1. IITF and UPCs UPC Specific Requirements Processing  0 analysis e + e - analysis (by Vladimir Morozov) Conclusions Spencer Klein

2. S. Klein, LBNL UPC specific requirements n Small events u Mostly 2 tracks F Low multiplicity vertexing is critical F Vertex position resolution matters F ‘per event’ memory leaks are painful n Low p T tracks u The e + e - analysis goes down to p T ~ 60 MeV F That’s where the signal is F Half field data e/  separation u dE/dx

3. S. Klein, LBNL  0 analysis 100,000  0 -->  +  - w/ IITF and old tracker  |Y  |< 1.5 u 20 files of 5,000 events each u 1 IITF file crashed F Why? F Use 95,000 ‘common’ events  IITF  0.event.root files are twice as large as old tracker event.root F Why? u Work from StRoot output u No Monte Carlo info on uDsts F Can’t show resolutions

4. S. Klein, LBNL Basic event properties n Old: 93815 events n New: 70414 events u Where did the rest go? F Events with no tracks lost?? n Nprim & Ntot distributions are totally different u Massive track splitting in IITF?? u Technical Problem?? F IITF +’old’ tracks together n Events with vertex u Old: 29620 events u New: 28482 events F 4% lower Ntot Red=IITF Blue – old tracker Nprim Red=IITF Blue – old tracker

5. S. Klein, LBNL Vertex Properties n Look very similar Zvertex Red=IITF Blue – old tracker Rvertex Red=IITF Blue – old tracker

6. S. Klein, LBNL  0 properties n Take particle pairs n Track splitting largely kills the analysis u Old: 29596 pairs u New: 144501 pairs n nPrim==2 cut u Most IITF events lost  M  spectra look similar  M  wrong F Generated at half field, reconstructed at full? Pair Mass (all) Red=IITF Blue – old tracker Pair Mass (nPrim==2) Red=IITF Blue – old tracker

7. S. Klein, LBNL  0 p T n Require nPrim==2 n Spectra look similar u Is scale correct? n Old RMS p T =194 MeV/c n New RMS p T =258 MeV/c n Substantially different Pair p T (nPrim==2) Red=IITF Blue – old tracker

8. S. Klein, LBNL e+e-e+e- n Analysis by Vladimir Morozov n Should be generated and reconstructed with half field n 100,000 events n IITF finds many more tracks than non-IITF n The number of primary pairs is comparable?? n IITF creates an awful lot of 1- track events u Because of the kinematics, we expect a lot, but why are so many more in IITF N tracks N prim prs

9. S. Klein, LBNL e + e - p T and M inv n Look fairly similar n Limited statistics u Not enough for detailed comparison

10. S. Klein, LBNL Conclusions The IITF  0 sample showed massive track splitting u Algorithmic flaw or technical problem? F There is no documentation for the files that were produced u The problems made it impossible to make sensitive comparisons  RMS P T for the 2  0 samples was substantially different n The IITF e + e - sample had a huge number of 1-track events u Would a similar analysis on real data behave similarly? n These tests give us no confidence that IITF is ready for use. n We request a well-documented production of our files and adequate time to test the output files. u Ideally, this test will include studies with data as well as simulations.