slide : here

file : all3.root (mix of day 116,118 and 119)
The split btw positive and negative contribution in LdL is at L/dL =10

 

 

1. when the cut is L/dL > 10

 bin = 21 bin center = 0.482 curr sig = 73 curr back =31.0555 sum sig = 73 sum back = 31.0555
 bin = 22 bin center = 0.486 curr sig = 173 curr back =31.2972 sum sig = 246 sum back = 62.3527
 bin = 23 bin center = 0.49 curr sig = 501 curr back =31.5389 sum sig = 747 sum back = 93.8916
 bin = 24 bin center = 0.494 curr sig = 958 curr back =31.7806 sum sig = 1705 sum back = 125.672
 bin = 25 bin center = 0.498 curr sig = 989 curr back =32.0223 sum sig = 2694 sum back = 157.694
 bin = 26 bin center = 0.502 curr sig = 673 curr back =32.264 sum sig = 3367 sum back = 189.958
 bin = 27 bin center = 0.506 curr sig = 242 curr back =32.5057 sum sig = 3609 sum back = 222.464
 bin = 28 bin center = 0.51 curr sig = 92 curr back =32.7473 sum sig = 3701 sum back = 255.211
 signal = 3701 back = 255.211 s/b =14.5017 significance = 56.6408
 # bin left = 4 backleft =111.171
 # bin right = 4 backright =142.107
 check from histo : left = 88 right = 130

       2.  when the cut is L/dL > 8
bin = 21 bin center = 0.482 curr sig = 122 curr back =81.7407 sum sig = 122 sum back = 81.7407
 bin = 22 bin center = 0.486 curr sig = 235 curr back =82.3248 sum sig = 357 sum back = 164.065
 bin = 23 bin center = 0.49 curr sig = 586 curr back =82.909 sum sig = 943 sum back = 246.974
 bin = 24 bin center = 0.494 curr sig = 1075 curr back =83.4931 sum sig = 2018 sum back = 330.468
 bin = 25 bin center = 0.498 curr sig = 1108 curr back =84.0773 sum sig = 3126 sum back = 414.545
 bin = 26 bin center = 0.502 curr sig = 777 curr back =84.6614 sum sig = 3903 sum back = 499.206
 bin = 27 bin center = 0.506 curr sig = 316 curr back =85.2456 sum sig = 4219 sum back = 584.452
 bin = 28 bin center = 0.51 curr sig = 150 curr back =85.8297 sum sig = 4369 sum back = 670.281
 signal = 4369 back = 670.281 s/b =6.51816 significance = 55.9578
 # bin left = 4 backleft =295.418
 # bin right = 4 backright =370.19
 check from histo : left = 230 right = 367

--> significance is ~ the same but the S/B ratio is divided by 2

• 06/19 : study as a function of probability

file used : updated.root (mix of 116,118,119,120)

1. L/dL distribution : cuts are |n#sigma_{kaon}|<3 && |n#sigma{pion}|<3

results of the fit :

 fit , par 0 = 3.30751 par 1 = 85.4781
 bin = 21 bin center = 0.482 curr sig = 111 curr back =44.508 sum sig = 111 sum back = 44.508
 bin = 22 bin center = 0.486 curr sig = 265 curr back =44.8499 sum sig = 376 sum back = 89.3579
 bin = 23 bin center = 0.49 curr sig = 760 curr back =45.1918 sum sig = 1136 sum back = 134.55
 bin = 24 bin center = 0.494 curr sig = 1401 curr back =45.5337 sum sig = 2537 sum back = 180.083
 bin = 25 bin center = 0.498 curr sig = 1424 curr back =45.8756 sum sig = 3961 sum back = 225.959
 bin = 26 bin center = 0.502 curr sig = 938 curr back =46.2175 sum sig = 4899 sum back = 272.177
 bin = 27 bin center = 0.506 curr sig = 322 curr back =46.5595 sum sig = 5221 sum back = 318.736
 bin = 28 bin center = 0.51 curr sig = 125 curr back =46.9014 sum sig = 5346 sum back = 365.637
 signal = 5346 back = 365.637 s/b =14.621 significance = 68.1156
 # bin left = 4 backleft =159.569
 # bin right = 4 backright =203.333
 check from histo : left = 121 right = 196

