad: worked on the kinematic file for replay launch;
2. Xiaochao: Received setting 3 calculation from the Golak group. This concludes the calculations we expect from this group (their model is non relativistic and won't work for setting 4 the largest angle). Also received an initialization email from the second group that they are dusting off their code now.
3. Mike: added the target collimator and is using all available data to "fine tune" the collimator position.
4. Bill: Recall the SHMS resolution problem shown in Hall C meeting (Abishek's talk)? Turns out that was solved by turning the flag of the 2nd GC "on" in the simulation, see MC_SHMS_Cher_flag_slide.pdf.
5. Mingyu: made some progress on simulation. May need to shift the window position of the pol3He cell a bit along z, and add N2 simulation for the gas outside the cell.
6. separate meeting on 2/11 am: Radcor - maybe ask David Flay for the tool, for both asymmetry and xsections; d2n simulation: okay for Murchhana to learn how to calculate yield since it helps to gauge if the simulation is correct (even if we eventually use the acceptance method to extract xsec for d2n).
7. separate meeting on 2/11 pm: We will send pseudo data to Wally, and have invited him for our March collaboration meeting. Need to know theoretical uncertainty and its dependence on "how long hte theory extraction will take" (varies from 5 minutes of a crude extraction using Pn,Pp model to "never" for the ideal extraction -- "ideal" meaning "exact exact exact" in the eye of theorists).

1. Drift chamber calibration: Junhao completed the 2nd round of DC calibration following the hodoscope calib completed last week, see elog41. Some channels has wider residue than others but the consensus is that's normal (Mark, Brad) . Checkings for some settings are still ongoing but the params are ready to go.
2. Replay (Brad): will do a "sanity" check of DC params and then launch the full replay.
3. Data analysis script: Sylvester will work with Melanie on this. The output of the code should be N+ and N-'s binned in (x,Q2) -- similar to online, with PID cuts for electrons and ytar (or ztar) cuts for windows; beam trip cuts; "decent" cuts on the solid angle and delta should also be applied (can have two sets: loser acceptance cuts for A1n, tigher ones for d2n, etc); and the W greater than 2GeV cut. All cuts can be exported as a flag (i.e. 0=not passing, 1=passing). Once we understand better the beam scraping, some beam position cuts should also be applied.
4. Mike's elastic analysis: MC/data normalization looks pretty good now. There are a few comments and suggestions:
1. XSTree code used Bosted's fit. From Jianping: For QE, Bosted's works well for carbon and other heavy targets, but not so well for 3He. Eric is still working on improving the 3He fit.
2. theta (OOP) angle looks symmetric in MC but not in data. Suggest cutting tightly on delta: (+/-8%) for HMS, (-10%,22%) for SHMS.
3. check MC error bars, they look underestimated;
4. for HMS, add N2 gas simulation for the background;
5. for SHMS, (1) remove the "collimator" flag in mc_single_arm input (that's the spectrometer collimator, not target collimator); (2) Add target collimator to mc_single_arm (Jianping will provide a correspondant); (3) Add smearing to MC to mimic data (this smearing is due to tracking algorithm and data indicates it is not a Gaussian, but can start with Gaussian for simplicity); -- Shujie can help with both (2) and (3); (4) add 3He elastic; (5) modify XSTree to add spin asymmetries.
5. Scratch notes of useful points from various Hall C User's meeting talks last week:
1. F2 talk (Bill): e+ background, Q2 had 1.8% tune to match MC with data on the focal plane (note this was for early SHMS experiments before the magnet tunes were fully understood)
2. NPD talk: used ESR (known as VM2000 reflector) as reflective material.
3. EMC talk (Abishek): (1) 12C data ztar width MC is wider than data (mostly due to optics tune, but Dave Mack argues some of the non-Gaussian spread is due to tracking algorithm and can be improved); (2) Some normalization problem that can be checked by re-measuring 12C foil thickness (now, since these foils don't explode).

1. Hodoscope: Finally found the bug in HMS Vp code: the Vp fitting histogram range is too small and the data are out of range. This caused the low stat problem (see last two months' minutes) and the failed fitting repeatedly. After extending the upper range from 50 to 100 (thanks to Mark for finding it out), now Vp calib code works perfectly fine. Mingyu redid the calibration. Also checked SHMS to make sure the histogram range is fine. Final hodoscope parameters are posted on elog38. Suggestion to Carlos: add a 2D plot and ask the user to check the histogram range prior to fitting.
2. Calorimeter: Melanie emailed this morning that the calibration is final. She is now updating the PID results using the updated Calo calib in preparation for the Hall C talk this Thursday.
3. Since hodo calib is now final, we asked Junhao to rerun the DC calibration. Three settings: HMS, SHMS A1/d2n, and SHMS d2n*. If 3-4 hours per setting the DC calib may be completed within 24 hours.
4. Replay goal: launch the replay jobs to farm before the Hall C meeting -- Brad.
5. Monitoring scripts: may need fine tuning but probably more efficient if we test the code with the replay; Physics script: Melanie will look into this after PID work and the Hall C meeting; helicity code: Carlos will ask Steve to accept the git version, maybe during the Hall C meeting.
6. Mike's carbon QE work: MC/data seems to off by factor 10, will continue working on it;
1. Shujie offer that Hall A 3H/3He experiment had new results/fits on 3He QE -- Jianping commented it's highly Q2 dependent so the Q2 min of 0.6 of the 3H experiment may not be good enough for us;
2. The model used is F(y)-scaling (or y-scaling?), again Jianping commented that it may work okay for heavy targets (like 12C) but not for 3He where few-body is important, also the Hall A work won't provide us the asymmetry.
3. Nevertheless, since we start with carbon QE comparison as a benchmark, suggest Mike contact Shujie to get 12C QE calcuations. For kinematic grid see QE elog25 (bottom of file Golak_comm).
7. Xiaochao received setting 2 calcuations from Golak group. See same entry as before: elog25.
8. Discussed upcoming Hall C meeting talks:
1. Went through Junhao's d2n slides;
2. Mingyu's talk slides aren't ready (put all time into hodo calibration this week), suggest using some cool pictures from this this google drive directory;
3. Melanie's slides will be DNP slides with updated PID results so probably okay;
4. Bill's Moller talk was informative -- take away message is we can do day-by-day corrections to Moller results based on Q.E. and beam energy (electron spin precession varies by 1.75 deg for every 1MeV change in energy, and depends also on where the energy shift comes from - NL, SL, or both;
5. Mike: no slide yet(?)
6. Suggest all presenters to cross-reference each other's talk and acknowledge all RCs on the slide (since this is the first a1n/d2n talks at the Hall C meeting).
9. Other misc items: Xiaochao talked through how to calculate yield from mc_single_arm with Mingyu and Murchhana.

