STAR QA Board

[Geary Eppley <eppley@rice.edu>]

Hi Gene:

Thanks for the detailed explanation. Then the Vx vs Vy distributions  
shown by Bulkcorr and HF last week were in fact not physical, but the  
result of a "software" update mid-analysis.

The central Vx vs Vy central distribution shown Friday has a radius of  
order 1mm. The L4 distribution has a radius of order 1 cm. Do you know  
what causes that difference?

We are interested in VPD-TPC Vz offset to see if any adjustments need  
to be made. L4 consistently shows and offset of -2 cm or less. The RMS  
is just slightly worse than Run 17, which is not bad considering the  
current problems with the active splitter.

What offset do you see in your analysis?

-Geary

Quoting "Van Buren, Gene" <gene@bnl.gov>:

> Hi, all
>
> I am attaching the plot that I assume is in question. I believe what  
> you are seeing is the result of the following:
>
> 1) The FastOffline reconstruction uses the "BeamLine3D" option  
> developed by Dmitri Smirnov several years ago, which allows the  
> BeamLine constraint to be used with finite weighting. The "non-3D"  
> "BeamLine" option, in contrast, gave the BeamLine infinite weight,  
> such that plots like this only ever show a point for a non-tilting  
> BeamLine definition, or a line for a tilting one. With the finite  
> weighting of BeamLine3D, the reconstructed vertices lie in a  
> reasonably tight cloud centered around the BeamLine...except...
>
> Another feature of the BeamLine3D option is that if the tracks point  
> to a vertex far enough away from the BeamLine, the BeamLine itself  
> can be excluded from the fit. The end result of this is that there  
> is usually an empty ring of no vertices just outside that tight  
> cloud, beyond which vertices again appear. The various points we see  
> randomly distributed out at several mm and beyond are this class of  
> vertices.
>
> 2) When I sent the note to starops about the reconstructed BeamLine,  
> I changed the BeamLine parameters in the database. Before that, the  
> Beamline was all 0's, which means the z axis. After that, I settled  
> (after a few iterations of trying to exclude pile-up vertices) on  
> the following parameters:
>
>    x0  = -0.053 cm
>    dx/dz  = -0.00113
>    y0  = -0.215 cm
>    dy/dz  = 0.00023
>
> (these are slightly different from my starops post as I decided to  
> try one more iteration after that post).
>
> Slide 2 of Yingjie's slides notes that the data presented is: "Run  
> ID : 22349011– 22357028 (day 349-357, 167 runs)"
>
> ...which means that it includes data both before and after the  
> database BeamLine parameters were changed by at the end of the day  
> on Dec. 22 The effect of this is clear in the below plot: much of  
> the data had one non-tilting BeamLine near (0,0), and the rest of  
> the data passes through
> (x0,y0) = (-0.053,-0.215)
> with the tilt in this plot visible as
> (dy/dz) / (dx/dz) = (dy/dx) = 0.00023 / -0.00113 = -0.2035
>
> I assume that if Yingjie plots this data vs. run number, it will be  
> more clear how _when_ the data was produced (which is highly, but  
> not perfectly, correlated with when the data was taken) defines  
> whether the vertices are generally in the upper or lower bunch.
>
> Hope that helps,
> -Gene
>
>
>
>
> [cid:4D5E5865-7E0B-48CC-B5BE-84E5127DCCAA@attlocal.net]
>
> On Dec 26, 2021, at 6:39 PM, Ting Lin  
> <tinglin@sdu.edu.cn<mailto:tinglin@sdu.edu.cn>> wrote:
>
> Hi Geary:
>
> I think Gene can comment on this. What they use are the Min-bias  
> events from the fast production files (picoDst or MuDst files)  
> generated by the production group.
