upgr15 test production
11/30 :
I made a test sample of Au+Au Hijing events with upgr 15 with :
- 100 events
- |zvertex| < 5 cm
- sigma(zvertex) = 5 cm
- fast simulator for ssd
- hit smearing for PIXEL and IST as well (the values are in the following pdf)
The files are located at :
/star/institutions/emn/bouchet/code/hijing_10Ev
The log of the ouput is in the same directory (0F15DBF20A9DB9D9AEFEBCFEF673B95F_0.out : ~ 15 MB)
Some plots for this sample are here.
12/01 :
another sample with the same vertex properties but with 100 events has been done.
The root files are at :
/star/institutions/emn/bouchet/code/hijing_100Ev
The log of the job is also in the same directory (B414010D4F7C0E6C15AD63F0003F4A20_0.out)
--> to do : plots
12/02 :
some plots for the check of the smearing are here
12/06 : last change
Few changes were applied to he code to :
- run without a local /StarDb/VmcGeometry/ : upgr15.h was updated in CVS
- code changes in StPixelFastSimMaker.cxx and StiPixelDetectorBuilder.cxx to reflect the change of upgr15.h
- code changes in StIstDetectorBuilder.cxx to account dead materials
- a bug found that did not allowed previously the tracker to accept hits below a given radius :
in /Sti/StiKalmanTrackFinder.cxx :
--- StRoot/Sti/StiKalmanTrackFinder.cxx 2008/06/10 00:58:40 2.101
+++ StRoot/Sti/StiKalmanTrackFinder.cxx 2008/12/05 23:22:59 2.102
@@ -791,7 +791,7 @@ void StiKalmanTrackFinder::nodeQA(StiKal
double xg = node->x_g();
double yg = node->y_g();
- if ((xg*xg + yg*yg) < 4.2*4.2) qa.qa= -2;
+//VP?? if ((xg*xg + yg*yg) < 4.2*4.2) qa.qa= -2;
}
//______________________________________________________________________________
An important change is that we have now to run in DEV (stardev).
So, to test the code I ran :
- the new code over 1 Hijing event : code in DEV + no local /StarDb/VmcGeometry (1)
- the old code over the same Hijijng event : code in SL08e + dependancy to /StarDb/VmcGeometry (2)
- the new code over the same Hijing event with pileup (3)
for (1)
the output (log +root files are at :
/star/institutions/emn/bouchet/latest_code/1Ev/
Form the log :
QA :INFO - # TPC hits: 111072: deconvoluted: 0: Used in Fit: 74234
QA :INFO - # SSD hits: 2016: Bad ones(flag>3): 0: Used in Fit: 1361
QA :INFO - # Hft hits: 4685: Bad ones: 0: Used in Fit: 2578
QA :INFO - # Ist hits: 2245: Bad ones: 0: Used in Fit: 1503
for (2) :
log+root files are at :
/star/institutions/emn/bouchet/code_final_mirror
from the ouput :
QA :INFO - # TPC hits: 110969: deconvoluted: 0: Used in Fit: 73720
QA :INFO - # SSD hits: 2016: Bad ones(flag>3): 0: Used in Fit: 1231
QA :INFO - # Hft hits: 4685: Bad ones: 0: Used in Fit: 2440
QA :INFO - # Ist hits: 2245: Bad ones: 0: Used in Fit: 1462
QA :INFO - # UnKnown hits: 7286: Bad ones: 0: Used in Fit: 0
So we see that we have the same number of hits (first colum).
The number in the second are the hits used in tracking. As I ran here in SL08e, whereas the previous log was done in DEV, it may explain the difference.
There are also the dead material in IST that could have an effect.
for (3):
the output (log +root files are at :
/star/institutions/emn/bouchet/latest_code/1Ev/
Form the log :
QA :INFO - # TPC hits: 111072: deconvoluted: 0: Used in Fit: 74292
QA :INFO - # SSD hits: 2010: Bad ones(flag>3): 0: Used in Fit: 1382
QA :INFO - # Hft hits: 13919: Bad ones: 0: Used in Fit: 2740
QA :INFO - # Ist hits: 2245: Bad ones: 0: Used in Fit: 1523
So here we see the number of HFT hits increased by a factor of 2.97, due to pileup
Finally I ran over 40 events with the new code :
root files+ log are at :
/star/institutions/emn/bouchet/latest_code/1-40/
The following plot show :
- first row (pixel) : lYHit-lYFit (left) ; lZHit-lZFit (right)
- second row (ist) : lYHit-lYFit (left) ; lZHit-lZFit (right)
The distribution looks ok, what is a bit odd is the values of the rms : it is lower than the older code.
For pixel, rms ~ 2-3 mu whereas it was rms ~ 6-7 previously
for ist, it was rms ~ 0.023 (0.17 in Z) , now it's 0.015 (0.18)
Maybe changes in Sti from SL08e to DEV can explain those differences.
I also ran 40 events (the older file was done with 100 events)
Except of this, the code is running and ready for small production.
The how-to is here
- bouchet's blog
- Login or register to post comments