1. Hodoscope: still not done, Vp calibration not working due to low stat. Could tune the beta to 1.0 by adjusting the LCoeff in Vp param but we should avoid such tweaking method.
2. Calorimeter: Melanie is doing final check. Goal is to find outlier blocks (param in principle should all be constant for all blocks after online gain match) and tweak by hand.
3. Some analyzer/replay readiness discussion
4. Mike showed some preliminary comparison of carbon foil data. Suggestion: (1) Run the QE option of mc_single_arm + xsections (not the elastic mode because the data is mostly QE at this kinematics); (2) In addition to comparing target variables, also compare focal plane ones.

1. Focus on replay readiness (notes by Brad)
2. Hall C analyzer (hcana) repo status -- Sylvester
1. hcana code readiness for pass1 replay
• Sylvester is working on integrating upstream hcana changes
• moved hel_scalers_*.json files to REPORT_OUTPUT/
2. discussed how an upstream change in how FADC hit selection is handled will impact us
• decided to merge that patch (best to stay close to upstream) and make needed updates to our config files
  • we will need to establish 'default pedestals' (run new script)
  • change introduces a timing shift in the adc_tdc-tdc hit selection formula
    • need to add offset(s) in tdc hit times for the hodoscope
  • Sylvester + Mark will work together to work out optimal changes
3. Sylvester will run some tests and push out a new singularity container
3. Parameter file status:
   • Hodoscope calibrations -- Mingyu
     • We now understand the shift in SHMS beta noted earlier
       • due to shift in s2y plane that correlates to replacement of a failed scaler module (https://logbooks.jlab.org/entry/3846102) that shares the same ribbon cables with the associated TDCs between runs 11372 and 11376. Best guess is that the termination in the replacement scaler wasn't matched and this resulted in signal distortion and the observed ~2ns s2y timing shift
       • Brad will look at signals and follow up for future experiments
       • Solution for us is to go with Mingyu's updated calibration for runs
     • Issue with the shoulder on HMS beta in -7.5 GeV/c kinematic remains
       • Recalibration using a different run reduces the shoulder somewhat, but HMS beta peak is still non-Gaussian. Indicates an underlying problem with the HMS calibration.
       • The beta peak with the 'old' hodoscope calibration is narrower and more symmetric, but is only slightly offset from 1. Probably a better calibration than the new one, but this is a solvable issue.
       • Mingyu will plot beta vs. paddle number with the new calibration, and take a look at how the calibration constants changed between the 'old' config and the new one. We'll see a report on this next week.
   • Calorimeter calibrations -- Melanie/Sylvester
     • Melanie reports that the single 'defocussed' calorimeter calibration is looking optimal for all runs (as hoped). She is doing a few final checks, but it is looking good
     • Pion cross calibration results look consistent with the defocussed calibration results as well. Also a good sign.
   • NOTE: We'll need to double check the PARAM settings with the updated hcana and ensure all is well. (Also verify/address the expected timing shift noted earlier.)
4. Confirm what replay script will be used for 'pass1' replay
   • SCRIPTS/hcreplay “-m all” ... is probably a bit too verbose on the output
   • Brad and Sylvester will pick/make a more balanced output.def list
     • Mark suggested we pack as many “QA” check/monitoring histograms into the replay output as possible.
     • Brad, Sylvester will followup with Mike N. to see what can be added
5. Per-run Monitoring script readiness/update -- Mike N.
   • Couldn't make today's meeting, will follow-up with Brad, Sylvester offline.
6. Simulation comparison with data, adding raster, formula checks? -- Murchhana, Mingyu
   • found an angle sign inconsistency in the cross section code: it used Hall A's code to transform from HCS to Spectrometer CS, which treats Left/Right HRS as positive/negative angles, but when calling this code, Jixie's code gave positive angles for both SHMS and HMS (basically it used the wrong angle for the HMS). This can be easily corrected by always input negative angle setting for the HMS when calling Jixie's x-section code.
   • There was a follow up discussion on how 'our' code differed from the Hall C maintained upstream code. It sounds like Murchhana and Mingyu are working off a fork of the upstream repo that Jixie had modified but were not sure what or how significant the changes were.
     • Murchhana and Mingyu will push documented git repos up to Github or ANL's GitLab (ie. with clear commit descriptions AND a clear ReadMe describing how to run the code and generate the output we need.
     • Murchanna and Mingyu will email the repo links to the analysis mailing list
     • Mike N and Sylvester will see if we can get the important changes merged upstream
7. Elastic/QE simulation update? -- Mike N.
   • No report/update this week.
8. Miscellaneous items:
   • New a1n_d2n_analysis@jlab.org analysis-focussed mailing list created: Subscribe if you aren't already on the list.
   • Junhao emailed that he is working on AFP loss analysis and will report next week.
   • Students sent the requested param files vs run-list to Brad. In most cases we have a common run-set that will work with all runs
     • Hodoscope calibrations still have some issues (plans to address noted above)
     • Need to confirm that Cerenkov cuts are consistent/optimal with Melanie's work
     • Once the hodoscope calibration issues are understood AND the upstream hcana changes are merged we will need to do one (hopefully) last sanity check and we are ready to go for the pass1 replay
   • Michael also found an error in the HMS standard kinematics file (typo/error in run range).
     • https://eicweb.phy.anl.gov/jlab/hallc/exp/polhe3/hallc_replay/-/issues/2

