Data access and analysis: CSC data

Page under construction

Please see the page "Data access and analysis" for a general introduction.

The Computing System Commissioning (CSC) production is designed to exercise the ATLAS computing model and give the physics groups the possibility to train for future data taking.

Useful references

Requests for data samples
Status of production
Status of the AOD replicas. You can navigate from this page to specific sites, e.g. the CERNCAF site.
Dataset Number Broker for Monte-Carlo Production -- AMI interface to dataset number use and requests
JobOptions for generators and each dataset -- LXR location
JobOptions for event generation, see also the README file.

Problems of CSC data

The following list is probably incomplete.

Problem of 30μm vs 1mm LAr simulation
- Discussion of the problem and actions taken, Minutes of Physics Coordination 2007-03-21
- 30μm or 1mm? -- In the displayed list: click on the data set number
Release 13 and earlier releases.
In general, backwards readability between release 12 and 13 for ESD and AOD is not guaranteed. It is planned that backwards readability is available from 13 onwards.
Naming conventions for the trigger -- changed for releases later than 12.0.6. The meaning of the threshold value is different for some trigger menus.
Wrong cross section when running AlpGen
Reported in HerwigForAtlas and the HN Forum, with recommendation how to correct the error.

Availability of data

The ATLAS computing model foresees that data are available at Tier-1/2 centers and jobs are sent where the data are located to minimize network traffic. This model is today not yet fully operational. Hence you may choose to download the data to your local disk. Avoid to make private replicas on CERN or Tier-1 storage. The most reliable Tier-1 centers are at present BNL, CC-Lyon, FZK and CERN. To minimize your problems, consider whether you can work with existing DPDs or CBNTs; preferably work with one of the physics groups to profit from their experience.

The situation of CSC data (Physics coordination, 2007-02-20): For reasons of saving disk space, some CSC datasets will be archived, i.e. put on tape: for release 11 -- all hits and RDO datasets; for release 12 -- after reconstruction 90% of hits and 80% of RDO datasets; the additional 20% of RDO datasets are kept until pileup is finished. AOD data are replicated automatically to all Tier-1 centers and Tier-2 clouds by subscription. RDO data are analogous to real data for which there will be one copy (mainly on tape). All datasets with streaming data are located at BNL, and are partly replicated to RAL and CC-Lyon.

Availability of data on Tier1

The status of the AODs at different Tier1s can be accessed from the monitoring section of the Computing Operations TWiki page.

Naming conventions for CSC datasets and files

The data consist of datasets, which group together many datafiles, which were produced under the same conditions (jobOptions, selection cuts). Meta data describe each dataset, e.g. number of events, physics channel and responsible physics group.

Datasets and files follow naming conventions. For the CSC datasets the front part xxxcsc.mmggvv of the logical dataset name describes the magnetic field, the geometry and the version of release.

mm (magn field)	gg (geometry + material)	vv version (+ bug fix)
00 old	00 ideal	00
01 new	01 misaligned
	02 misaligned + distortions

For detailed naming rules see the page File naming conventions. For documentation of detector geometry layouts see Tags in the ATLAS geometry database, Distorted material, and Detector layouts in release 13 [pdf].

Note: this convention is not followed strictly, nor is it sufficiently documented. The BPhysics group has defined its own BPhys naming conventions.

Magnetic field and distorted material in CSC simulation

Magnetic field maps -
Distorted material -- Inner detector and LAr; plots of X/X0 as function of eta, for InDet and before first active EM calorimeter layer