Difference between revisions of "G2p Analysis Minutes"

From Hall A Wiki
Jump to: navigation, search
(→‎8/28/2013)
(→‎9/4/2013)
Line 30: Line 30:
 
match the target plane variables.
 
match the target plane variables.
 
**Will tune quartroples later.
 
**Will tune quartroples later.
 +
 +
*Jie:
 +
**Just back from China. Will work on writing the multi-tracks-analysis-note.
  
 
==8/28/2013==
 
==8/28/2013==

Revision as of 17:51, 4 September 2013

Minutes of the weekly analysis meetings

Agenda

9/4/2013

Present: Kalyan, Chao, JP, Jixie, Min, Jie
By Phone: Ellie, Toby, Ryan, Karl


Feature presentations:

  • Chao:
    • From last meeting we start to doubt the beam position. So Chao modified the program optimizor

such that it can also calculate the beam position at the end of the fit. The detail is that chi^2
is now defined as Chi^2=dX^2+dY^2, where dX and dY are the distance of projected-back-position to
beam position, and this projected-back-position is based on surveyed sieve hole position and reconstructed theta | phi angles. The program is ready to run and the result shows that the beam position given by BPM
on March 4 is OK while on March 14 is not correct. The uncertainty of the beam position fitted by this program
is about 1.5 mm.

    • His talk can be found here.
  • Min:
    • Gave an update on SNAKE tuning. The focal plane variable of March 4 can match the real data with uncertainty

when select only the center row. But can not match well with March 14 data. The septum current ratio optimized by
matching focal plane of these data set agrees with Jixie's Geant4 prediction.

    • Mannually shift the BPM horizontal position of March 14 optics runs, she found that the focal plane variables

can be matched within unsertainty when shifted from 4.5 mm to -3 mm or -4 mm. This indicates that the BPM position
of March 14 runs are not reliable.

    • Started to adjust aperture on SNAKE. JP suggested also using reconstruction matrix to do reconstruction and then

match the target plane variables.

    • Will tune quartroples later.
  • Jie:
    • Just back from China. Will work on writing the multi-tracks-analysis-note.

8/28/2013

Present: Kalyan, Chao, JP, Jixie, Min, Melissa
By Phone: Ellie, Toby, Ryan, Karl, Pengjia


Feature presentations:

  • Ryan
    • Looked at livetime asymmetries for all production runs over each momentum/energy setting and determined the mean
      value, error and standard deviation for each resulting histogram. The livetime correction to the yield is done on a run-by-run
      basis, but this method will allow Ryan to look for any overall systematic shifts in the livetime. JP suggested correcting the
      livetime asymmetry within each momentum setting with the "expected value" so that all runs can be combined to increase the
      total statistics. Also, each run should be weighted by statistics to make the standard deviation a meaningful representation of
      the uncertainty. His slides can be seen here.
  • Pengjia
    • Gave an update on the status of the BPM calibration. Showed the uncertainties provided by the survey group, which gives an
      uncertainty at the BPM of 0.25mm. He needs to contact the survey group to better understand the values provided.
    • Also showed a timeline of calibration settings for early in the run period (March). Looking first at BPMA, it seems that the
      calibration is fairly stable over this period. However, the calibration for BPMB is not very stable. It seems that using different "div"
      settings (2 or 4) caused the BPMB settings to change but did not affect BPMA. JP suggested looking at runs immediately after the
      calibration is done to see when the calibration for BPMB goes bad. Pengjia's slides can be seen here.


