For some studies, I have generated with starsim 10k D0's (1D0 per event) with power law pt (in 0<pT<5) but with a larger timelife :

• 100ctau sample where the <decay> of the D0 is 122microns*100 = 1.2 cm
• 1000ctau sample  where the <decay> of the D0 is 122microns*1000 = 12 cm

We want to test if the DCA to Primary Vertex filled in MuKpi (from MuDst) is agree with large decays.

This sample can also be used to test the accuracy of the second vertex fit

1. 100ctau sample : pdf is here

With this sample, I also looked at the minimc files. The goal is to evaluate some efficiencies of our reco. code.

# D0 in geant (N) --> effic_1 --> # D0 in minimc (n1) --> effic_2 --> # of D0 reconstructed with MuKpi (n2)

To find (n1), I loop over entries in minimc files and built the inv. mass with the tracks in same event.

Cut are applied later to select 'real' D0, ie where daughters arer actually a pion (geantid = 8) and a kaon  (geant id =12) , and for which their parent is a D0 (geant id = 37)

• parentKey_Pair == parentKey  (assure that the D0 is the same, in case of several D0 embedded per event)
• geantId ==8 && geantId_pair==12 (select pion(-) and kaon(+)) 
• sign>0 && sign_pair<0 (somehow useless) 
• parentGeant==37 && parentGeant_pair==37 (assure that the parent of each track is a D0)

(n2) is the results of Mukpi (production 3 version) and I added a cut on the charged of daughters

Fig1 : inv. mass and pt from the minimc files

Fig2 : inv. mass and pt from the reco files

Fig3 : ratio reco/minimc (effic_2)

comments :

• inv. mass from the reco looks better than the minimc (?)
• ratio is > 1 for pT>2.5 : possible explanations :

1. too more Kpi pairs reconstructed in MuKpi because of particle_id not enough to select correctly pion and kaon
2. there are events where 2 candidates are reconstructed, meaning there is at least 3tracks in this event : pion, kaon and some other track

Requiring # candidates / event ==1 reduces a bit the efficiency 

Fig4 : ratio (reco/minimc) with # candidates = 1