1. We first focused on the full-replay readiness:
2. Hodoscope calibration status:
1. Mingyu modified the calib code so if there are not enough events (less than 1000 for SHMS or 200 for HMS) for a good fit of the drift speed, then set to 15.0 cm/ns for that paddle/channel
2. A "shoulder" in beta remains for HMS, could plot beta vs. focal-plane x to check if that's related to a specific paddel - to do for Mingyu. Bill mentioned there is also a "good sc paddle" branch that can be used to choose hits at middle of the paddle only.
3. The summer 7.5-GeV setting problem (shift in beta) remain. Comparison of calib params shows c1 varied quite a bit from A1n to the d2n* (7.5GeV summer) calibration but c1 is not even used in hcana (only c2 is). Cearly we need a focus group on this problem now -- Mingyu, Carlos, Xiaochao, Mark
3. Calorimeter calibration status - Brad, Sylvester: some blocks stil have zero or even negative params that are not reasonable. To recap:
   1. Melanie has done several iteration of calibration, first using the defocused run and then tried "merging" with params from DIS runs. The change in dE/E is negligible, see 07/28/2020; Brad argued that the merging method may not be reliable and so we should use the defocused run results for all runs.
   2. But the defocused run params may still have the "too strong correlation" problem. Sylvester/Brad will check the params and see if any block is signfiicantly different from others (in principle, all blocks are gain-matched which means all params should be the same, see minutes from 2020/11/10.)
   3. Sylvester will check with Melanie on the status of the pion peak check. This will be useful as supplemental info for the param check above.
4. Brad asked all students to send him the params and the run number ranges for each set of param.
5. Murchhana showed formula checking between mc_single_arm, hcana, and her own derivation using the optics notes (or HCS, TCS definitions), see elog33. So far, mc_single_arm and hcana, and the optics notes are all consistent though Murchhana's own derivation is off by a minus sign. In a followup discussion after the meeting, we found if we use Jixie's code to calculate the x-section and the rates, the input angle for HMS side should be negative (same as for hcana and the optics note). We will re-run HMS simulation and update next week. (60 years after the era of Enrico Fermi and the separation of "physicists" into experimentalists and theorists, we can say this is where both groups still share the same struggle: spend hours or days to chase a minus sign, or multiple minus sign errors here and there.)
6. Mike N. provided a short update on elastic/QE simulations. Continued working on the energy loss. Also tried integrating the QE results from elog25 and tried a spline fit for continuous coverage in E'. In fact Xiaochao has just updated this entry to include a finer-bin version from Golak group dated 1/1/2021, for 9.25-deg and 4-MeV E' bins. The E' values are different from the previous "0.5-deg theta and 10-MeV E' bins", so technically one could combine both coarse-bin and the new results for a better input to the spline.

1. Last meeting of 2020! Does anyone recall when we had our last meeting of 2019? We had it on 12/30 (I think) in the counting house!
2. APS/Hall C talk plan: Hall C: A1n-Melanie, d2n-Junhao, target-Mingyu, elastic-Mike; APS: A1n-Melanie, d2n-Murchhana, target-Junhao, elastic-Mike. We will review abstracts on Jan. 5th.
3. Carlos' notes on helicity scaler decoding can be found at doc1105, including a sample script.
4. Replay script (hcswif): see minutes from 12/18 (last Friday). Bill found an error in the script and will fix it after consulting Steve. Both Melanie and Junhao tried the script and managed to set the parameter using run number ranges, see Junhao's brief instructions under folder 20201222/
5. Replay plan -- Brad will check all calibration/params and launch the first round over X'mas. Will focus on cooking partial set of data per each setting and check output.
6. Replay scripts: for cutting beam trips, it appears there are now 3 ways:
   1. First is Sanghwa's script that is already in the Hall C replay calib (Bill uses it for F2 currently), see instructions;
   2. Second is to use Carlos' new helicity scaler code. The beam trip cut can be done within the same pass (with a 1/120-sec delay, Carlos already have this cut in the sample script);
   3. Melanie was also working on something with Sylvester, but only half-finished.
   4. Melanie will look into all 3 and come up with a plan.
7. Replay script regarding physics extraction -- later.
8. Hodoscope mystery: Brad checked the hodoscope calibration and some paddles on the edges have incorrect drift speed results due to poor fitting (or low stat). Mark suggested Mingyu to determine certain criteria and to set those paddles ot 15.0 if fitting fails. Then plot beta again to see if it fixes the 7.5-GeV run problem. - We may meet tomorrow or next week to follow up on this.
9. Melanie on pion peaks in calorimeter blocks: now added the "single block" cut. Most channels are fine but some are not. For SHMS shower blocks, Mark suggested to require "MIP peak in corresponding preshower block in front of that shower block" which may help to cleanup the spectra. For HMS, suggest requring "MIP peaks in all 4 blocks that line up" to clean up. Will try to fit the peak, fine tune the param, and plot dE/E to see if it improves or stays the same. (We will only need to revisit this if the tuned params give worse dE/E).
10. Melanie on pion contamination: Looking okay/good enoughf or now. Will document so will not forget the details when writing thesis.
11. Simulation formula checking: Mingyu is still working on it (comparing optics notes, hallc_replay, and mc_single_arm). Bill added that we should also check Jixie's code where the physics xsections are calculated.