8/21/2013

Present: Kalyan, Chao, JP, Jixie, Min, Melissa
By Phone: Ellie, Toby, Ryan, Pengjia, Alex


Feature presentations:

  • Melissa
    • Updated previous asymmetry results to include charge/livetime/prescale normalization for all energy settings. Also
      showed the corresponding physics asymmetries. The 1.1 and 1.7 GeV settings need some work, as the scaling from raw
      to physics asymmetries doesn't quite make sense. Also need to look at asymmetries run by run to identify any possible bad
      runs. Her slides can be seen here.
  • Chao
    • Last time he found an offset between two sets of straight-thru data and is working on determining the reason for this
      offset. He first tried using only first order matrix elements for the optimization, but the resulting sieve pattern was not
      aligned very well. Next he tried including second order matrix elements in the optimization, using the data from a run with
      liquid helium. Using this database to look at the run without liquid helium revealed a horizontal offset of 7.3mm. A possible
      reason for this offset is that the BPM information is not quite correct. To check this he looked at simulations for two different
      beam positions. The results looked suggested only a -0.3mm horizontal offset. Finally, he compared the focal plane plots for
      settings with and without liquid helium. The bpm tells us there should be a ~7mm offset (~8mrad) difference between the
      settings, but the data shows only ~2mrad difference. Chao will work with Pengjia to try to determine if there is a problem with
      the bpm information. His slides can be seen here.
  • Min
    • Merged the db_run.dat and db_run.dat files for left and right HRS into one file to fix problems with replay on batch farm.
    • Working on tuning the SNAKE parameters. She found the best results by adjusting the position of the septum and HRS along
      the beam direction. Looking at the focal plane distribution for SNAKE and the data, there is a point where the SNAKE model "turns"
      away from the data. Moving the HRS from its original position (1.96mm) to 0 does not seem to help, but moving the HRS and
      septum both in the negative-y direction seems to correct for this "turn". Looking at a run with a large beam offset, there seems to be
      a much larger difference between the data and SNAKE. Further, the plot of x vs y doesn't seem to match at all. It's possible that the
      issue with bpm information that Chao found is causing similar problems for Min; she will try fixing the magnet positions and adjusting
      the beam position to see how this effects her results. Her slides can be seen here.



8/14/2013

Present: Kalyan, JP, Chao, Min, Melissa
By Phone: Ellie, Toby, Ryan, Karl, Pengjia, Alex, Guy


Feature presentations:

  • Pengjia
    • Showed an update on the status of BPM calibrations. He is currently using a method where he looks at the linear
      portion of the BPM current vs. signal, and extrapolates the fit to determine and offest. This offset is used instead
      of the pedestal value. JP is concerned about using this method - he and Pengjia will discuss it more offline.
    • Also mentioned the makeup of the uncertainty for BPM calibrations. It will come from the BPM hardware resolution
      (plus addition noise from radiation), uncertainty in the survey, and possibly a third contributer. The Beampackage is
      ready to calibrate bpms and rasters for all run periods, with the exception of the fast raster which still has some issues.
      His slides can be found here
  • Toby
    • Presented raw asymmetries from pass 2. The results looking promising but still need to be corrected for charge,
      livetime and prescale, and should use focal plane variables for acceptance cuts instead of "gold" variables.
    • Also, there are still a few issues with the replay for pass 2 that need to be corrected. His slides can be seen here.
  • Ryan
    • Showed the results of livetime asymmetry analysis. He cross checked his results using scalar totals from both the
      TTree and hel_ring tree, which also turned out to be a good check of the helicity decoder. In general, he found most
      runs had a livetime asymmetry of less than 200 ppm. He will check to see if the distribution of his results is gaussian.
      The results are also in the mysql database. His slides can be seen here.


8/7/2013

Present: Kalyan, Jixie, Chao, Alex, JP, Min, Melissa
By Phone: Ellie, Toby, Ryan, Karl, Pengjia

General Discussion:

  • Karl posted a couple of new papers to the wiki, they can be found here and here.


