TupleToolVtxDaughters sometimes fails to create branches
When running the full D02HHHH
Charm WG production, we observed that some of the files were missing certain branches that should have been created by TupleToolVtxDaughters
.
The expected behaviour of the tool is to create & fit so-called "sub-vertices" using subsets of decay children. For example, for a D0 -> h1 h2 h3
decay, it is expected to create three groups of branches: D0_SubVtx_12_*
, D0_SubVtx_13_*
, and D0_SubVtx_23_*
. Each of these groups should each contain branches with info including vertex position & error, flight distance, fit quality, etc.
However, in a small fraction of files from our production, we observed that certain groups weren't being created at all (eg. all of the D0_SubVtx_12_*
and D0_SubVtx_13_*
branches were present, but all of the D0_SubVtx_23_*
branches were entirely missing. This only appears to affect a very number of files, so this is an uncommon occurrence which is why it was missed during testing.
To view an example of an affected file:
[djwhite@lxplus769 resolution_profiles]$ root -l root://eoslhcb.cern.ch//eos/lhcb/grid/prod/lhcb/LHCb/Collision18/CHARM_D02HHHH_DVNTUPLE.ROOT/00111438/0000/00111438_00000296_1.charm_d02hhhh_dvntuple.root
root [0]
Attaching file /eos/lhcb/grid/prod/lhcb/LHCb/Collision18/CHARM_D02HHHH_DVNTUPLE.ROOT/00111438/0000/00111438_00000296_1.charm_d02hhhh_dvntuple.root as _file0...
(TFile *) 0x1380530
root [1] TTree* tr = (TTree*)_file0->Get("Hlt2Dstp2D0Pip_D02KmPimPipPip_Tuple/DecayTree")
(TTree *) 0x19c1f00
root [2] tr->Show(0)
In this file, the list from tr->Show(0)
will be missing the branches D0_SubVtx_12_*
. Other files exist that are missing different groups of branches (eg. D0_SubVtx_23_*
).
However, most other files from this production have their branches created & filled properly (eg. root://eoslhcb.cern.ch//eos/lhcb/grid/prod/lhcb/LHCb/Collision18/CHARM_D02HHHH_DVNTUPLE.ROOT/00111438/0000/00111438_00000296_1.charm_d02hhhh_dvntuple.root
). Since most files are unaffected, this is a relatively low-priority issue for our analysis, but others using the tool may be more affected than us.