Print

---

Hi,

 in the last version of recoilNtp I have:

   if (chbcand) {
    lundB = (TMath::Abs(tmpdauB)/tmpdauB)*521;
    if (TMath::Abs(lundB) == 521) {
               -----------------
      fBrecoCharge = 1;
      fBrecoFlavor = 1;
    }
    if (TMath::Abs(lundB) == -521) {
               ------------------
      fBrecoCharge = -1;
      fBrecoFlavor = -1;
    }
  } else {
    lundB = (TMath::Abs(tmpdauB)/tmpdauB)*511;
    if (TMath::Abs(lundB) == 511) {
               -----------------
      fBrecoCharge = 0;
      fBrecoFlavor = 1;
    }
    if (TMath::Abs(lundB) == -511) {
               ------------------
      fBrecoCharge = 0;
      fBrecoFlavor = -1;
    }
  }

this could be the reason why the efficiency seems to be too low (50%
due from this bug). I removed the Abs and commited. Take the HEAD.

 There is another suspect cut. The number of charged kaons in the recoil
is higher than the one in the old Cpf analysis (mean 0.4 -> 1.4). As a
consequence, this cut is very inefficient even in Vub events (30% of
efficiency instead of 90% of the old analysis). Since kaons are not
present in Vub events this sounds really bad.

 Daniele