Feature presentations:

  • Melissa
    • Showed raw asymmetry results for 2.2 GeV, 2.5T setting. JP suggested only using focal plane variables (not
      reconstructed variables) for acceptance cuts until optics with target field is completed. Also, the prescale value
      needs to be included to account for total statistics. Once the issues with the second farm pass are resolved, she
      will include the normalization for charge/livetime and complete the analysis for all energy settings. Her slides can
      be seen here.
  • Chao
    • Showed an update on first order matrix calibration for the straight-thru conditions. Min did a study where there
      was helium in the target nose, now Chao is doing a similar study but using runs where there was helium in the target
      nose. Using the most updated BPM information, there is still a ~1.5mm offset in the horizontal direction. He is able
      to clean up the data by cutting on focal plane and y-target, then makes a dp cut to distinguish between helium and
      carbon peaks. Comparing his results for the first order matrix to the results without helium, the matrix changes quite
      a bit. He will try to figure out what is causing this change and will compare the results to the SNAKE model. His slides
      can be found here.
  • Min
    • Working on matching SNAKE model to the data. Showed the results (for two sieve holes) of moving the septum and/or
      HRS magnets by 1-2mm. There seemed to be some improvement, but she will continue to study this. She will also look
      at the rest of the sieve holes to see the distribution/acceptance. Her slides can be found here.



7/31/2013

Present: Vince, Kalyan, Chao, Min, Jixie, Melissa
By Phone: Ellie, Toby, Ryan, Pengjia, Alex

Feature presentations:

  • Pengjia
    • Gave an update on the status of the BPMS. Showed plots of current vs raw BPM ADC values for several different
      beam positions and saw that the distribution is linear above 30nA. Using the offset of the fit to the current,
      he can determine the pedestal for each channel. Overall, the results look much better using this method,
      although BPMB may still have some issues. Also showed a plot of raw current vs entry number. The spread in
      position looks better then before using the new values for the pedestals. Next he will try to improve his
      method to minimize current dependence and will try to use a non-linear fitting of calibration data. His
      presentation can be found here
  • Toby
    • Working on pass-2 of farm replay. There were some issues using the batch farm, but they seem to be fixed now.
      Rootfiles should be ready soon!
  • Ryan
    • Was delayed by problems with helicity decoder, but should be able to finish helicity dependent deadtime analysis soon.


7/24/2013

Present: Vince, Kalyan, Chao, Min, Jixie, Melissa
By Phone: Moshe, Pengjia, Ellie, Toby, Ryan, Guy

Feature presentations:

  • Chao:
    • Gave an update on simulations using the straight through optics conditions. He showed results comparing
      the reconstructed and thrown variables for 3 cases. First, assuming an uncertainty in bpmA/bpmB of 0.2/0.4 mm.
      Second, assuming no resolution in the bpms . And third, using only bpmA. From this he concluded the BPM
      resolution will contribute 2e-4 to the delta reconstruction. A question was raised as to why the phi distribution was
      so wide. Chao will check this for next time.
    • Also updated the simulation package. The most recent version of the event generator uses a “random walk”
      algorithm to generate particles following the cross section distribution. The simulation code is available for use,
      and Chao will work on producing some documentation.
    • Next up is working on optimizing the optics matrix for March 4th data to help Min, since the simulation package
      will need input from her SNAKE model. His presentation can be found here.
  • Min
    • Gave an update on her SNAKE model. Using the most updated beam package and looked at the distributions of
      the current, bpmA, bpmB and target variables versus entry for run 3185 (for both x and y). The distribution of the
      target variables seems to follow the bpmB distribution. She selected a region of events in this run to remove
      oscillations. Looking at the results for the uncertainty from the first order matrix, the uncertainty seems to be smaller
      then before with delta-y = 1.90mm and dela-phi = 0.68mrad. However, looking at the current results for tuning the
      septum field, the comparison between SNAKE at the data is not much better.
    • Next she will try moving the location of the quads by 1-2mm and comparing the trajectories of runs 2725 and 3185.
      Her presentation can be found here.

