Project

General

Profile

Support #1873

Cooling channel optics data analysis from July run

Added by Liu, Ao over 7 years ago. Updated about 7 years ago.

Status:
Closed
Priority:
Normal
Assignee:
-
Start date:
20 September 2016
Due date:
% Done:

100%

Estimated time:

Description

This issue tracker is used to keep track of the optics data analysis for the July run.

This will include magnet alignment studies and beam optics studies.


Files

Screen_Shot_2016-09-20_at_11.40.36_AM.png (375 KB) Screen_Shot_2016-09-20_at_11.40.36_AM.png Takes forever to process one event.. Liu, Ao, 20 September 2016 17:41
SepRunData.pptx (6.04 MB) SepRunData.pptx Liu, Ao, 26 September 2016 07:20
#1

Updated by Liu, Ao over 7 years ago

Here is the list of Chris' ideas, and my comments.

  • Build a table; store each event spill number and recon event number;
    make a list of boolean flags, for each event flagging whether it passes
    or fails a given cut. The point is you will need, as the analysis
    matures, to make plots of e.g. the events failing a particular cut
    ****Ao's comments: Done. I have scripts to extract data from recon, and make specific cuts. One can apply cuts as one needs. I'll create a github project to upload the files;
  • For each event, reject the events if they do not meet sufficient
    quality criteria:
    1. single space point in each TOF
    2. single track in each tracker
    3. > 10 track points in each track (there are up to 15, one for each
    tracker view)
    • Ao's comments: Partially done. Currenlty it requires single track in each tracker, a user-defined number of track points in both trackers, but it does not require single space point in each TOF yet;
  • Extrapolate the events from TKU to TKD. Add another cut
    4. reject events if they go near to an aperture. Initially you can just
    use a e.g. 20 mm fiducial cut; later on you may want to do something
    more clever. The point is, particles which, when extrapolated, go near
    to an aperture, have some finite probability of scattering out of the
    accelerator and this makes a bias in the analysis. We want to reject
    events which bias the measurement in this way.
    • Ao's comments: I think, what will bias the measurement more could be the ones that are not filtered by the cuts. i.e. ones that should have left the tracker volume but did not because they were scattered into the volume. It shouldn't be too hard to filter them out though.
  • Plot the residual between the extrapolated TKU track and the measured
    TKD track in x, y, px, py. e.g. dx = (TKU extrapolated x) - (TKD x).
    • Ao's comments: Almost done. Struggling with MAUS. It's being so slow and I don't know why.
  • You may also like to plot dx against x, y, px, py
  • Check for features in these residuals plots. Do the plots look
    reasonable? I would expect ~ 10 - 20 mrad residuals in x' and ~ 10 - 20
    mm residuals in x, based on typical track widths.
    • Ao's comments: I think I can do this by next week. Of course, probably with some help on MAUS... (bugging Durga because I experienced this slowness already)

Attached is my MAUS run with a G4BL formatted beam as the input.
It's stuck there forever to process one event.

#2

Updated by Rogers, Chris over 7 years ago

Ao's comments: Almost done. Struggling with MAUS. It's being so slow and I don't know why.

How are you doing the track extrapolation? What are your config.py/etc?

#3

Updated by Liu, Ao over 7 years ago

Hi,

The data cards I changed include:
beam = {
"particle_generator":"file", # routine for generating empty primaries
"beam_file_format":"g4beamline_bl_track_file",
"beam_file":"/home/aoliu/maus/MAUS-v2.5.0/data/08154_recon.datTK0STN5.gbeam",
"file_particles_per_spill":100,
"random_seed": 10,
}

default_keep_or_kill = False

#4

Updated by Rogers, Chris over 7 years ago

Try:

maximum_number_of_steps = 10000

#5

Updated by Liu, Ao over 7 years ago

Changing max. steps did not help. Suspect that the complicated, segmented geometry slows things down. Consulted Durga and was told that this problem is on his to-do list, but no clue yet. Durga will try to set up the GRID to run tracking for me, hopefully soon, or the tracking will take forever.

#6

Updated by Liu, Ao over 7 years ago

I've came to the stage of requiring a lot of MAUS runs - getting help from Durga and Dimitrije on that but still is not progressing as fast as I expected. JB is here at Fermilab for 2 weeks and we could work out something together.

Since JB did Cooling Demo with MAUS, it should be not hard for JB to set up MAUS-2.5.0 for the analyses. In the meantime while MAUS is building and running, JB could you try to write a Python/C program to calculate the residuals of X, Y, PX, and PY if I provide you X,Y,PX,PY for the reconstructed track and MAUS-simulated track?

#7

Updated by Liu, Ao over 7 years ago

Discussed with JB today.