1. Bluejean recording link: part 1 is on farm replay, part 2 on how to define different parameters for specific run number range - unfinished, part 3 on simulation;
2. Bill's slides, and a couple of other stuff can be found under 20201218/.
3. This should supercede the previous recording on 11/20.
4. Example input files on A1n setting: shms_B.inp, shms_C.inp, test3.inp, test4.inp
5. To do on farm replay: 1) learn how to use hcswif to submit jobs; 2) find out how to set params for specific run number ranges (on ifarm); 3) submit list of short (50k) replay jobs to farm, 1-2 runs per kinematic setting and verify outputs, verify params used.
6. To do on simulation: 1) continue formula check; 2) learn how to write an input file; 3) run mc_single_arm then getXsec and plot rate vs. ztar.
7. Followup on setting runnumber ranges: looks like we were on the right track on this screen. To change cut params and calib params, replace pointers g_ctp_parm_filename and g_ctp_calib_filename by the appropirate files. Also see brief instructions under 20201218/.

1. Sylvester sent an email that the latest HMS optics have been added to hallc replay. Those doing replay should download the latest version.
2. Carlos: working on the BCM asymmetry error calculation in the scaler code. Brad will check Carlos' notes on the helicity scaler. We would also like to see a detailed description of where the uncertainty of BCM comes from and how it's propagated.
3. Melanie on pion peaks in calorimeter (see elog31): wrote code based on Vardan's but we are not sure if the peaks we see are real physics (pion) peaks, or a cut-out of the low-energy background or even pedestals. Need to add the parts from Vardan's code that select events that have no energy deposit in all other blocks, and remake all plots. This may take some time, and we decided to "do as much as we can this week, but can proceed to the first round of replay without it."
4. Melanie on pion contamination (see elog30): Added "method3".
1. Now the values for HMS are 0.1% (method 1), 0.44% (method 2), and 0.03% (method 3 by scaling down the pion spectrum). Method 3 assumes the pion sample in the final analysis has the same shape as the "clean pion" sample and this adds to uncertainty. Note: previously we thought the 0.1% is an underestimation because the pions in our analysis sample may be subject to a different pion rejection from the "clean" pions used in the rejection analysis. But now looking at these numbers (that 0.03 is way below 0.1), it is equally possible that the "real" pion rejection is higher than the value we are using (due to "clean" pion samples "not really clean").
2. Overall we think the 0.1% from method 1 is good to a factor of 2, i.e. the real value may be between 0.05% and 0.2% which is good enough.
3. For SHMS though, the 3 numbers are 0.2%, 0.4%, and 0.03%. The 0.2% is on the high side and we asked Melanie to refit the pion contribution (by tuning the functional form), since the red fit has a clear overestimate of the actual data near E/p=0.8 and also above 1.2, for both HMS and SHMS.
5. Mingyu on hodoscope:
1. Bill pointed out hclog 3818947 in which Simona noted that "starting with runs 10815 in SHMS and 3633 in HMS", "the timing changed so you will have to revisit your timing cuts". This solves the mystery why all summer runs have different diffTime for hodoscope paddle for both HMS/SHMS, and now we have the exact run numbers!
2. The beta shift for -7.5GeV momentum runs is still now mystery: the -7.5GeV d2n runs from March do not show the shift, only -7.5GeV runs from the summer do. This means the shift is not caused by physics (and not by the difftime shift since other momentum settings do not show it). While we do not use the beta spectrum for analysis, the shift is worrisome as it may indicate deeper problems in the replay. Mingyu will send a couple of rootfiles to Brad for checking.
6. We discussed replay script readiness. Bill will work with Carlos on getting the scripts ready and hand them over to students by Friday.
7. Murchahana and Mingyu showed some mc_single simulation comparison with data. For A1n 30 deg, the SHMS downstream 3He simulation clearly shows a faster drop than data, see mc_shms_30deg_screenshot.png. For HMS the ztar seems reversed. To do: (1) add circular raster; (2) check ztar calculation. Here is a breakdown:
   1. check ztar calculation in hallc_replay in e.g. analyze.C, pay attention to whether the spectrometer angle is defined with a sign.
   2. Derive how ztar should be calculated following the definitions of various coordinate systems. A good reference is Carlo's thesis, for example on page1.png. You should derive equations that look like page3.png (note: I am not saying these Eqs.() are incorrect, but we do need to check everything with the sign convention of our choice).
   3. Similar to Carlo's thesis that shows how SHMS coordinates are defined, this HMS optics note should tell you how HMS coordinate systems are defined. Also see Holly's slides on HMS optics
   4. Another reference is the Hall A technote 02-012 on HRS optics.
   5. check ztar calculation in the latest mc_single code e.g. mc_single_arm.f, note the mc_single input defines spectrometer angle to be positive for both SHMS and HMS.
   6. Put everything together and make sure we have a consistent (i.e. correct) picture.
8. Mike N. showed slides on mc_single, see elog32, with a focus on the formula used for multiple scattering and ionization loss. Jianping noted need to work on Bremsstrahlung too.

