Meeting minutes for first period:

2017

02/19/2017
  • One problem from experimental analysis: get solid angle. Solid angle is changing with number of binning. As number of binning increase, the shape at target look better (as what it is supposed to be).
    • Solid angle get from run phase space for central foil, then use 4D table to cut at focal plane and then look at target phase space.
    • Depend on number of bin, solid angle is different.
    • --> This solid angle help to solve part of the data cross section dropping as a function of number of binning. But problem still there.
  • Tried:
    • Cut in theta, make the cut become smaller, but results are the same, still 40% difference.
    • Different transport functions --> not help.
    • Limit of each focal plane variables in the table??? --> not this.
    • Create table for single  foil at center only --> not help.
  • To do:
    • Check 2D vs 4D: result from 2D cut was good within 10%. --> Check the consistent between 2 methods.

02/24/2017
  • Look-up table method does not work well. The boundary is well-behaved but too many events were cut inside the boundary. Less events are accepted as the number of binning increase.
  • I did a comparison just to get the sense of cross section from data and simulation with 2D cutting at focal plane (phifp vs yfp). and 4D cut at focal plane using look-up table. Detail can be found here.
  • In the plot, the distribution between 2D and 4D are different. So many events loss in 4D even they have the same boundary.
  • Now we have another new method, which using the transport function directly.

03/01/2017
  • New method to get cross section. This use the transport function, which is the lastest version, to make the cut at focal plane.
  • Detail how to make this cut:
    • The cut is combine of 2D (phi-y)+ 2D(theta-y) + 2D(x-y) at focal plane, where these 2D are defined by the sieve positions from the transport functions.
    • Use transport function: x, y, theta, phi (at focal plane) = function( y, dp, theta,phi --- at target). By fix 3 variables, we can get, i.e. xfoc = f(phi_target) with phi_target run from phi_min to phi_max
    • For a fix ytarget, dp, we can know limits of theta_target and phi_target. Keep one of the angle (either phi_target or theta_target) constant and make a "continous" point on the other variables, i.e. phi(i) = phi_min + i*delta_phi, phi run from phi_min to phi_max.
    • Do the same, then we can get the boundary of 4 lines.
    • There are 2 ways to make this cut:
      • From phi_min to phi_max, there are npoints, we can do fitting in this range and then just use a linear line ( y = a*x+b) to cut. Combine all 4 lines together to define 4D cut at focal plane.
      • Or we can use TCutG from ROOT to apply a polygon cut (which is a closed area defined by the points).
  • Detail how to get cross section:
    • Simulation:
      • Get solid angle by: run phase space + no radiation effect, apply cut at focal plane, look at target angle and get solid angle. In addition, this also give the boundary at target.
      • Run elastic carbon, count how many events inside the boundary at target (which is mentioned above) called "tacc".
      • Count how many events survived after focal plane cut, called "cacc". Sum cross section inside this cut called xsc.
      • Cross section = xsc/tacc.
    • Data:
      • Apply PID + dp cut, and 4D cut.
      • Subtract background. Result between simulation and data can be found here.
  • To do:
    • Why 25%, 32% different between data and simulation?
    • Check dp plot for simulation and data.
    • Why 4D cut at focal plane with the right boundary still can not reproduce an exact boundary at target?

03/15/2017:
  • Figure out the problem with experimental cross section using look-up table.
  • The problem is:
    • Only a small area in focal plane is covered for all kinematic (especially in y focal, as well as theta and phi focal).
    • Another limitation is the resolution of sieve holes at focal plane. From optic information in each dimension, there is an associate resolution.
    • Combine above 2 information, we can not make binning as many as we want. There is a limit for each dimension. As shown in here.
  • With these limitation, we can not get cross section better than 20% agreement between data and simulation.

04/05/2017:
  • Uncertainty in sieve survey was quoted as 0.5 mm but the final number should be use is around 0.2 mm. This uncertainty in sieve slit will transfer into uncertainty in target angles.
  • Number of binning in focal plane is determined by uncertainty in how well we know the sieve center. This is done by select 5 samples each with 5k events from 1 sieve slit hole at focal plane. This uncertainty is much better than the resolution. So, the table still not work.
  • Another method is to reconstructed 2 angles at target: theta and phi. Apply a cut at reconstructed quantities on both simulation and data to select events.
  • Comparing cross section result for central foil with different momentum settings are shown here.

04/26/2017
  • Reconstructed 2 angles and W (invariant mass). Analysis cuts are: cut to choose 1C 1D 2C 2D (square).
    • Cut on both reconstructed variables + momentum (dp) from simulation and data, as shown here. For dp=2%, 2D&2E was chosen instead of 2C&2D. Both results from 2 beam energy are ~40% off compare with simulation.
    • W cut was applied and Choose 2C&2D for all settings, one of the dp=2% (1.5GeV) become better but still the same for other beam energy, here.
    • W cut was applied in wider range (-0.005,0.01) instead of (-0.002, 0.004). And correct for the contribution from carbon excited state, here. The problem with dp=2% is still there for 1.1 GeV. Suggestion is that maybe the background contribution, normalization for bkgd is not right.
    • So far I have worked on single carbon elastic at center.

