We measure plane efficiency for plane D (for example), when there is one and only one cluster in A/C and B planes in which the clusters in A and C are matched within 0.5 mm.  [ It's known that the planes E2U-A (index 8) and W2D-C (index 30) were missing a SVX and therefore they should get lower efficiencies as 1/4 of the planes couldn't register any signal. ]

If we count plane D as "efficient" whenever there is non-zero cluster on plane D, we obtain the following efficiencies on various planes:

Nevertheless, if count plane D as "efficient" only if one of the cluster on plane D matched with the only cluster on plane B (with the same above-mentioned criterion), we find the following efficiencies:

If we change the cluster matching criterion from 0.5 mm to 0.25 mm, one'd obtain the following result when we don't match cluster in B & D (in our example of narration):

If we require matching clusters in B and D to be < 0.25 mm (as well as between A/C), we see the following :

Energy (ADC) of matched clusters in the 4th plane when the other 3 planes have only 1 cluster for clusters with length ≤ 5 (from 7 runs 18174055-18176018) :

[ There was scintillator trigger cuts but NO silicon bunch cuts. ]

 
 
 
 
 

As requested by Wlodek Guryn,I did the above for two groups of runs which Wlodek called "hpos" (high luminosity) and "lpos" (low luminosity).  And since all these would have problems for the W1D-C plane, I also did the same for 14 runs before W1D-C plane had problem.  The result has been stored in the tar-gzipped file: xyeff_0.5mm.tar.gz .

The above were all done using the matching criterion of within 0.5 mm between the two clusters in two planes.

When I changed the criterion to be 0.25 mm, the result has been stored in this tar-gzipped file: xyeff_0.25mm.tar.gz .

The same efficiencies of roman-pot planes in the x-y coordinates as above, except this time we only consider clusters with length ≤ 5 with energies (ADC's) higher than thresholds for their given lengths:

[  The energy thresholds that I've used are (in the order of "E1U, E1D, E2U, E2D, W1U, W1D, W2U, W2D" and lengths of "1, 2, 3, 4&5" ) :
{  {20, 28, 50, 66}, {20, 28, 50, 60}, {20, 28, 50, 60}, {20, 28, 45, 60}, {20, 28, 50, 60}, {20, 26, 50, 60}, {20, 28, 50, 60}, {20, 28, 50, 60} } ;  ]

 
  
  
  
  
  
  
  
  
The above was done with the 0.5 mm matching criterion and I've stored the histograms in root files in xyeff_0.5mm_enlencuts.tar.gz .

XY coordinates (in m) on the roman pots of the tracks on two opposite arms (EU-WD or ED-WU) :

[ Each track is formed by two roman-pots on each side.  In each pot, if there are more than one pair of clusters in the A/C or B/D planes, I use the pair with the closest match in positions; if there is one cluster in A (B) but not C (D), or vice versa, I use that one existing cluster as the cluster position.  Events in the circumstance that there are more than 1 cluster on one plane but none on the other are rejected.

The angle (θ_X or θ_Y ) is formed by the two roman-pots in one arm is compared with that of the opposite arm, only when they agree within 3σ's, these two opposite tracks are accepted.  Moreover, a fit is also performed in the form of X_RP = X_IP + θ_X • (Z_RP - Z_IP), and similarly for θ_Y.   The fit has the effect of assuring that both tracks come from the IP (Interaction Point).   θ_X/θ_Y's obtained by these two methods are also compared and the track pair/event is accepted only when they agree (within 0.000125 radians). This procedure would eliminate tracks which don't come from the IP and often have very small fitted θ_X/θ_Y  ≈ 0 (as the fit fails). ]

Plots of -t (GeV/c)² in UA4 bins from 0.19 to 1.11 with the same cuts/criteria as above :

   
 

After "fiducial cut" attempt, XY coordinates (in m) on the roman pots of the tracks on two opposite arms (EU-WD or ED-WU) :

No. of planes that a roman-port point has within the final tracks selected for analysis :

For the first 14 runs which still had a working W1D-C plane :






Including all the 62 runs,  in which the latter runs didn't have a working W1D-C plane :





 

Two more significant changes:
--- The matching criterion of 0.5 mm has been changed to 0.3 mm. 
--- The method described in (point 9) above (by comparing  θ_X/θ_Y's obtained by two methods) is replaced by extrapolating the tracks in the East and West to z=0, then comparing their x₀/y₀ and finally choosing only those track pairs which are within 3 σ's are accepted.