>
> There is Vx vs. Vy plot in fast offline, but do not get report if  
> that is problematic. For the fast offline, Gene said that current  
> setup require that there should be a primary vertex exist, then it  
> is working.
>
> As it is holiday, there is no expert present during the meeting, we  
> will continue to check using this week's data and discuss in next QA  
> board meeting (that will be Dec. 31th, Friday at noon BNL Time).
>
> Thanks,
> Ting
>
>
> -----Original Messages-----
> From: "Geary Eppley via STAR-QAboard-l"  
> <star-qaboard-l@lists.bnl.gov<mailto:star-qaboard-l@lists.bnl.gov>>
> Sent Time: 2021-12-27 07:04:27 (Monday)
> To: STAR-QAboard-l  
> <star-qaboard-l@lists.bnl.gov<mailto:star-qaboard-l@lists.bnl.gov>>
> Cc:
> Subject: Re: [STAR-QAboard] 2. Vz – vpdVz distribution is shifted by  
> about 5cm.
>
> HF & Bulkcorr QA reps:
>
> The vertex Vx vs Vy plot doesn't look physical. Can the beam really do
> that? Do you understand why your plot looks so different than L4 for
> that period? Your central distribution is an order of magnitude more
> compact than the L4 plot.
>
> Is your result consistent with the mean vertex position Gene sent out?
> What plot and analysis chain does Gene use? What analysis chain do you
> use? Is there a Vx vs Vy plot in Fast offline? If there a Vz VPD-TPC
> plot in fast offline?
>
> Could you plot Vz VPD-TPC for just the central Vx vs Vy distribution?
>
> -Geary
>
>
> Re: Re: [STAR-QAboard] QA Board Meeting December 24
>
> Date 12/24/2021 (10:51:06 PM CST)
> From Ting Lin <tinglin@sdu.edu.cn<mailto:tinglin@sdu.edu.cn>>
> To geary eppley <eppley@rice.edu<mailto:eppley@rice.edu>>
> text/plain Text (3 KB)
> Hi Geary:
>
> We found this issue from offline QA by HF and Bulkcorr group.
>
> You can see the slide 3 from HF group by Yingzhou and Kaifeng below:
> https://drupal.star.bnl.gov/STAR/system/files/1225_HF_QA_pp500_0.pdf
>
> Also slide 3 from Bulkcorr group by Ashish:
> https://drupal.star.bnl.gov/STAR/system/files/Bulkcorr_pp_500GeV_QA_week1_new.pdf
>
> Best,
> Ting
>
>
>
> Ting Lin tinglin at sdu.edu.cn
> Fri Dec 24 13:17:40 EST 2021
> Previous message: [STAR-QAboard] QA Board Meeting December 24
> Next message: [STAR-QAboard] QA Board Meeting December 24
> Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
> Hi all:
>
> A brief minutes for today's meeting:
> https://drupal.star.bnl.gov/STAR/comp/qa/pwg-qa/run-22-qa
>
> For Run-22 offline QA:
>
> From Lanny Ray:
> 1. Offline QA software get stuck this morning.
> 2. Run 22355023 – 22357011 have excessive hits in various RDOs. Tonko
> power cycled the TPX FEE’s and that seems to have fixed the issue.
> This will very likely to affect the track quality. Need to check with
> TPC expert if we want to mark these as bad or not.
> 3. Mean dE/dx has been slowly decreasing since the first run
> e.g.22349015 with MIP at 2.4 keV/cm, which continued on days 351 &
> 352, but then reduced to more normal values (2.2 keV/cm) on day 353 to
> the present. Also need to check with expert about the real reason if
> it happen again.
>
> From pwg:
> 1. Most of the problematic runs have been identified by offline QA
> shift, also noted in the shift-log.
> 2. Vz – vpdVz distribution is shifted by about 5cm.
> 3. Vr shifted during the last few runs in day 356.
>
>       BulkCorr (Ashish)
>       LF (Chenliang)
>       HF (Yingjie, Kaifeng)
>       JetCorr (Tong)
>       ColdQCD (Bassam): .
>
> Best,
> Ting
>
>
> _______________________________________________
> STAR-QAboard-l mailing list
> STAR-QAboard-l@lists.bnl.gov
> https://lists.bnl.gov/mailman/listinfo/star-qaboard-l