Difference: PhysicsValidation (1 vs. 2)

Problems found:

Real data reconstruction job with 6.5.5 on SLC4_32bit machines can't get the same results as on SLC5_64bit machines.

Examples:

(1) EsTimeAlg/src/EsTimeAlg.cxx double ttof[30]={0.},helztof[30],mcztof,forevtime,backevtime; changed to: double ttof[30]={0.},helztof[30]={0.0},mcztof=0.0,forevtime=0.0,backevtime=0.0;

double cor_evtime,corev[15],tmp_time=0.; changed to: double cor_evtime=0.,corev[15]={0.0},tmp_time=0.;

(2)MdcDedxAlg/src/MdcDedxRecon.cxx add zhit_pol = 0.0; in MdcDedxRecon::execute() and MdcDedxRecon::kaltrackrec

add in MdcDedxRecon::execute() : else continue; to avoid push_back resdedx which not set initial value. dedxList->push_back( resdedx );

(3)TrkReco src/TMDCTsf.cxx int nHits[4]; //in each layer. changed to: int nHits[4]={0,0,0,0}; //in each layer.

src/TCurlFinder.cxx TTrack* TCurlFinder::merge3DTrack

at the end of the member function, at : return track;

Three physics processes and related physics distribution checked

• Reconstruct real data
• Simulate single particles with generator tester
• Simulate J/psi->rho pi with generator BesEvtGen

Status:

• No difference in the reconstruction
• Almost the same in the simulation with high statistics for three instances

Histograms for comparisons:

PDF files for comparisons:

-- ZhangXiaomei - 2010-10-25