1. Holly: merged SHMS optics to hallc replay as "default". (Sylvester confirmed the merge by email). Note: this does not affect simulation since mc_single is based on COSY for both forward and backward transports. Still not clear of HMS optics have been merged (Mark wasn't at the meeting). With this progress in optics (SHMS only) we can now start doing a couple of things:
1. Mike N. can start comparing simulation with data for elastic scattering, with a focus on focal plane comparison
2. Murchhana and Mingyu, both completed simulation of 3He gas and glass windows separately, can now replay some data runs and try a first comparison of simulation vs. data on the pol3He cell production data. (Murchhana will learn how to use the DIS cross section code first).
2. Carlos: finished helicity decoding, Steve is checking the code before merging. His section of the presentation was recorded in bluejean, see recording link (must use jlab login to access). Highlights:
1. Outputs of his code is the same as report files (Steve's version)
2. We discussed how to cut on physics events.
3. Working on a document to describe what the code does.
4. Bill mentioned there is an existing script to add varaibles (such as BCM or current) to replay output event tree "T", see bcm_check_module.pdf. Can adopt this for our helicity and current cut variables.
3. Melanie: Still trying to make Vardan's script work for plotting pions in the calo blocks -- will move this to higher priority for the coming week. Meanwhile, making good progress on pion contamination. We reviewed the slides (see elog29 and made a few suggestions:
1. For method 1 (applying PS cuts to Etot/P spectrum, fitting pion and electrons separately, divide pion fit/electron fit by NGC rejection factor): when dividing NGC rejection factor, should use the low-momentum-setting rejection factor results for both low- and high-momentum settings. This is because NGC performance should be independent of the kinematics. Low-P setting has more pions so probably cleaner pion samples (and more accurate results on the rejection factor).
2. The value given by method 1 is probably an underestimation. This is because the actual pion event causing the contamination is different from the pion clean sample used in NGC PID analysis.
3. For method 2 (applying PS cuts and NGC electron cut to Etot/P spectrum, try fitting pion and electron separately, divide pion fit/electron fit which gives directly pion contamination): the pion fit looks too high (either constant or rising at high Etot/p). If possible, use the same functional form as method #1 and scale it down to the green spectrum here. We can call this "method 3".
4. Overall good agreements between the two methods and the range between method 1 and "mod method 2" should give us a good idea of the uncertainty on the final pion contamination
5. All above is for HMS. Still working on HMS.
4. Mingyu on hodoscope check, see elog28
1. d2n runs from March 2020 have the same difftime range as A1n runs. So we should use two sets of difftime cuts: one for "spring" and one for "summer".
2. A1n hodo calib works for all d2n settings except the 7.5 GeV setting. For 7.5 GeV, did another calibration that seems to work fine for all runs at this setting. So we could use two calibrations: one for "low-medium momentum" setting and one for 7.5-GeV setting. But consulted Mark, and he said the beta spectrum should not change with momentum. So we may still have some problem here.
3. Will continue checking all runs for difftime and calibrated beta spectrum.
5. Xiaochao will send her SAMC code to Mike N. (It's fortran!)
6. Looks like we are on-track for launching the reply before the winter break. Will need some farm/batch scripts. Bill will look into it (looks like John Matter is the expert here) and tell us (MMMJ) how to use them. We could split up the runs to 4 sets so everyone can practice using farm.
7. There are a couple of conferences with deadline in early Jan. We should discuss this by the end of year (if and when we get invitations) and get us organized on a full coverage of talks - APS April 2021 and DIS 2021.

1. Burcu warned us that we have been using the same momentum shifts in our replay as J/Psi and CT. About momentum offset: this can only be determined from precision ep data. We will continue using what we have been using but keep this in mind when we do precision e-3He elastic analysis. (Not important for DIS). Mark commented this shift is typically at the 0.2-0.3% level for SHMS and ~0.1% of HMS if P0 is below 5 GeV.
2. Optics: Holly commented she has been replaying data without the momentum offset. She is not optimizing delta so this offset should have little effect on the optics of SHMS she has been working. Tried xtar correction but so far not working (yet).
3. Both Holly and Mark will send the optics files to Sylvester for merging.
4. Carlos on helicity scaler coding: showed some screen shots of current work. One problem is the clock does nto seem to move forward. Another problem is the 1MHz clock rate does nto look exactly at 1MHz. Suggestions: (1) Is it possible the scaler event clocks are read differently and what is shown is actually the new reading every (1/120) seconds? In this case, one would see values different from 8243 or 8244 at the start and end of the run where a helicity cycle is cut short. (2) There is a "undefined" helicity time window every time the helicity is flipped so one should expect to see not 8333 (1E6/120) but ratehr some values smaller. That's maybe why we always see 8243 or 8244.
5. Mike on elastic QE: looked into details of all Eloss and MSC needed and will implement these into mc_single. Will ask Chao and Mike Paolone for the laste SAMC code.
6. Mingyu on hodoscope checking:
1. Tried another d2n run but all timing shifts persists. But applying A1n calibration seems to work okay for these runs if applying very wide timing cuts. (Both HMS and SHMS). So maybe the only thing that needs changing is the timing cuts?
2. When trying to update timing cuts for the newly fitted d2n peaks, ran into (possibly) formatting problems that result in some histograms being empty. This may be the reason why previouse attempt to rerun calibration using d2n runs failed (reported last week but did not write down in minutes - XZ). Will look into this formatting problem next.
3. Junhao double checked the script and got the same results (shifting in all channels).
4. Next, Mingyu will check d2n runs from March and see if we need to split d2n runs into Spring vs. Summer and using different difftime cuts.
7. Murchhana showed some simulated spectra for the glass windows and 3He (no xs weighting). Will be good if we can compare these with real data with the latest optics implemented.
8. Xiaochao received the first QE calculation from the Golak group for the 8.5 deg setting only, see elog25. These are for 0.5-deg theta and 10-MeV E' bins, 10 times coarser than our original request.
9. Next week we will meet at 3pm.