7/17/2013

Present: Vince, Chao, Min, Jixie, Kalyan
By Phone: Ellie, Alex, Toby, Moshe, Pengjia, Karl

Feature presentations:

  • Pengjia:
    • Present the pedestal as a function of time for the whole data set. The pedestal were
      drifting along time. They were not stable at all. Currently Pengjia subtracts only the averaged pedestal
      of a period of time to reconstruct beam position for that period. He will need to do the subtraction run by run.
    • Also presented are the BPM calibration against Harp position for 100, 75 and 50 nA. It seems that the
      calibration for 50 nA need to be redone. He also shown that the calibration constant is sensertive to
      beam current. Whenever the beam tripped and then the beam position is not stable during the current was
      ramping up.
    • See detail here.
  • Moshe:
    • Working on an event generator for simulation. Right now he is working on the energy loss caculation
      using both EStar and Geant4. He mentioned that the energy loss result from EStar
      does not agree with that from Geant4.
    • He is also trying to get cross section models for helium-4 and nitrogen-14. He has QFS model
      in hand but this model is known to be poor. Vince agreed to provided him with some N-14
      data and Moshe will try to compare P.Bosted's model with Vince's data. Then decide which model to use.
    • See his presentation here.

7/10/2013

Present: Vince, Chao, Min, Jixie, Melissa
By Phone: Ellie, Ryan, Toby, Moshe, Pengjia, Karl

Feature presentations:

  • Chao
    • Gave an update and summary of the helicity decoder There were a couple of bugs that needed to be fixed,
      including the one bit delay. Also, the ring buffer creates a set of unused, extra events at the beginning of each
      run, which will now be marked as an error. Finally, there was an alignment issue when inserting helicity information
      back into the rootfile. These issues should all be fixed now. When using the decoder, the variable that should be
      used to obtain the correct helicity of an event is "hel_act", and all events with hel_error /= 0 should be cut out.
    • Also working on improving the decoder to deal with beam trip cuts. The cut should be made in units of helicity
      patterns, so time information (from the fast clock) will be added to help make these cuts. His slides can be found here.
  • Melissa
    • Gave a summary of her method to determine raw asymmetries including the applied cuts and method for extracting
      the asymmetry. However, the issue with the helicity decoder described in Chao's talk means that the results shown are
      not reliable. Concern was expressed over the error bars shown, she will carefully check this when she re-does the
      analysis with the correct helicity information. Her slides can be seen here.
  • Min
    • Updated the LHRS optics calibration by including beam trip cuts and using the latest bpm information from Pengjia.
      The results seem better! Cuts still need to be applied for when the beam moves. Since BPMB is not always stable, she
      thought about just using BPMA to calculate the beam position at the target. Vince suggested that by correcting for the
      beam movement the poor resolution in BPM B might be less significant and Min could still get reasonable results. Her
      slides can be found here.



7/3/2013

Present: Kalyan, Chao, Min, Melissa
By Phone: Ellie, Ryan, Toby, Pengjia, Alex

Feature presentations:

  • Ryan
    • Updated the MySQL database to include Jie's results from multi-track analysis and Toby's results from target
      polarization analysis. This information is also now included in the ROOT library. Also fixed several bugs that caused
      missing information for some runs in the database. He is also working on helicity dependent deadtime analysis.
      His slides can be seen here.
  • Pengjia
    • Gave an update on BPM analysis. He compared two techniques to reduce noise in the pedestal; applying a filter
      and taking the average of every two seconds of data. The two methods seem to give similar results. For next time
      he will look at how the change in current affects the position, and how using a 2Hz filter affects the mean value of
      the BPM. His slides can be seen here.
  • Toby
    • Finished writing a technote for the target polarization analysis and will be looking for feedback soon. He will also
      start preparing for a second round of farm replay, since no event-type information exists in the rootfiles currently.


6/25/2013

Present: Karl, JP, Chao, Kalyan, Min, Melissa
By-Phone: Ryan, Toby, Ellie, Pengjia, Alex, Moshe

General Discussion:

  • Ryan will submit an abstract for a general g2p talk for the DNP meeting this fall.