After requiring the presence of the RP_ET (elastic trigger) triggerID in the events, we have the following statistics after each cut and the -t plots.

[ All numbers in the first plot below shown are numbers after a real cut in the analysis code except  "Trig & RP".  When I look for clusters in each RP, I require that the ADC/TAC of the trigger scintillators pass the online criteria.  I do so for all 8 RP's and in each event, if there aren't at least 2 RP's with clusters and passing trigger criteria, I mark that event as a "fail".  But no event is actually cut / got rid of here.  If there are not enough valid hits (no matching cluster and not satisfying trigger criteria), the event would not pass the "Opp Track" cut.  ]







 

Now if we place the RP_ET triggerID cut at the end of the analysis flow, we can see what triggers are there before the RP_ET triggerID cut and how many events the requirement of triggerID events have removed.





 
 

When ET should be fired (determined by using the TAC/ADC of the two roman-pot PMT's and the logics as described in this TOF-MTD-PP2PP_document), the plot below shows whether the triggerID of ET (=590709 except for earlier runs like 18174055 which has triggerID as "4") was present (1) or not (0).  This was done for all events, without any selection cuts.


 



The presence (1) or absence (0) of ET triggerID does not reflect whether an event should be an ET or not logically (or physically).   This turns out to be due to the "prescale" !!  Because of the "prescale", even an event should be an ET logically, the ET triggerID would not be present if it was "prescaled away".

But I have checked (and Akio Ogawa confirmed) that the DSM (e.g., the lastDSM(0) --- this is the 3rd bit in the TCU output as described in the top table of the last page of the above-mentioned TOF-MTD-PP2PP document ) seems always consistent with the determination of ET by using the TAC/ADC of the PMT's.
 
Now, at the end of my selection process, if I use the 3rd bit of lastDSM(0) --- ET bit --- as an ET requirement, this ET requirement would eliminate NO/0 event !  Because my selection codes earlier in the process have already made the same requirements on the TAC/ADC's of the two PMT's as well as the EU-WD or ED-WU opposite side coincidence.  Now, this is what I have expected. 

( The daq-file, i.e., raw data, has the same feature.  If one looks at the "daqbits" from Tonko's daqReader as I described here, the ET bit would be 0 if it's prescaled even logically this ET bit should be "1".  Again, the DSM/TCU information would tell you that that event should be an ET. )

No. of TOF Hits for events with a CPTnoBBCL triggerID vs those events without CPTnoBBCL triggerID :

When the comparison cut of the scattering angles (x/y) between East and West is reduced from 3 σ's to 2 σ's, below is the plot that shows what triggerId's exist in the final sample (after all selection cuts).






The ratio of the no. of events with "CPTnoBBCL-triggerId"  to the no. of events of "ET-triggerId" is about the same as the plot shown in point 14, at almost exactly 1.8.

 

Plot of AnalysisFlow after I use the proper ET trigger signal to make ET trigger cut [ 3^rd bit of lastDSM(0) as defined in the first table of the last page of the TOF-MTD-PP2PP document ]:


 

With the cuts mentioned in point 13 [using the 3^rd bit of lastDSM(0) as the real ET criterion], the -t distribution of all selected events (with all bins, not just the UA4 bins) is shown below:

 

 

Asymmetries and A_N:

_{RHIC proton polarization vector is normally vertical +y.  But during the RHICf period in 2017, the spin rotator rotated the polarization axis to the direction +x.   Here, we treat the +x (polarization axis) as the +y axis before,  and thus φ = φ + 90° [ if φ > 180°, φ = φ - 360° ].  After that, we can just use the square-root formula and fit for ε/(PB+PY) ~ AN cos(φ) .}  

           

Background Determination (2)

Previously, the x0 vertex plot was made after we have placed 3σ cuts on the θ_x and θ_y.   Now, we made 3σ cuts on vertices x0 and y0 first and then we look at the difference of  θ_x (say) between the West and the East.  I have performed a fit of the form of "Normalized-Gaussian + p₀ + p₁*x + p₂*x^₂ + p₃*x³" and the result is shown below:



Just like what we did previously, since the bin width is cancelled out in fraction calculation, we don't mention it below.  The total no. of events under the Normalized Gaussians (from  -3σ to 3σ) is S = 362.1*0.9973.   And the background under the polynomial (green) is the integral of the polynomial from -3σ to 3σ, which is B = [ p₀*3σ*2 + 1/3 * p₂ * (3σ)³ * 2 ].

So, the background fraction is B/(B+S) ~ 0.12%.