Here are the two steps we use to improve the W(lnu)W(jj) mass resolution: 1.) Compute the neutrino pz by requiring the nominal W mass (i.e., 80.4 GeV) for the leptonic W. M_W^2 = (E_lepton + p_neutrino)^2 - (px_lepton + px_neutrino)^2 - (py_lepton + py_neutrino)^2 - (pz_lepton + pz_neutrino)^2 ------- Eq. MET = sqrt(px_neutrino^2 + py_neutrino^2) ------- Eq. If there are two acceptable physical solutions for neutrino pz then we take the one which is closest to the lepton pz. If the solutions are imaginary (which happens some small fraction of the time), then we take the real part. This means assuming the discriminator of the quadratic equation to be 0, which in turn gives another quadratic equation involving MET. We then redefine the MET to be the solution of this Eq. The code for this step in (cvs HEAD version) ElectroWeakAnalysis/VPlusJets/src/METzCalculator.cc (.h) 2.) Now we perform a kinematic fit constraining the hadronic W mass to the nominal W mass and leptonic W mass also to the nominal W mass. We take the netrino pz from step 1 to form the starting 4-vectors for neutrino. This kinematic fit gives us the m_WW which has better resolution than the simple addition of 4-vectors of the final state particles. The code can be found in lines 793--803, i.e., just below the comment "Do kinematic fit" in ElectroWeakAnalysis/VPlusJets/test/kanamuon.C (or kanaelectron.C) You also need to check out HEAD version of PhysicsTools/KinFitter.
/afs/cern.ch/work/u/usernameSo, the new workspace is not hanging from your $home but on a different path which contains the first letter of your loginname and then your loginname.