It was reported that the tuning of the scale factor (Normalization Factor in TRS) was giving different results in SL08c library at PDSF and RCF (see here). It has not been possible to reproduce this with SL08c owing to the difficulty of checking out and compiling all of the updated packages (it should be possible but I was not able to.)

Instead the newly-tagged SL08d library was used at both RCF and PDSF. At each site embedding was run over all the available daq files in

/eliza13/starprod/daq/2008/ and /star/institutions/lbl/daq/2008/ respectively. In each case two sets were run with scale factors of 1.38 and 2.67 used.

Jobs were run with the following parameters

star-submit-template -template embed_template.xml -entities PTLOW=0.1,PTHIGH=2,ETALOW=-0.5,ETAHIGH=0.5,PID=8,MULT=0.15,TRSNORM=2.67 [or 1.38]

Note that the multiplicity was set to 15% of the event multiplicity in order to build up adequate statistics from a single pass.

The plots below, generated from the minimc.root files, shows the dE/dx vs p_T for "ghost pairs" (the real data) in black, overlaid with that from "matched pairs" (the embedded tracks) in red. For PDSF and RCF the scale factor of 1.38 is most appropriate (further fine tuning might improve this) and 2.67 is clearly not appropriate.

Things to note. At PDSF the I only got 26 completed jobs instead of 34 at RCF so the stats are different (don't know why yet).

Embedded tracks are flat p_T so they dominate the real data as higher p_T which is why they look wider. There are no track quality cuts sued to generate the plots.

For reference the Job IDs were:

PDSF

SF=1.38 5CB3516941A64D0AFEB8FFB5003106C5

SF=2.67 78522EB1476CDA7E191F611BC6E7C909

RCF

SF=1.38 F0E4E999804F6E669910B8F79ECA5FF0

SF=2.67 042A7ED7A99964487B6E96AC2E324370

Output can be found under

/eliza13/starprod/embedding/P08id/AuAu9/

/star/data05/scratch/lbarnby/embedding/P08id/AuAu9/