05/24/2017
  • Fix problem with 1.1 GeV, dp =2%. The reason is that run 1888 was used instead of run 1889. There is no end for run 1888. Updated result with run 1889 is shown in here.
  • Several problems has been appeared:
    • I figure out I used spectrometer angle 5.99 deg (which is Vince used in 2nd period) instead of 5.91 (which is from 1st period survey). The above slide have cross section with both angles.
    • We can not study multifoil. Reason: combination of mis-wired septum and target collimator block most of upstream and downstream foil.
    • I did try to optimize theta_rec with higher order (include cross terms) but results are bad.
    • 0% dp seem different from the rests. So I start to use sieve's center from run 1805 (single foil) instead of run 1866(multifoil). Then when I change the central angle, 5.91 deg, the cross section different between data and sim seem the same for both sieve position sets.
  • Uncertainty budget for simulation and data are listed in above slide.

06/28/2017
  • Study inelastic dp: choose inelastic carbon run.
    • Dp spectrum from simulation is obtained by using Peter Bosted model with internal/external radiation length (include the equivalent radiator). From dp shape of simulation and data, as shown in here, we can extract the acceptance.
    • To do: add in elastic tail from ROSETAIL to correct for the contribution from elastic tail.
  • Study extended target: we have problem with extended target. We took multifoil carbon with z=+/-10,0 cm but with target collimator IN. With this collimator, it will block most of upstream and downstream foil. Hence, multifoil with target collimator IN can not be used to study ytg acceptance. Instead, we do the following:
    • Choose a sieve hole which is seen by all target positions (get the count with target collimator IN). Only hole 2D is seen by both downstream and center carbon foil. The count is shown in this slide.
    • In above slide, run 1900 (multifoil with target collimator OUT) and run 1899 (single carbon foil with target collimator OUT) ytg are plotted with the 2D (theta vs phi reconstructed angle) cut. We see that, seem there is no contribution from upstream and downstream.
    • To do: Get cross section from sieve slit run (with sieve slit in but target collimator OUT).

07/12/2017
  • Study the acceptance with target collimator OUT:
  • To do:
    • Get acceptance function from these data point.

08/09/2017
  • Obtain acceptance of all sieve holes, shown in here. Seen holes are labeled in red color, not seen ones are labeled in black color.
  • After get the acceptance for each hole. Need to determine acceptance function. Acceptance function is determined by:
    • For 1 momentum setting, determine boundary for 3 foils together. And the chosen holes should be the same for all momentum setting. Determine 6 corners.
    • For each corner (from dp =-3.0%), get a linear line which go through these 2 points. Similarly get the rests 5 corners.
    • The acceptance function is then applied to sieve slit multifoi run (1866, dp=0 and 1869, dp=-3%).
    • Reasonable results are obtained, as shown here.
  • To do:
    • How to extend this acceptance to dp=+2%?

09/27/2017
  • The acceptance function works in dp range from (-3% , 2%).
    • The function is y_fp(dp) and ph_fp(dp). With each y_fp and ph_fp corner is a function of dp. For example: for 1st corner:
      • y_fp = -0.227*dp +0.010
      • ph_fp = -0.121*dp +0.027
      • There are 6 corners. From 6 corners, form a boundary, only accept event stay inside, throw events stay outside.
    • This function is added into simulation and data to apply same cut on both.
  • Different ice thicknesses were used for 1.1 GeV and 1.5 GeV: 1.5 GeV was taken before 1.1GeV. In my analysis, I used 8mm ice for 1.5 GeV and 15mm ice for 1.1 GeV. Cross section for 2 beam energies can be found here.
Then I start with extended target: He3.
  • Tried to get asymmetry, experimental asymmetry was extracted using this formula. Seem this formula is not right. Using this formula, dilution is corrected with N_0 term. This way, we don't need to get into N2 and He3 cross section ratio in order to extract the dilution factor. However, this solution did not give right answer for asymmetry with all dp and 2 beam energies. Asymmetry from simulation is not right either.
  • He3 spectra: ytg, phitg, thetatg, W are shown here.

11/15/2017
  • Get elastic 3He cross section for 1.1 GeV (dp = -2, 0, +2%), 1.5 GeV (dp=0%) and 2.2 GeV (dp=0%)
    • 1.1 GeV result agree between simulation and data are within 5%, results shown here. For this energy setting, I used 15mm of ice for simulation . These  ata were taken during 05/11/2003.
    • 1.5 GeV result is ok as well, as shown here.  W plot is not so good. On the low W side and high W side have problem. But cross section result is good within 5%. These data were taken during 05/04/2003 so I used less ice, 8 mm of ice.
    • 2.2 GeV has problem, a 30% disagreement. This run was taken on  05/17/2003, so I should have added more ice, ice thickness was used is 15 mm. There are several suggestions:
      • Septum settings were changed, I checked the set and read back value of septum current, it seems ok.
      • W plot is not so good, suggestions: angle reconstruction is not so good. So I can look at dp instead.
      • Optic is not good enough? The reason why I did not include 2.2 GeV optic into my optimization is because at that point, I did not have reconstructed W, so I can't determine the sieve center well because of the huge contribution from quasi elastic (of carbon).
      • Wrong normalization?
      • Use different runs.
      • Correct for quasi elastic contribution.
  • To dos:
    • Get n2 density inside 3he target.
    • From 2.2 GeV acceptance run (carbon target without sieve) determine ice thickness?
    • Add 2.2 optic data into optimization ??? Then, need to redo everything :(