Reconstruction tutorial: Part 4

1
Reconstruction tutorial Part 4

• Writing a (very) simple Reconstruction/Analysis
  algorithm Z?ee-
• Simplified from a real algorithm by Maarten
  Boonekamp (thanks!)
• Access egamma objects, select some, compute
  invariant mass of pairs, and save a Zee object
• Check the Z in the truth

2
Where you should be now

• At that stage, you should have a running ZeeRec
  algorithm
• which prints a few things (grep Zee AskAthena.log
  )
• book variables in the ntuple but do not fill them
• loop on egamma object and fill ZeeObject but no
  computation of the mass have been done

3
Compute kinematic quantities

• In method ZeeObjectsetDaughters() called by
  the constructor, fill two HepLorentzVectors (
  http//wwwinfo.cern.ch/asd/lhc/clhep/doxygen/htm
  l/index.html ) with electron 4- momentum
• m_egamma0eg1 m_egamma1eg2
• HepLorentzVector hlv2
• // get kinematics of the two electron and
  store in HepLorentzVector
• for ( int i0 ilt2 i )
• const LArCluster clus m_egammai-gtget_Cluste
  r()
• const double eta (clus)-gteta()
• const double phi (clus)-gtphi()
• const double et (clus)-gtet()
• const px et sin(phi)
• const py et cos(phi)
• const pz et sinh(eta)
• const en et cosh(eta)
• hlvi.set(px,py,pz,en)

4
kinematic quantities

• Add 4-vectors
• // sum kinematics in the internal 4-vector
• m_hlvZhlv0hlv1
• // already coded for you in ZeeObjectmass()
  etc
• // mass() m_hlvZ.m()
• // et() m_hlvZ.perp()
• // phi() m_hlvZ.phi()
• // eta() m_hlvZ.pseudoRapidity()
• Run again and check the variables in the ntuple
  make sense (Z mass ?)

5
Note on const policy

• We clearly want to avoid that, say, the person
  writing egamma identification modify (on purpose
  or by mistake) the content of, say, the track
  collection, which would affect downstream, say,
  primary vertex search
• (in some very specific case, it can be that, on
  purpose, a collection is  finished  by another
  algorithm. Allowed only if it makes the code much
  simpler)
• Collection in StoreGate are/will be locked?only
  accessible through const datahandle
• The rest of the code need to be const-correct
• Note that current compiler (gcc 2.95) is not very
  strict on that, but the next one (gcc 3.2, will
  become the default in the course of the year)
  will be!

6
Tips on const correct code

• Const double a 1 // I promise not to modify a
• Const MyObject a MyObject() // I promise
  never to modify a
• Const MyObject a new MyObject() //I promise
  never to modify the MyObject on which a points
• MyObject a new MyObject() const //I promise
  never to modify a
• MyObjectMyDoer(const AnotherObject a, const
  AnotherObject b) //MyDoer promises not to
  modify a (which is passed by reference) nor the
  object pointed at by b
• ?MyDoer accepts objects that are declared const
  or not
• Note that the const-ness is part of the signature
  of the function so MyObjectMyDoer(const
  AnotherObject a, AnotherObject b) is another
  function

7
Tips (2)

• MyObjectDoer() const //Doer promises not to
  modify the object it belongs to
• ?only such const methods can be used from an
  object that was declared const
• MyObjectDoerNonConst()
• const MyObject theObject
• theObject.DoerNonConst() //will not compile
  (error message no matching method best
  match..)
• theObject.Doer() //will compile
• So if you suddenly want to be const correct, all
  the code which depend on you may not compile any
  more if const-uncorrect
• Hence the need to take this habit early

8
Tips (3)

• Code with const is more readable because you
  immediately spot const variables (obviously)
• especially when reading header files
• Many mistakes can be detected at compile/link
  time through const correctness
• A trivial example
• Const a theObject.doer()
• If (a1) // will not compile, where is the
  bug ?

9
Compute Z mass from tracks only

• Add variable m_massFromTracks in ZeeObject and
  output in the ntuple
• Elements of code (not complete !)
• Add variable in ntuple (cut and paste from
  z_mass)
• In ZeeRec/CBNT_Zee.h add m_z_massFromTracks
• In src/CBNT_Zee.cxx declare the ntuple variable
  and fill it from ZeeObj
• Navigation from egamma object to the track
• EMTrackMatch trackMatch m_egammai-gtget_EMTra
  ckMatch()
• SimpleTrack track trackMatch-gtgetTrack()
• Do not forget to test pointers are non-null
• you will need include files SimpleTrack/SimpleTrac
  k.h and egammaEvent/EMTrackMatch.h
• Computation of kinematics
• Pxcos(track-gtPhiVert())/track-gtPTInvVert()
• Pysin(
• Pztrack-gtCotThVert()/ track-gtPTInvVert()
• ?4-momentum then mass
• Please try now

10
HepMC

• HepMC is the truth event record. It has vertex
  and particles connected in a tree.
• (see http//cern.ch/hepmc )
• It stores the output from the generator, physical
  particles as well as some history/hard scatter
  (GenStat3)
• Exercise find the Zs and its daughters in the
  truth and printout some quantities in the log
  file

11
NoteGeant particle in HepMC

• Output from Geant3 is also stored in HepMC
• allow smooth navigation between generator world
  and geant world
• Avoid duplication of particles
• Allow same interface
• However difficult to store geant specific
  information so this will need to be eventually
  improved
• The original kine Geant3 number need to be
  recovered because it is still used to label
  detector digits
• It is stored in barcode() or status() depending
  of the cases (see Generators/GenzModule/doc/GenzMo
  dule.pdf )

12
Look for the Z

• //in Zeerecexecute get pointer of MC
  collection
• const McEventCollection mcCollptr
• if ( m_storeGate-gtretrieve(mcCollptr).isFailure(
  ) )
• mlog ltlt MSGERROR ltlt "Could not retrieve
  McEventCollection"ltlt endreq
• return StatusCodeFAILURE
• //loop on event in event collection
• McEventCollectionconst_iterator itr
• for (itr mcCollptr-gtbegin()
  itr!mcCollptr-gtend() itr)
• const HepMCGenEvent myEvent(itr)-gtpGenE
  vt()
• HepMCGenEventparticle_const_iterator
  Part
• //loop on particles in event
• for (Part myEvent-gtparticles_begin()
  Part!myEvent-gtparticles_end() Part )
• // good particle
• if ((Part)-gtpdg_id()23)
• //dump in log file (note that a HepMC
  particle is-a HepLorentzvector
• mlog ltlt MSGINFO ltlt "Z found with mass " ltlt
  (Part)-gtmomentum().m()
• ltlt " barcode " ltlt (Part)-gtbarcode() ltlt "
  status " ltlt (Part)-gtstatus()

13
Find daughters

• // find decay vertex
• const HepMCGenVertex
  endVert(Part)-gtend_vertex()
• //loop on daughters
• for (HepMCGenVertexparticles_out_const_iter
  ator
• itrDauendVert-gtparticles_out_const_begin(
  )
• itrDau!endVert-gtparticles_out_const_end()
  itrDau )
• mlog ltlt MSGINFO ltlt " daughter found with
  type " ltlt (itrDau)-gtpdg_id()
• ltlt " barcode " ltlt (itrDau)-gtbarcode()ltlt
  " status " ltlt (itrDau)-gtstatus()
• ltlt " Pt " ltlt (itrDau)-gtmomentum().perp()
  ltlt endreq
• Please try now