LVL1 trigger bias in SEP05 - ArKCl at 1.75 A GeV

This page tries to show the effect of the LVL1 trigger in the SEP05 experiment on selection of the reaction events. Bias of LVL2 trigger is not discussed here.

In the production run, the main LVL1 trigger (96% of all events written to tape) was START && !(VETO) && MULT_TOFTOFINO>=16. Appart of that, we stored downscaled events with the trigger START && !(VETO) && MULT_TOFTOFINO>=4 (~4% of all events) and START (~2% of all events). These LVL1 events are called later as M16, M4 and ST.
Comment: The sum is not 100% because some events fullfill 2 or more trigger conditions (if downscaling was 1 for all triggers, there would be always trigger M4 in case of M16). The correct condition how to select events with a needed trigger bit is if(EventHeader.fTBit&N)==N) where N is desired trigger bit (16 for M16, 1 for M4 etc.)

Additionally, we had a short run (4 hours, 13-OCT-2005 day286 18:00 - 22:15) with a so called "minimum bias" trigger. The condition on no veto signal was not applied. The main LVL1 triggers were START && MULT_TOFTOFINO>=4 and START && MULT_TOFTOFINO>=16. These LVL1 events are later called M4_mb and M16_mb. We can apply in analysis condition NO VETO, these events are called M4_mb_nV and M16mb_nV.

Table 1: List of triggers:

production run		trigbit		downscaling		trig.rate/1 START signal*
ST		2		4096		1.00
M4 && START && !(VETO)		1		32		0.017
M16 && START && !(VETO)		16		1		0.010

"minimum bias" run		trigbit		downscaling		trig.rate/1 START signal*
M4 && START		1		2		0.031
M16 && START		16		1		0.014

* taken from on-line scaler data from 1 hld file, not very precise due to time fluctuations

Now we show figure with (RK) track multiplicities from various triggers, normalized per one START event. Note that 97.3% of START events has no tracks.

• Track multiplicity distribution for various triggers in SEP05:
  

In attachment you can find clones of this figure in logy and zoomed in y axis (fig1_*).

We can see following in the figure above (and attached):
1) M4 trigger differs considerably from M4_min_bias, so NO VETO condition is important (it is seen already in trigger rates).
2) M4 trigger differs considerably from M4_min_bias with NO VETO condition applied by software. Somehow we cannot reproduce the NO VETO condition from the data.
3) For M16 and M16_min_bias trigger the distributions look similar, but M16 is scaled down by 15%.
4) We can see an "empty target" events admixture, i.e. contribution of reactions outside of target. Main affect is in M4_min_bias distribution: 23% of events has no tracks. Also M16 events have enhanced number of events with no tracks. For production run this admixture can be subtracted, because we have a dedicate run with no target, see Filip's entry in forum empty target.

Now let's compare data to UrQMD simulation made by Filip. There are UrQMD events processed through Geant and hydra, with emulation of M4 and M16 LVL1 trigger, and without any LVL1 trigger condition. The last data sample should correspond to true minimum bias. Note that there is no VETO detector in simulations, so it corresponds to data from the run without VETO condition ("minimum bias").

• Track multiplicity distribution for various triggers in SEP05 - data vs UrQMD:
  

In attachment you can find clones of this figure in logy (fig2_logy).

The normalization of UrQMD is done in following way: for START trigger (no trigger selection in UrQMD) we normalize UrQMD distribution to data in all bins except bin 1 (no tracks). For M4 and M16 triggers we use the reduction factors: ratio of UrQMD events passed through the LVL1 condition, which is 74% for M4 and 35% for M16.
UrQMD describes data surprisingly well. For M16 trigger it seems to describe well both "min. bias" and production run, the NO VETO condition is apparently not important.

Following figure and table summarizes the results:
Figure shows (again) track multiplicity distribution for data and UrQMD for START (or min. bias in UrQMD case), M4 and M16 triggers.

• Track multiplicity distribution for various triggers in SEP05 - data vs UrQMD.:
  

In attachment you can find clones of this figure in logy and zoom in y axis (fig3_*).

Table 2: Rates of LVL1 triggers per beam particle (==START trigger):

trigger		EXP		EXP-empty*		UrQMD
ST		==1.00
ST&&1track		0.0267				==0.0267
min bias						==0.0310
M4mb		0.0287		0.0231?		0.0230
M4mb_nV		0.0201
M4		0.0178		0.0162
M16mb		0.0127		0.0114?		0.0107
M16mb_nV		0.0105
M16		0.0108		0.0101

* rate with subtracted admixture of "empty target events". For production run the measured "empty target" rates were used, for "minimum bias" run the admixture was estimated from the shape of track multiplicity distribution.

Based on observed good agreement of data and UrQMD, we can use the UrQMD to estimate the centrality selection of the LVL1 trigger, as it was done by Filip in centrality


For completeness, we checked the multiplicity of charged pi mesons in the data samples taken with various triggers:

This table is outdated, see new Table 1 in LVL1TriggerBiasInSEP05-piMultiplicities

Table 3: pi+/pi- multiplicities per event, in HADES acceptance, efficiency corrected, empty target subtracted.

trigger		EXP		UrQMD
ST		0.031/ 0.038
min bias		1.00 / 1.22*		1.18 / 1.48
M4mb		1.51 / 1.82		1.53 / 1.92
M4		1.73 / 2.09
M16mb		2.17 / 2.63		2.11 / 2.60
M16		2.14 / 2.70

* for EXP min. bias was taken as multiplicity of START events multiplied by 0.0310, which is estimated min. bias reaction rate, see Table 2.

The pi meson multiplicities for M4 and M16 triggers are reproduced well in UrQMD, for minimum bias we observe 20% difference. It is not clear whether it is real difference between pi multiplicities, and/or mistake in normalization == estimate of real minimum bias both in experiment and simulation.

Following figure shows distribution of polar angle of protons from reactions with various triggers. All distributions are efficiency corrected and normalized to 1 (The efficiency correction is same for all distributions). The effect of the M16 trigger is visible only under 22 deg.

• distribution of polar angle of protons from reactions with various triggers:
  

• the same, not corrected for efficiency:
  

-- PavelTlusty - 18 Dec 2007