      2. L/dL distribution : cuts are |n#sigma_{kaon}|<3 && |n#sigma{pion}|<3 && probability >.5

results of the fit :

 fit , par 0 = 5.68107 par 1 = 9.25827
 bin = 21 bin center = 0.482 curr sig = 38 curr back =10.1436 sum sig = 38 sum back = 10.1436
 bin = 22 bin center = 0.486 curr sig = 98 curr back =10.1806 sum sig = 136 sum back = 20.3242
 bin = 23 bin center = 0.49 curr sig = 278 curr back =10.2176 sum sig = 414 sum back = 30.5418
 bin = 24 bin center = 0.494 curr sig = 601 curr back =10.2547 sum sig = 1015 sum back = 40.7964
 bin = 25 bin center = 0.498 curr sig = 591 curr back =10.2917 sum sig = 1606 sum back = 51.0881
 bin = 26 bin center = 0.502 curr sig = 358 curr back =10.3287 sum sig = 1964 sum back = 61.4169
 bin = 27 bin center = 0.506 curr sig = 121 curr back =10.3658 sum sig = 2085 sum back = 71.7826
 bin = 28 bin center = 0.51 curr sig = 30 curr back =10.4028 sum sig = 2115 sum back = 82.1854
 signal = 2115 back = 82.1854 s/b =25.7345 significance = 44.2021
 # bin left = 4 backleft =38.5745
 # bin right = 4 backright =43.3147
 check from histo : left = 26 right = 50

• the significance decreases from 68 to 44
• the s/B increases from 14 to 25
• signal(default)/signal(prob>.5) ~ 2.5
• back(default)/signal(prob>.5) ~ 4.5
• so it seems the background is more supress with this cut

 

updated.root

•  default

|nsigma|<3 && L/dL>10

signal = 5346 back = 383.993 s/b =13.9221 significance = 67.8645

 

• vs rapidity

default+|y|<.5

signal = 3593 back = 210.438 s/b =17.0739 significance = 56.4309

 

but L/dL shows a split at L/dL>=11

signal = 3407 back = 143.571 s/b =23.7304 significance = 55.9098

 

--> so we see that a cut on the rapidity improves the s/b ratio and decreases slightly (from 67 to 55) the significance

 

• vs |nsigma|

the cut here is |nsigma|<2

signal = 4982 back = 345.591 s/b =14.4159 significance = 65.6871

 

the cut here is |nsigma|<1.5

signal = 4127 back = 277.874 s/b =14.8521 significance = 59.9163

 

--> it seems to have a very small dependence with the nsigma cut

 

• vs gRefMult

for run 7 :

 

 

a. for 40-80% : equivalent to 10 <= gRefMult<=114

signal = 603 back = 13.0869 s/b =46.0768 significance = 24.0231

 

b. for 0-5% : equivalent to 485 <= gRefMult

signal = 397 back = 21.1262 s/b =18.7918 significance = 18.8646

 

• multidimensional cuts : Sl vs probability

In the following plots, I have varied 2 cuts at the same time : the probability and the L/dL cut and shows the significance and s/b ratio for each bin (the values are the z-axis)

1. for the significance

 

     2.  for the s/b ratio

• multidimensional cuts : Sl vs pT_{D0}

1. for the significance

From this plot, I then plotted the significance and s/b ratio as a function of p_{T} of D0 ; I choose a cut L/dL>8 for all p_{T} bins from the plot above that suggested a max. significance for L/dL>8 :

Note : the error bars are only statistical , I should have propagate the error from both signal and backgroung

And finally for all p_{T} bins, I counted the signal in -1<y<1 , which gives me the following spectra :

I used ~700k events before Zvertex cuts, which makes ~400k events in |Zvertex|<10 cm

 

• multidimensional cuts : Sl vs product of Dca : DcaToPV_{pion1} * DcaToPV_{pion2}

1. for the significance

      2.  for the s/b ratio