Feature presentations:

  • Chao
    • Gave an update on optics simulation package. He showed the results of comparing the reconstructed and thrown
      variables from simulation to see if he could reproduce the central value of delta, theta and phi. He determines the
      average for each variable by taking the average of each point and weighting by the cross section. Currently, his
      reconstruction stops at the target plane with z=0, which assumes that the target is symmetric along the z-direction.
      He needs to find a careful way to correct for this. His presentation can be found here
  • Min
    • Gave an update of LHRS optics calibration. She worked with Pengjia to get the bpm package with filter working
      for ifarm replay (only LHRS for now, RHRS will be working soon). She used the updated bpm information to update
      her results for straight-thru optics. She found that the results for theta were larger by a factor of ~3. She will try to
      figure out if this is a result of the bpm filter or something else. She will also update her results to include cuts for
      beam trips and beam moves. Her presentation can be found here
  • Melissa
    • Gave a followup on beam charge asymmetries. She updated her results to include beam trip cuts to compare with
      Pengjia's results. The effect of adding the beam trip cuts is small (~ >0.05%), and Pengjia and Melissa's results agree
      within ~1% or less. She also showed a comparison between results taken from ring buffer data and happex data.
      There is a >1% difference between the two systems. Her slides can be seen here


6/19/2013

Present: Karl, JP, Vince, Jixie, Ellie, Kalyan, Chao, Jie, Min, Melissa
By-Phone: Ryan, Toby, Pengjia, Alex

Feature presentations:

  • Ryan
    • Gave a recap of the y-target calibration for the LHRS. Last time he showed that he got strange results
      during the optimization, specifically for the (0,0) and (-4,0) settings. By using a different run for
      the (-4,0) setting he was able to improve the results. He then tried maximizing the event number per hole,
      while still keeping the number of events approximately the same per hole. This improved the resolution of
      the results. He also tried increasing the order of the optimization in the database. This also improved
      the resolution, but may not be a reasonable solution since increasing the order can over constrain the matrix.
      His presentation can be found here.
  • Pengjia
    • Gave an update on BPM analysis. He is currently rewriting the insert and calibration piece of the
      beampackage to include the new filter method presented last time.
    • Also working on a beam move check that uses rms detection for fast beam moves, looks at current trip
      information, and splits events if the beam moves more than 0.3mm. He is still debugging this. He
      also discussed pedestal subtraction for the bpm calibration. The pedestal shown was quite wide and had
      multiple peaks. Pengjia will study this more to determine if this behavior is consistent through all runs,
      or if this was a specific period when conditions were changing. His presentation can be found here.
  • Toby
    • Worked with Josh and James to determine that there was in fact a bug in the online code to determine
      target polarizations. This means the 2.5T polarization is actually ~50% of what was reported online.
      This does not affect the 5T results.
    • Showed details for his method of determining the TE points. He takes the time reported online for the
      TE as the approximate start time, does a 0th order fit of the points and looks at the chi-squared value.
      He then adds points until he finds the end point of the TE. Finally, he adds points at the beginning of
      the TE to minimize the chi-squared value. The polarization is then averaged over each run. He questioned
      how the polarization decay contributes to the uncertainty for each run, since the poarization is actually
      decaying over each run. JP suggested using the average of the decay curve to determine this. Toby's
      presentation can be found here.
  • Jie
    • Has completed the efficiency study for multi-track events. This information will be in the mysql
      database soon.
    • Also working on data quality checks for optics variables. He looked at the t0 calibration first. One
      concern is how "negative time" events are being dealt with. Jie will check that they aren't being cut out.
      He also looked at the transport and rotated x,y,theta and phi variables. Within each kinematic settings,
      these varibales look stable over all production runs. Jie will take a closer look the elastic settings next.
      His presentation can be found here.

