70 likes | 148 Views
Inv. Mass for the recent production. details. Not all the production (only for day124 to day153 ; I merged those files before we finished the full prod. this week) So it represents ~ 22M events
E N D
details • Not all the production (only for day124 to day153 ; I merged those files before we finished the full prod. this week) • So it represents ~ 22M events • Idea : previous look by tightening nSigma(dEdx) cuts did not improve a lot, so I don’t use them (|nSigma|<2 by default) and I tried to use geometrical cuts (as TCFiT is a geometrical method) • I rerun over all the files produced with additionnal cuts
Additionnal cuts • |cos*|<.6 • cos ( pointing) > .90 • gRefMult > 800 • pT>1 • Probability(TCFIT) >.75 • 0.02 < slength(TCFIT)<.07 • siliconHits ==4 for both daughters • It seems that gRefMult and pT are efficient to remove the background • For the pT cut, it takes advantage of tracks less curved and also we gain in pointing resolution
Inv. Mass K-π+ • The single bin remains above the background line even by varying the cuts so I’m not sure it’s only a fluctuation (it’s also far above the background line)
Inv. mass fit (i) • In that case, BRTW fit gives a significance of 3.56 • The mean is not 1.86 but slightly shifted (1.855)
Inv. Mass fit (ii) • Here cos ( pointing) > .97 • Fit by BRTW gives significance = 4.13
comments • With tighter cuts to decrease the background, a ‘signal’ can be seen • It’s a bit worrying that it’s only 1 bin but it remains above the background line even by varying cuts • Mixing and cutting the kinematical variables (pT, angle in lab and rest frame) could be a source of fake mass reconstruction • Will look at the full production (~ 55Mevents we produced)