The STAR experiment

I deployed a private build of SUMS (roughly 1.8.10) on CMSAF and made the following changes to globalConfig.xml to get local job submission working:

In the Queue List

In the Policy List

Now for the Dispatcher

And finally, here's the site configuration block

More details from L2-jet algo in PDF

First things first. I went to http://www2.lns.mit.edu/compserv/cms-acctappl.html and applied for a local account. The welcome message contained a link to the CMSAF User Guide found on this TWiki page.

AFS isn’t available on CMSAF, so I started a local tree at /osg/app/star/afs_rhic and began to copy over stuff. Here’s a list of what I copied so far (nodes are running SL 4.4):

I copied these precompiled libraries over instead of building them myself because of a tricky problem with the interactive nodes’ configuration. The main gateway node is a 64-bit machine, so regular attempts at compilation produce 64-bit libraries that we can’t use. CMSAF has a node reserved for 32-bit builds, but it’s running SL 3.0.5. We’re still working on a proper resolution of that problem. Perhaps we can force cons to do 32-bit compilations.

The environment scripts are working, although I had to add more hacks than I thought were necessary. I only changed the following files:

1. ~/.login
2. ~/.cshrc
3. $GROUP_DIR/site_post_setup.csh

It doesn’t seem possible to change the default login shell (chsh and ypchsh both fail), so when you login you need to type “tcsh” to get a working STAR environment (after copying my .login and .cshrc to your home directory, of course).

Basic interactive tests look good, and I’ve got a SUMS configuration that will do local job submissions to the Condor system (that’s a topic for another post). DB calls use the MIT database mirror. I think that’s all for now.

MIT’s CMS Analysis Facility is a large Tier-2 computing center built for CMS user analyses. We’re looking into the viability of using it for STAR computing.

Here are the plots from 5000 gamma events, with a seed tower energy cutoff of 1 GeV and a cluster energy cutoff of 4 GeV.  The first page has on the upper left total cluster energy, on the upper

See the below pdf.  This was made from 5000 piminus events.  The error bars were calculated from the formula delta-eff=sqrt(B*C/(B+C)^3), where B is the number of accepted tracks and C the n

See the pdfs below.  The second page has reconstruction efficiencies.  Note that a cut of pt>5 GeV was imposed on the reconstructed tracks to try to reduce the impact of secondary particl

We've generated the following MC datasets, all located in /star/institutions/mit/corliss/magic_single_particle/

See the pdfs attached below, each covering 1000 events of the given particle type in the given detector.  Page 1 has the standard MC QA plots, page 2 has plots for MC primary tracks only, page 3