1. Optics: Mark HMS optics is ready, will email Sylvester for merging. Holly: will work on xtar correction for SHMS (expect 4 times smaller effect than HMS though).
2. Melanie on calorimeter calib:
   1. showed pion peaks in SHMS preshower blocks, some blocks are shifted from the expected value of 80 MeV. Next will make plots for the shower blocks. Need to use cuts for single block only (so the pion event goes through the whole block and does not share energy with other blocks). (XZ) emailed Vardan for his scripts and received them within the week.
   2. But what do we do if we do see shifts? Discussion: 1) first we may want to calibrate all blocks using pion peaks (using Vardan's scripts); 2) then compare the two params for the pion and electron peaks in Etot to see if one is visibly better than the other. 3) In principle, the pion method should be better since it removes the strong correlation of params of adjacent blocks.
3. Mingyu on hodoscope check of d2n:
1. For unknown reasons, all d2n hodoscope panels show different timing (by 5 to 10ns) from A1n data. All difftime cuts need to be refitted for d2n. What is causing the shifts??
2. To double-check, MIngyu will use a different d2n run. If the shifts remain then will study if we need a different hodo calib for d2n.
3. Also to double check, Junhao will try to reproduce the plots using a d2n run.
4. Pion peak check in calorimeter -- next time.
5. Junhao: d2n DC calibration seems fine.
6. Melanie on PID:
1. After applying low calorimeter cuts on the pion samples, now the pion rejection for NGC is much higher. (where is the elog?). Also found that the "peacock feather" structure in the PSH vs. Etot plot is from binning effects.
2. Realized the previous pion contamination wasn't calculated correctly (these were shown on page 39 and 47 in elog 24 slides). What needs to be done is to: Method 1: make one Etot/p spectrum with preshower electron analysis cut (blue), one spectrum with preshower electron cut and NGC pion sample cut (red). After that, apply Etot electron cut and calculate the ratio of integrated red to integrated blue (integrate above the Etot cut). THis would be the pion contamination without NGC rejection. Divide this ratio by NGC's rejection factor that corresponds to the NGC electron cut of our choice should give us the final contamination. Note: if the red spectrum clearly shows an electron peak then will need to use fits (sum of a quadratic function for pion and a Gaussian peak for electron) to separate electrons from pions and take only the pion fitted spectrum to integrate when calculating the pion contamination.
3. Or using Method 2: make the same blue spectrum as in method 1 with preshower electron analysis cut, then make one spectrum with preshower analysis electron cut and NGC electron analysis cut (green). One should see the green spectrum is dominated by the electron peak but some pion background may remain. After that, do a two function fit (e+pi) to the green and see if can separate electrons from the pion background. If the green spectrum is too small at low Etot/p, may use the value at low Etot/p as an upper limit of pion level. Now, apply Etot electron cut and calculate the ratio of the two functions (pi over e) in the two-function fit of the green spectrum. This would be the pion contamination with NGC rejection (since NGC cut is already in the green spectrum).
4. Compare method 1 vs. 2 to make sure the results are consistent with each other.
5. Note: applying the NGC electron analysis cut will change the pion contamination slightly since the cut efficiency is not 100%. But this should be a small effect.
6. For a graphical explanation see PID_Black_Friday.pdf
7. Carlos reported having technical difficulties to create the helicity scaler tree. -> see more details the week of 12/1. Also Hall A does not have a class for writing helicity scalers.
8. Mike Nycz will be doing the elastic/QE analysis! We discussed the difficulties of modifying mc_single for the elastic setting: 1) focal-plane matching (important for scintillator bar matching); 2) multiple scattering in the spectrometer: to do this precisely need cosy for magnetic transport of the MSC spread. THe best we can do without such COSY model is to use approximations.

