[Rgc_analysis] [EXTERNAL] Second Pass Results for Raster Calibration

Mon Jul 3 17:15:35 EDT 2023

Hi Sebastian,

Thank you for your feedback!

I presented these slides at the software meeting last Thursday, and Rafaella, Veronique, Nathan, and others gave me some helpful information.

You're right, the method that I used to get the data was incorrect. Others at the software meeting suggested that I do a profile fit to the data. What that entails is to choose a fixed ADC signal value and make bins in the vertex positions for that fixed ADC value. This should yield a gaussian distribution in the data. By fitting the histogram to a gaussian, one can take the mean of the fit as the vertex position and the width of the gaussian as the error bars.

I attached some preliminary plots and histograms to this email. For the particle vertices, I averaged the electron and pion vertices like I described in the slides I had sent previously. However, I made no cut to the x,y,z vertex positions on these plots. I do have other plots where I did repeat those cuts, but I'm still in the process of putting everything together coherently.

Using a fit of the form [cid:image005.png at 01D9ADD1.F6A4B5E0] , I got the following for the x and y parameters:

My Fit Values for X-vertices: [cid:image006.png at 01D9ADD1.F6A4B5E0]
CCDB  Values for X-vertices: [cid:image007.png at 01D9ADD1.F6A4B5E0]
My Fit Values for Y-vertices: [cid:image008.png at 01D9ADD1.F6A4B5E0]
CCDB  Values for Y-vertices: [cid:image014.png at 01D9ADD1.F6A4B5E0]
These results are still very preliminary and I wouldn't stake too much in them, since some other cuts I was messing around with changed these significantly.

In the "e_pi_avg_vx_adcx" graphs, I fit the results of the procedure I described above. While the error bars are still very large, the fit to the data still looks much better. The "histos" pdfs show all the histograms for each of the ADC signal bins that I fit.

I hope that this is the procedure that Veronique and the others had described to me last week. I should have some slides put together that discusses everything in greater detail by the next software meeting this Thursday.

Thank you all again for your help! I very much appreciate your feedback.

Sincerely,
Derek Holmberg

From: Sebastian Kuhn <kuhn at jlab.org>
Sent: Monday, July 3, 2023 3:48 PM
To: Holmberg, Derek <deholmberg at wm.edu>
Cc: Silvia Niccolai <silvia at jlab.org>; Griffioen, Keith <griff at wm.edu>; Carlos Ayerbe Gayoso <gayoso at jlab.org>; harut avagyan via Rgc_analysis <rgc_analysis at jlab.org>; Raffaella De Vita <devita at jlab.org>; Veronique Ziegler <ziegler at jlab.org>
Subject: Re: [EXTERNAL] Second Pass Results for Raster Calibration

Hi Derek,

your findings are quite interesting - sorry for no feedback earlier. I feel we can say 2 things:
1) Your fit is probably still biased by the outliers (the dots that fill the rectangle and don't seem to follow the slope). One way to fix that may be to use your fit result and make a further selection: Only included data points where the vertex x or y is within +/- 0.7 cm of the PREDICTED vertex from your fit vs. ADC. This won't cut too many data points, but maybe reduces the bias towards "flatness" of the fit. You could even iterate this: Make a new fit, and then use THAT to select the data points, etc.
2) There seems to be something not quite right yet with the way the vertex of closest approach is being calculated. I do not understand how the mean deviation of the mid-point between the 2 vertices from the fit can be an unchanging sigma of 0.38 cm no matter how tightly you control the DIFFERENCE between the 2 vertices. In other words, when you require the pi+ and e- to reconstruct to within 0.2 cm from each other in the perpendicular plane at their closest approach (which is the meaning of S_V < 0.2 cm), why you don't reconstruct the correct position (average vertex) to at least a LITTLE BIT better precision than if you don't have any cut on S_V at all.

Be that as it may, please keep in mind that those 0.38 cm are NOT the intrinsic uncertainty of the FIT constant - once we avoid bias (see 1 above), those will be determined to much better (in principle by a factor 1/√N) due to the high statistics of the sample. So, I am not at all ready yet to give up on this method! ;-)

Here is another thing you could try: Instead of cutting on Vz for each particle individually, you could also cut on the z-value of the point of closest approach. (Presumably, the POCA is a given by a z-value along the beam line where the 2 particle trajectories, traced back to that point, are closest to each other). That way you are really picking the events that we are most interested in = the ones which have a 3-dim. vertex inside the target.

Finally, for diagnostic purposes, it might be useful to make histograms, e.g., of this quantity (z-value of the POCA), as well as the difference between predicted vertex x and y (from the linear fit) vs. the inferred vertex from the POCA.

Thanks for working on all of this. I am not sure when there will be another RG-C analysis meeting (tomorrow is 7/4 and next week is CLAS Collaboration Meeting), but maybe we can talk during the latter.

