QA for the d-Au 2008 St Upsilon files

UPDATED on Aug 8, All upadates marked in BLUE

Global Run/Event Characteristics

SL08C dAu Events total found using the getfilelist script with the conditions

production=P08ic,filename~st_upsilon, trgsetupname=production_dAu2008, filetype=daq_reco_MuDst,storage=nfs

This returns 1,210,691 events

Next the events are made into a tree and tracks with a corresponding tower are selected

The triggers and tracks per event were tested.

Events without the trigger Id 210601: 42578

Note: These numbers match those seen by Haidong

Events with no 210601 trig Id removed from sample, (these have trig Id. = 3)

The run numbers of these events were recorded, they are the following:

After track-matching to the BEMC, events with no tracks: 77994

With these events removed this left the following total number of good events: 1090119.

This give a total set of good runs: 1508

Number of interaction vertices and ranking

The following plot show the number of interaction vertices (up to 15) in an event. For d-Au it has an average of 8 vertices per event, giving about 8M "sub-events".

The ranking is shown here, for all vertices

Now the ranking values for vertex[0] (the first vertex found)

Vz for different ranking values


If rank =0 and = 1 for a vertex it will be saved. This leaves us with 1.06 Million events

UPDATE Vertex and ranking

It the dAu vertex was constructed using Minuit not the PPV finder, Marco sent a link to more information on this,

Vertex information

We found that there is a improvment in the quality of the VZ distribution if we use the first primary vertex in the array and select on the ranking between the following: 0 ≤ Ranking < 2.

Although it does not remove the wings, it can be seen that the rankings 0 to 1 converge with this selection giving cleaner Vz

Vertex[0], 0 ≤ Ranking < 2

All primary vertices

First Vertex in array (vertex[0])

      Select First primay vertex in array: vertex index = 0
      Select ranking between the following: 0 ≤ Ranking[0] (vertex index == 0) < 2

Multiplicity and VZ plots

Vertex Z distribution:

Vertex Z distribution after ranking cut (rank==0 or rank==1)

VZ Wings vs run

Run number have been separated for clarity

Y7=Run-number within 8M-9M Y8:Run-number > 9M

A small investigation was done to see if these "wings" in Vz were run dependent

The multiplicity varied but it seems that the wings happen at all stages though the run. There is one case of very wide Vz in a run around 9014/9015

VZ vs. run Y7

VZ vs. run Y8

Mult vs run

Ref_Mult averaged over vs. run Y7

Ref_Mult averaged over vs. run Y8

Vertex Z Y7 average vs. run

Vertex Z Y8 average vs. run

Pt Z Y7 average vs. run

Z Y8 average vs. run

Conclusion and future work here

Most of these run-by-run averages look reasonable. There are several runs that have outliers, for example, low multiplicity runs, some large vertex shifts and shifted Pt. The next step would be to pick the outliers and check the log books and see if there were beam/detector issues, maybe these should be removed to improve the quality of the dataset

Multiplicity of positive vs. negative tracks

Within an event the number of positive and negative tracks were counted. This plot looks stable

L2 trigger online monitoring (These are thumbnails so please click to enlarge)

This is a summary of the L2ResultArray information. They all look OK apart from spikes in the L2 Z position, I checked my binning, it seems to be OK. Another point was the double peak in the L2 time is this as expected, I nievly thought that any timings would look smooth on average. Finally, L2 hot towers peaked at 25 and is by far the largest peak, it looks like it has hit a maximum value limit.

L2 Cos

L0 eng.


L2 eng.



L2 z vertex

L2 Triggers

If we match the L0 triggered tower from the L2ResultsArray with the tower in an event and bit-shift the ADC >>4 this condition should give us back the L0 trigger distribution from the data.

We can also match the L2 tower from the L2ResultsArray with the tower.

This time we use (ADC-OnlinePedestal)>>4 this condition should give us back the L2 trigger distribution from the data.

ADC >> 4 L0tower==tower

Energy with L0tower==tower

(ADC - OnlinePed)>4 L2tower==tower

Energy with L2tower==tower

Electron ID plots

Electron ID

We used the following for this investigation







dEdx tight






P/E tight



With L0 and L2 trigger conditions

With a mimic trigger L0==(ADC)>>4 >18 and L2==(ADC-OnlinePed)>>4 ≥ 5

the dEdx distributions are

dEdx with no selections

dEdx with mimic L0 trigger condition

dEdx with mimic L2 trigger condition

dE/dx with tight P/E cut

To really cut into the electron rich region, a tight P/E cut was applied and dE/dx plotted

dEdx, tight P/E with L0 mimic condition

dEdx, tight P/E with L2 mimic condition


Now onto P/E distributions


P/E with (p > 1.5)

P/E with mimic trigger condition

P/E with L0 mimic condition

P/E with L2 mimic condition

P/E with loose dEdx cut and mimic L0 condition

Remaining L0 electron candidates for loose cuts

26941 tracks

Remaining L2 electron candidates for loose cuts

62430 tracks

Electron Candidate Pairing

++ vs. --

First we looked at ++ and -- signals



Is there something strange for ++ pairs around 11 GeV ?

Signal vs. background

With the dEdx cut 3.3e-6 to 4.7e-6

With P/E less then 2 (see above table)

Momentum of paired tracks > 1 GeV

Opening angle cos(theta) < 0.5 and Mass cut > 2 GeV/c

BG Pairs(++/--) scaled to 2sqrt(++*--)

Pair satisfied if it has mimic L0 condition applied to one track and mimic L2 condition to the other

Signal (black) Background (red)

Signal Minus Background

Fit to peak





UPDATE: Mass peak

The multiple ranks of same value (see above) were causing my mixing code to smear the peak

This is corrected with the following with a sharper peak with ranking and "sub event" vertexId set as:

  1. First vertex only (vertex[0])
  2. "Sub Events" with Rank 0 or Rank 1

For a crude fit I tried a Gaussian between 8.5 and 10.5 GeV and a Gaussian + Straight Line fit (gauss+[3]) between 4 and 14 GeV

Gauss only 250 MeV bin

Gauss + Poly x 250 MeV bin

Gauss only 500 MeV bin

Gauss + Poly x 500 MeV bin

Gauss only 1 GeV bin

Gauss + Ploy x 1GeV MeV bin

This is a table of the fit values

Gauss 250 MeV bin
1 Constant19.803.72
2 Mean9.480.08
3 Sigma0.540.11
Gauss +Pol 250 MeV
1 Constant14.611.73
2 Mean9.350.13
3 Sigma1.040.10
4 Poly-0.500.44
Gauss 500 MeV bin
1 Constant39.736.61
2 Mean9.520.09
3 Sigma0.590.11
Gauss +Pol 500MeV bin
1 Constant30.390.12
2 Mean9.380.12
3 Sigma1.080.51
4 Poly-1.120.78
Gauss 1GeV bin
1 Constant73.0610.62
2 Mean9.490.10
3 Sigma0.720.092
Guass + Pol 1GeV
1 Constant59.959.2
2 Mean9.410.13
3 Sigma1.030.15
4 Poly-1.120.82