1. Optics: Based on last week's progress, tried 5th order and also more iterations but did not see improvement for the negative z foils. Will summarize next time.
2. Calibration check following last week's merge:
1. Murchhana on NGC check: all SHMS are good; all HMS ref time plots have less background than previoiusly, while all difftime and calibration histograms are the same as before. (Note: this was solved the week after. discovered a wrong rootfile was used.)
2. Melanie on Calorimeter check: all difftime and calibration histograms look good. Pion peaks check -> need to wait for next time.
3. Mingyu showed hodoscope compairson of old replay and old calib vs. new replay with new calib. Slight improvement for both HMS and SHMS. Hodo diff time is also good for A1n data set. Will email this to Brad for merging and proceed to d2n data check.
4. Junhao: previous DC calib already used Mingyu's hodo calibration so should be good.
5. Jianping is wondering if we can use beta from hodoscope for PID. However the 2m drift time of the hodoscope itself won't be long enough so we will need the 22m from the target. In this case we must use RF (beam bunch) time as the starting time and so far this is not in the hall c replay code. Resolution of scintillators is in the order of 300ps (Mark). WE may revisit this later for positron run analysis (where background is huge and we need to reject protons too).
3. Melanie showed PID work, see elog24
1. Showed calorimeter pion rejection which is around 4000+/-1500 (on page 5, to be more specific, the rejection is 27641/7 and error bar is in the order of 1/sqrt(7)). To do: Will look into higher statistics and maybe a better way to evaluate error bar. (Evaluation of the error in N1/N0 where N1 is a subset of N0 has always been a topic of discussion!); Other comments:
2. Showed NGC pion rejection, still has problem because only applied "high cuts" on the calorimeter for the pion samples. There appears to be large background in the calorimeter spectrum that is near zero in both directions. These must be removed by using "low cuts" on the calorimeter for the pion samples.
3. There is no need to apply window (z) cuts when doing PID.
4. For evaluating pion contamination (for example in the Etot spectrum on page 39 and 47), got about 1% pion contaminationfor the low mementum setting with high PID cuts, electron eff is about 90% (low); We noted that need to separate real pions from the knock-on electrons (the spectrum shows small electron peak after the pion cut is applied). -- see update on 11/24 though, the 1% result is not correct.(in a good way!)
4. We discussed briefly how to use simulation to extract cross sections (mostly for d2n). Bill mentioned F2 is using the simulation code to fully simulate the yield (with both internal and external radiation losses) and then compare with yield from data. Then, the experimental xs is (Data/MC)*(xs)_rad. We may not use the same method. (Jianping promotes the method of using simulation to get the acceptance only), correct the data yield by the acceptance to get the cross section.
5. Along the same simulation topic, we noted for elastic or QE analysis, we need to fully rely on simulation for the itnegration of xs model and acceptance. To do so we also need to cut out region where the simulation does nto agree with data on the focal plane.
6. Next step: Melanie will work on analysis script in preparation of the first cooking; Junhao will look into pressure curve, Murchhana into simulation (for learning), Bill will get the Moller reanalysis organized (now 3%, will be 2% after re-analysis, has a plan from Dav G. and may take 4 person * weeks); looking for a plan for elastic analysis.

1. Optics: Mark reported trying more iteration on xtar-xptar fitting. For negative z, more iteration helped the optics at 18 deg. At 30 deg, it is also much better but the z=-20 foil is still not that perfect. Will try more with iteration and then the odd-order terms suggsted last week. Holly: will try the xtar fit this week for SHMS (meanwhile, latest SHMS optics can be found at doc1101).
2. Murchhana on NGC calibration: HMS methods 1 vs. 2 and the difference is 1/8.4 (method 1) vs 1/6.2 (method 2), see HMS_CER_Calib_method1_d2n.pdf, HMS_CER_Calib_method2_d2n.pdf and SHMS_NGCER_calib_method2_d2n.pdf. Previously we thought the difference could have come from extra electronic noise that widens the peak, but a short check shows the change in gain (param) is the opposite of what we would expect from a noisy widening of the peak. Please note that the NGC spectra in the method 1 slides show very clear 2- and 3-pe peaks and their position seem to confirm the results from method1.
3. Calibration organization and a forward flow plan - Brad: checked all history of 4 detector calibrations. All but hodoscope look fine and have been pushed to github. Hodoscope calibration looks was done based on a very old replay code and should be repeated. To do:
1. All will download the latest "merge" repo from the directory Brad provided;
2. Mingyu will repeat hodoscope calib;
3. all other 3 detector timging and calibration will produce new sets of checking plots to make sure things are okay;
4. DC will need to be re-calibrated after hodo is done.
4. We discussed calorimeter calibration. Brad noticed a lot of params are different between different kinematics. This should not be the case. We noted this is probably because when normalizing E/p for electrons to be 1.0, the params of adjacent blocks are highly correlated (or, there are multiple minima for the same fitting). In principle, if the blocks had been gain-matched (they were, someone mentioned Simona did that using pion or MIP peaks), then there should be little variation in the params from block to block and from kine to kine. We decided to use pion peak to check that all blocks are aligned. (We have a lot of pions!) Melanie will make the plots, basically, use NGC to select pions and plot E/p for each block.
5. Brad requested to have the latest optics in the repo. (Mark said 1-2 weeks, Brad: that's okay).
6. Scaler readiness - Carlos: Took over from Melanie and working on adding the helicity scaler tree to hcana. All other scalers are already in the tree but the helicity scaler is different. (We noted if Hall A can do this in Podd then C should be able to do this in hcana).
7. PID - Melanie reported on PID (see elog21:
1. PID of DIS low momentum setting. Pion contamination is higher (as expected for the low momentum). Can achieve 1% pion contamination with electron efficiency 90%.
2. Pion sample is still not clean - really need to add "low cuts" to both preshower and Etot to reject the background. WE noted there is a strutured background in the ps vs. Etot 2D histogram that looks artificial. Maybe some integer problem? should we look into this?.
3. Note: checking the calorimeter params and make sure all pions line up in all blocks may help to narrow the pion peak and reduce pion contamination.
8. Analysis script (lower level) readiness: extracting electron and pion counts with all PID, acceptance, and good track cuts. Cut beam scraping. Cut beam trips(?), kinematic binning. Melanie volunteered on this because she had worked on this previously.
9. Simulation: We have the latest mc_single_arm and can now make a plan:
1. DIS: Bill mentioned for F2, they take Tree from mc_single_arm, and combine it with DIS cross section xs(E',theta) that has both internal and external radiation included. (So, external radiation is a simplified version taking the zreact=0, since there is no Z dependence in the model xs). Then, take (Data)/(MC)*(xs)_rad to give "measured xsec (from data)".
2. For QE, we will have Born xsec and Asymmetry, not radiated. But we can do the same since mc_single output Tree has both vertex and radiated kinematics.
3. For elastic we will need to modify mc_single_arm. (XC, Mingyu?) -- Thank goodness it is still in Fortran!
4. What do we do if simulated FP distribution does not look the same as data? This is important since we will need to simulate scintillator plane on/off to match data. Answer: cut out region where the disagreement is large (that's nearly all negative z region).
10. General discussion on analysis plan: 1) We would like to see if the 121A group can contribute manpower to elastic/QE analysis. 2) We will need to proceed to simulation work. Murchhana will pick up this for d2n; 3) Junhao will start look into pressure curve analysis (no simulation needed, all relative analysis), principles and script preparation.
11. Moller reanalysis: Bill has a list from Dave a while ago. It will take about 4 person*weeks. Now we have 3% Moller. With reanalysis can get to 2%. Will see if can do this with Murchhana in the next 3 months (complete by early spring).
12. Other: Xiaochao sent QE calculation need to 3 theory groups last week. For details see elog19