Greetings - Sebastian

On Jun 26, 2023, at 11:19 AM, Holmberg, Derek <deholmberg at wm.edu<mailto:deholmberg at wm.edu>> wrote:

Hi Everyone,

I hope you all had a great weekend.

I attached the results from the updated analysis I had done, taking into account the additional cuts you all had recommended. As a quick summary, here are the additional cuts I had made:

  1.  A cut on
[cid:image001.png at 01D9ADCF.91615190]

  1.
  2.  A cut of all data points separated by more than 1.25 cm from the center of the target
  3.  Various cuts on the "distance of separation" (which I just labeled as
[cid:image002.png at 01D9ADCF.91615190]

  1.  ) between the reconstructed particle vertices

I also found the residual standard deviation for each of the fits. However, while changing the cuts on the distance of separation seemed to affect the calibration constants by 5-10%, the residual standard deviation for each one was practically the same at 0.38cm. Given that the 2-sigma for the fits is around 0.75cm, would this detached vtx method be precise enough to be useful? My results certainly aren't gospel, and I'd love to hear your feedback.

When I performed the fits, I didn't include error bars on the individual vertex positions and just fit them as-is. Is that the correct approach? I just wasn't sure if we were going to have the analysis meeting this week because of the JLUO meeting.

I uploaded my slides to page of last week's meeting:https://clasweb.jlab.org/wiki/index.php/RGC_Jun20

Sincerely,
Derek Holmberg
[cid:image003.png at 01D9ADCF.91615190]
[cid:image001.png at 01D9ADCF.91615190]
[cid:image004.png at 01D9ADCF.91615190]
[cid:image002.png at 01D9ADCF.91615190]

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.jlab.org/pipermail/rgc_analysis/attachments/20230703/f7c70564/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image001.png
Type: image/png
Size: 889 bytes
Desc: image001.png
URL: <https://mailman.jlab.org/pipermail/rgc_analysis/attachments/20230703/f7c70564/attachment-0009.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image002.png
Type: image/png
Size: 999 bytes
Desc: image002.png
URL: <https://mailman.jlab.org/pipermail/rgc_analysis/attachments/20230703/f7c70564/attachment-0010.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image003.png
Type: image/png
Size: 1046 bytes
Desc: image003.png
URL: <https://mailman.jlab.org/pipermail/rgc_analysis/attachments/20230703/f7c70564/attachment-0011.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image004.png
Type: image/png
Size: 568 bytes
Desc: image004.png
URL: <https://mailman.jlab.org/pipermail/rgc_analysis/attachments/20230703/f7c70564/attachment-0012.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image005.png
Type: image/png
Size: 1230 bytes
Desc: image005.png
URL: <https://mailman.jlab.org/pipermail/rgc_analysis/attachments/20230703/f7c70564/attachment-0013.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image006.png
Type: image/png
Size: 5507 bytes
Desc: image006.png
URL: <https://mailman.jlab.org/pipermail/rgc_analysis/attachments/20230703/f7c70564/attachment-0014.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image007.png
Type: image/png
Size: 3994 bytes
Desc: image007.png
URL: <https://mailman.jlab.org/pipermail/rgc_analysis/attachments/20230703/f7c70564/attachment-0015.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image008.png
Type: image/png
Size: 5659 bytes
Desc: image008.png
URL: <https://mailman.jlab.org/pipermail/rgc_analysis/attachments/20230703/f7c70564/attachment-0016.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image014.png
Type: image/png
Size: 3819 bytes
Desc: image014.png
URL: <https://mailman.jlab.org/pipermail/rgc_analysis/attachments/20230703/f7c70564/attachment-0017.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: e_pi_avg_vx_adcx.pdf
Type: application/pdf
Size: 15308 bytes
Desc: e_pi_avg_vx_adcx.pdf
URL: <https://mailman.jlab.org/pipermail/rgc_analysis/attachments/20230703/f7c70564/attachment-0004.pdf>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: e_pi_avg_vy_adcy.pdf
Type: application/pdf
Size: 15306 bytes
Desc: e_pi_avg_vy_adcy.pdf
URL: <https://mailman.jlab.org/pipermail/rgc_analysis/attachments/20230703/f7c70564/attachment-0005.pdf>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: e_pi_adcx_histos.pdf
Type: application/pdf
Size: 25916 bytes
Desc: e_pi_adcx_histos.pdf
URL: <https://mailman.jlab.org/pipermail/rgc_analysis/attachments/20230703/f7c70564/attachment-0006.pdf>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: e_pi_adcy_histos.pdf
Type: application/pdf
Size: 26364 bytes
Desc: e_pi_adcy_histos.pdf
URL: <https://mailman.jlab.org/pipermail/rgc_analysis/attachments/20230703/f7c70564/attachment-0007.pdf>