GOAL : following the procedure of previous post, I should start with single track cut to remove background. (for D⁰ reco. with SVT/SSD)

Daughters of D⁰ should have longer DCA than background ,then a cut on the track significance S_t = DCA/σ_DCA should be useful

Samples used :

1. hijing file used for mixing 
2. single D0 with regular lifetime
3. single D0 with 100*ctau (100 * 120μm = 1.2 cm)

As there is 1 D0 per event, reconstructed daughter candidates from 2. and 3. are the real daughters, ie coming from a displaced vertex = larger DCA.

The following plot shows :

• upper left : S_t for hijing
• upper right : S_t for single regular ctau
• bottom left : S_t for single 100*ctau
• bottom right : all
• the cut used is : # of si hits =4 && |ndEdx_K|<2

It is hard to really discriminate the background and single D0 (the HFT studies show it clearly) but with the long ctau D⁰ sample, we see that S_t is broader.

** July 8th : UPDATE **

Fig. 2 : same cut as Fig. 1 but for DCA_XY

Fig. 3 : same cut as Fig. 1 but for DCA_Z

Fig. 4 : same cut as Fig. 1 but for the full DCA

The full DCA is calculated as DCA = √(DCA_XY²+DCA_Z²) [same for the errors]

Comments : for Fig. 4 , it seems the signal curve is slightly above the background for DCA/σ_DCA > 1

Fig. 5 : same cut as Fig. 4 but with a cut on the # of silicon hits decreased to 2 or more.

Fig. 6 : same cut as Fig. 5 but with a cut on the Pt of daughters more than 1

The cut on Pt of daughters is motivated but the plot below :

Fig. 7 : Pt distribution 

** September 10th : UPDATE **

Instead of using dcaXY, I used dca3D (from global) :  dcaG[0] = TVector3(mDCAGlobal_mX1Gl[kg],mDCAGlobal_mX2Gl[kg],mDCAGlobal_mX3Gl[kg])

It seems that the significance obtained with this DCA gives a much better discrimination between signal and background.

Results are here