1. All 3 DNP talks went well!
2. Mark on HMS optics:
1. Found a problem: xtar had been fixed by cosy (cosy-based recon from data) in the optics fitting code. This can be a problem because xtar can be at cm level for long targets and if cosy model is different from real situation. Tried fitting xtar in the code and it seems to have made positive z foils much better (no more curved pattern for xsieve vs. delta in the negative delta region!). On the other hand, neg z foils still have the same problem as before.
2. Plan: will consider using 3rd order coefficients for xtar fitting (since only odd power will cause a difference between pos and neg ztar).
3. Overall, looks like the problem is half-solved and we are looking forward to further progress on this next week1.
4. Note that this effect is large for HMS because (dxtar/dztar)=(1mard/1mm)*xtar. The slope is only about 1/4 of this value for SHMS but Mark will contact Holly (not at the meeting) to see if SHMS can be improved too.
3. Brad: still need the foundation of what was used in everyone's calibrations. In progress...
4. Xiaochao proposed we set a goal of having everything ready to go (repeat calibrations, iterate caibrations, and finalize optics) by Thanksgiving. This way we can get the scripts ready in December and launch pass1 replay by X'mas, take two week of vacation, and come back ready to analyze! Everyone agreed. (IN other words: if we are not ready to replay all data, then no X'mas...)
5. Melanie presented SHMS PID that now includes NGC efficiency and pion contamination using A1n's high momentum DIS data, see A1nd2n-ana-log18.
1. NGC pion rejection is very low, around 10. Typically should be 100-500. Could be that the pion sample cuts are not tight enough (there are only the high cut each on prsh and sh, not lower cut. Suggest plotting the pion region larger (and with NGC "less than something" cut to focus on the pions if possible). Some typical Hall A plots for pion and electron samples can be found from the 6 GeV Hall A test run page: LHRS and RHRS. To be exact, the links are LHRS samples, and RHRS samples.
2. Brad also suggested cutting out the mirror edges to see if that caused the low pion rejection. This is for understanding the problem at hand only, since for analysis we will probably include all regions.
3. By adjusting preshower, Etot/p and NGC cuts, now have reached 0.43% pion contamination while electron efficiency is still at the 95% level or above. This is very good. Will proceed to low momentum next (pion contamination may be higher and difficult to suppress to the 0.4% specified in the proposal), and then HMS.
4. We are reminded that we do have pion asymmetry data from 6 GeV A1n, g2n, and d2n data. With the high pion background for our DIS kinematics, applying the correction should not be a big problem.
6. Since we finished all discussions relatively early, we spent a few minutes to discuss the meaning of the subscripts in sigma_1/2 and sigma_3/2. Here are a few references in which these are defined: In summary, the subscripts refers to the helicity of the photon-nucleon system or the total spin projection of the photon-nucleon system.

1. Mark: tried single-foil fit but it didn't work any better. (Terms are very different but do not agree with data). Xseve vs delta has the same problem as before. We discussed why the optics does not work. Could it be that too many high-order terms are included and we do not have data to fit them all? Need to continue investigating.
2. Mingyu presented hodoscope timing checks, but all of them have shifts. We do not understand why. But while discussing this, we realized that: see next item
3. The calibrations of detectors may not have done in sequence. For example, Junhao's DC calibration was done using the old hodoscope params, not Mingyu's latest. In principle, hodoscope and DC (or any other correlated detector pairs) should be calibrated in iteration (hodo->DC->hodo->DC etc). Brad will talk to the students, figure out the condition under which each detector was calibrated, and figure out a master plan for going forward. (Will need to repeat some calib work). Meanwhile Carlos will update his document to reflect the flow of the calibration work.
4. We spent about 1.5 hours doing dry run for the upcoming DNP talks.

01/18/2018 Meeting to discuss A1n/d2n ERR:

12/12/2017 Meeting to discuss A1n/d2n ERR:

10/17/2017 Second meeting to discuss A1n/d2n ERR:

08/03/2017 First meeting to discuss A1n/d2n ERR: