| [FIXED] Possible bug in scrut14 release? [message #17215] | 
			Tue, 26 August 2014 12:50   | 
		 
		
			
				
				
				
					
						  
						Dmitry Morozov
						 Messages: 34 Registered: May 2007  Location: Protvino, Russia
						
					 | 
					continuous participant  | 
					From: *ihep.su
  | 
		 
		 
	 | 
 
	
		Dear colleagues, 
 
I faced the following strange behaviour which i don't understand. 
 
I'm trying to reconstruct and analyse hc -> J/psi pi0 pi0 process, giving e+e- 4gamma in the final state. 
I'm following the standard analysis strategy through ana_complete.C macro.  
So the relative part of the analysis macro: 
// *** Mass selector for the J/psi cands
	double m0_jpsi = TDatabasePDG::Instance()->GetParticle("J/psi")->Mass();  
	RhoMassParticleSelector *jpsiMassSel=new RhoMassParticleSelector("jpsi",m0_jpsi,0.3);
// *** Mass selector for the pi0 cands
	double m0_pi0 = TDatabasePDG::Instance()->GetParticle("pi0")->Mass(); 
	RhoMassParticleSelector *pi0MassSel = new RhoMassParticleSelector("pi0", m0_pi0, 0.05); 
// *** the lorentz vector of the initial hc
	TLorentzVector ini(0, 0, 5.5827, 6.598952222);
	theAnalysis->FillList(eplus,  "ElectronAllPlus");
	theAnalysis->FillList(eminus, "ElectronAllMinus");
	theAnalysis->FillList(gamma, "Neutral");
		
// *** combinatorics for J/psi -> e+ e-
	jpsi.Combine(eplus, eminus);
// *** combinatorics for pi0 -> gamma gamma
	pi0.Combine(gamma, gamma);
// *** jspi mass selection
        jpsi.Select(jpsiMassSel);
// *** pi0  mass selection
	pi0.Select(pi0MassSel);
// *** combinatorics for hc -> J/psi pi0 pi0
	hc.Combine(jpsi, pi0, pi0);
// ***
// *** do 4C FIT (initial hc system)
// ***
	for (j = 0; j < hc.GetLength(); ++j)
	{
		PndKinFitter fitter(hc[j]);	// instantiate the kin fitter in hc
		fitter.Add4MomConstraint(ini);	// set 4 constraint
		fitter.Fit();		            // do fit
		double chi2_4c = fitter.GetChi2();	// get chi2 of fit
		double prob_4c = fitter.GetProb();	// access probability of fit
		hhc_chi2_4c->Fill(chi2_4c);
		hhc_prob_4c->Fill(prob_4c);
		RhoCandidate *hc_4cf = hc[j]->GetFit(); //get fitted hc itself
		hhc_mass_4cf_prob0->Fill(hc_4cf->M());
        }
 
  
I tried to analysed two sets of simulated data - from scrut14 and trunk (rev. 25710). Version of analysis macro does not matter - the results are the same.  
 
I attached two png files - 3 histos in each: mass of fitted hc by 4c fit (by Add4MomConstraint), probability and chi^2 of the fit.  
 
  
 
In trunk case - I have all hc masses in one bin (which is how it should be? Am i right? Correct me please if I am confused.), probabilities is mostly populated around 0 and chi^2 is more or less spread over the histogramm. 
 
      
 
In scrut14 case - I have relatively wide hc mass distribution, probabilities are close to 1, and chi^2 are close to 0. 
 
Maybe somebody can comment on the source of this difference?  
 
Thank you in advance. 
		
		
		[Updated on: Wed, 10 September 2014 12:00] by Moderator Report message to a moderator  
 |  
	| 
		
	 | 
 
 
 | 
	| 
		
 | 
	| 
		
 | 
	| 
		
 |