6/12/2013

Present: Karl, JP, Vince, Jixie, Chao, Jie, Min, Alexandre, Kalyan, Melissa
By-Phone: Ellie, Toby, Ryan, Moshe, Pengjia

Feature presentations:

  • Chao
    • Has fixed the bit-shift problem in the offline helicity decoder. The next round of farm replay will
      have correct helicity information.
  • Melissa
    • Showed a comparison between her and Pengjia's results for beam charge asymmetries. The results
      show ~0.5% difference. It's possible that the beam trip cuts that Pengjia used is the reason for this
      difference. Her presentation can be found here.
  • Min
    • Showed an update of angle and vertex matrix calibration results using previous bpm calibrations. The results
      are not better. She is still working on improving it by applying beam current cuts and beam position cuts.
      She will also try using a different run to optimize that has the same conditions but a more stable current
      She will check that the beam position is also stable during this run. Her slides can be found here


6/5/2013

Present: Karl, Vince, Kalyan, Chao, Jie, Min, Melissa
By-Phone: Ellie, Toby, Ryan, Moshe, Guy, Pengjia

General discussion:

  • There will be a practice talk for Min's Hall A Collaboration Meeting talk sometime soon (Friday or Monday?)


Feature presentations:

  • Moshe
    • Gave an update on the status of GEp analysis. He did a first round of analysis without calibrations,
      to determine a general procedure for analysis. He showed preliminary results for elastic peak
      identification, binning optimization, dilution factors and asymmetry extraction. He is also working
      on writing a GEp event generator for HRSMC. His presentation can be found here.
    • Details of status of GEp analysis can be found in this document, written by Moshe.
  • Pengjia
    • Working on improving BPM resolution. He added a low pass, software FIR filter when processing
      the data, which seems to work very well. The previous results were 7-8X larger then the results that
      have been processed with the filter. He is working on incorporating this into his beampackage code,
      and will repeat his BPM noise study using the filter. Will also check if the central value changes at all
      as a result of the filter. His presentation can be found here.
  • Ryan
    • Working on the LHRS Y-target calibration. He optimized 3 runs in the 2.2 GeV (0T, 6deg) setting
      with different beam positions; (4,0), (0,0), (-4,0). The results seemed strange, specifically for the (0,0)
      and (-4,0) settings. He first tried optimizing each run individually, which gives results that make sense.
      He then tried optimizing them in pairs, and got reasonable results except for the combination of (0,0)
      and (-4,0). Finally, he tried to check these results using other runs with the same beam positions, but
      found that during these optics runs, the beam position was never moved back to (0,0). Vince pointed
      out that the typical resolution for y-target is ~1mm, so these results may be ok. Ryan's presentation
      can be found here.
  • Toby
    • Showed updated target polarization calibration constants. He decided to use a 3rd order polynomial
      to fit the wings of the baseline-subtracted signal. It's possible that the discrepancy between using a
      2nd and 3rd order polynomial could result from the signal "bleeding" into the wings. He described his
      method for minimizing the reduced chi-squared to find the TE points. He starts with a set of 15 points
      where the target is most thermalized, and adds points from the beginning/end of the TE to further reduce
      the chi-squared.
    • He also found that the 2.5T calibration constants found offline are ~50% smaller than what was determined
      online. He will work closely with James/Josh to confirm that this really is a problem. His presentation can
      be found here.
  • Jie
    • Showed an updated version of multi-track efficiency analysis. He determined that his new method of
      including background cuts is not as reliable as his previous method to determine uncertainty. By including
      background cuts, the systematic uncertainty is decreased by about 20%, but it very position dependent, so
      he suggests staying with his previous results. His presentation can be found here.


Jan-May 2013

Minutes_Jan2013_to_May2013

April-Dec 2012

Minutes_Apr2012_to_Dec2012

Jan-March 2012

Minutes_Jan2012_to_Mar2012

July-Dec 2011

Minutes_July2011_to_Dec_2011

Jan-June 2012

Minutes_Jan2011_to_June_2011

Retrieved from "https://hallaweb.jlab.org/wiki/index.php?title=G2p_Analysis_Minutes&oldid=19262"