Preparing for a new experiment
Setting up offline analysis for a new experiment typically
involves the following steps:
- Determine the experimental configuration (spectrometers,
detectors, beamline) to be analyzed and identify the corresponding
analysis modules.
- Create a database for the new experiment, using the
start date of the data taking as the time-stamp for new
new entries (see database overview).
At the minimum the database should contain up-to-date
detector map entries for every detector and rough starting
values for the spectrometer reconstruction matrix elements
and VDC timing offsets. Often this information can be
carried over from a previous experiment with only minor
modifications. Also, enter any other calibrations and geometry
data that are available, even if approximate.
- In the database, create initial run database values.
These are typically the starting beam energy and spectrometer
momentum and angle settings. If the experiment is already completed,
extract the history of these settings from logs and enter them
into the database. These values affect the kinematics calculations;
they are not important for detector checkout.
- Determine which physics calculations are needed for the
offline analysis and identify corresponding Physics Modules.
- Identify desired output histograms and tree variables.
Create an output definition file. The file
$ANALYZER/examples/output_example.def contains
most of the necessary documentation.
- If desired, create a definition file for logicals.
An example generating detailed VDC statistics is
given in $ANALYZER/examples/cuts_example.def.
- Write a CINT
script that sets up the configuration identified in
Step 1 and the physics analysis decided on
in Step 4. Often, a script from a
previous experiment, or one of the examples in
the directory $ANALYZER/examples, can serve as a guide.
The script usually also locates raw data files, creates
one or more THaRun objects, configures various
options of the event loop object THaAnalyzer, and
starts the replay. In particular, the names of the output
file, the output definition file, and the logicals definition
file must be given to THaAnalyzer.
- Identify the plots that you wish to generate from the analysis
results and write a script to create them. This may be part
of the script created in the previous step.
Note that there is no need to quit the Analyzer and start a new
session or another program after completion of the analysis; all of
ROOT's visualization tools are available from within the Analyzer.
Ole Hansen <ole@jlab.org>
Last modified: Mon Nov 17 11:21:21 EST 2003