I'll work on the scripts to do
1. Use recon data 0ABCD_recon.root, extract the beam at TKU station 5;
2. Apply cuts to select tracks that have certain number of trackpoints in each tracker, save them to a G4BL ASCII .gbeam file;
3. Depending on the availability, use this .gbeam or convert it to a json file for tracking on local Linux clusters/IC cluster/GRID;
4. Also provide beam extracted from data 0ABCD_recon.root

JB is more familiar with ASCII files, so he will do,
1. Set up MAUS 2.5 on IC cluster and get it ready to run;
2. Write a script to extract trackpoints in MAUS output ROOT file, at the tracker planes;
3. Write a program to get the bin-by-bin residual (bin on x, y, Px, Py, for example) with the simulation data and measurement data.

Will update this issue tracker with progress.

#8

Updated by Liu, Ao over 7 years ago

  • File 08278_TK0andTK1_STN5.zip added

Data for run 08278.

Plots are for TKU and TKD station 5.

Cuts applied: Hit all 5 planes, Within Fiducial, Transverse momentum < 100 MeV/c;
No NAN

To-do: Use G4BL to track them from TKU to TKD and do comparison.

Any questions please let me know. Going to sleep now. I'll be monitoring emails when I wake up.

#9

Updated by Liu, Ao over 7 years ago

  • File 08280_TK0STN5_xPx.jpg added
  • File 08280_TK1STN5_xPx.jpg added
  • File 08280_TK1STN5_xPx_sim.jpg added

Did more analyses and removed the confusing plots I attached.

Plz find the plots in the ppt file here. It explains.

I should be able to do MAUS comparison tomorrow - we'll see.

#10

Updated by Liu, Ao over 7 years ago

  • File deleted (08280_TK1STN5_xPx_sim.jpg)
#11

Updated by Liu, Ao over 7 years ago

  • File deleted (08280_TK1STN5_xPx.jpg)
#12

Updated by Liu, Ao over 7 years ago

  • File deleted (08280_TK0STN5_xPx.jpg)
#13

Updated by Liu, Ao over 7 years ago

  • File deleted (08278_TK0andTK1_STN5.zip)
#14

Updated by Liu, Ao over 7 years ago

One thing is absolutely critical: pion contamination, or PID.

Who usually does PID on data? Can collaborate, or it would be a waste of expertise.

My thought is - obviously in a dream world -
Recon data -> A good person who filters muons out -> I do the analyses like above, and start working on the optics.

I need to find that person and say a lot of thankful words.

Ao

#15

Updated by Rogers, Chris over 7 years ago

Nb: Ao, before we get much further down the route of data taking, can you do a sift of the data and check that it looks satisfactory for analysis. E.g. is the transmission good enough? Is the tracker recon okay? Are there any things you are unhappy about wrt TOF performance? Is the data consistent with MC on a run-by-run basis? Can you make a table, for each run, listing the things you checked?

The point is that we want to carefully check every run for problems early, so we can catch them early.

#16

Updated by Rogers, Chris over 7 years ago

Note slide 3 hall probe readings - can you see a drift in the optics corresponding to this magnet settings?

http://micewww.pp.rl.ac.uk/attachments/7338/MICE_Soak_test_Sept_2016.pdf

#17

Updated by Liu, Ao over 7 years ago

Hi Chris,

Regarding your note #15:
1. I have an idea on how to check the transmission and tracker recon (tracker recon is a bit trickier to me but I'll consult experts if needed).
2. As for TOF performance, I have no experience with that. Should we find an experienced detector folk to check that?
3. As for MC comparison on a run-by-run basis. I'm on it.

Regarding your note #16:
From the time that the gap happened, I think the best setting to check is the 170 MeV/c runs 8277 and 8280, which are before and after the gap, respectively. I have the data for 8280 but not for 8277. Will work on that too.

Some information on how these stuff used to be done will be helpful and time-saving.

Best,
Ao

#18

Updated by Rogers, Chris over 7 years ago

It might be worth having a look at our ops meeting page from today, in particular the detector validation.

http://micewww.pp.rl.ac.uk/projects/operations/wiki/20160927

A couple of huge features to check:

1. Do the TOF plots look okay e.g. the long tail in 8280. I note there are rather a lot of electrons, more than I recall in other similar runs. Did something change? Or am I imagining it? The TOF calibration has not been updated yet, and some of the gains changed, so errors at ~ few hundred ps may be expected.
2. Here is a killer feature that must be a SSD misalignment I think - the beam is kicked 100 mm off axis in the EMR! Do you see that in the TOF plots? Nb: EMR was misaligned in 2015 but we fixed that misalignment in 2016 (I forgot when) so the detector should be physically aligned. Let's dig.

#19

Updated by Rogers, Chris about 7 years ago

  • Status changed from Open to Closed
  • % Done changed from 0 to 100

Also available in: Atom PDF