Saikat Ray Majumder

1
Excess power method preparing for automated
analysis of LIGO data for burst sources

• Saikat Ray Majumder
• Patrick Brady
• University of Wisconsin-Milwaukee
• DCC G040179-00-Z

2
Introduction.

• Software
• LAL
• LALApps
• The excess power code runs in a stand alone mode
  using Condor as the job manager.
• Cluster
• Medusa cluster at UWM

3
The different components

• The components of excess power search
• segwizard
• LSCdataFind
• lalapps_binj
• lalapps_power
• lalapps_bread
• lalapps_burca
• lalapps_binj_find

4
What do they do?

• segwizard produces a .txt file containing the
  list of segments to be analyzed.
• LSCdataFind fetches the information about the
  location of the data in the cluster.
• lalapps_binj optional If one wants to do
  software injections then this can be used to
  generate the parameter files. Output is in both
  xml ASCII formats.
• lalapps_power the main search engine

5
What do they do? (contd.)

• lalapps_bread can read triggers from multiple
  xml files into a single xml file. The xml files
  should be listed in a .txt file.
• lalapps_burca optional looks for coincident
  events from two IFOs.
• lalapps_binj_find optional this looks for
  triggers corresponding to injected signals .
• 
  
• 
  

6
DAG

• How do we stitch in all these different
  components?
• DAG
• Directed Acyclic Graph
• The programs are nodes in the graph, and the
  edges identify the dependencies.

7
DAG (contd.)
LSCdataFind
8
DAG
LSCdataFind
lalapps_burca
lalapps_burca
9
DAG ( Contd.)

• The DAG itself is defined by the contents of a
  DAGMan (Directed Acyclic Graph Manager) input
  file.
• Hence one needs to write this DAGMan input file,
  specifying the list of programs in the DAG, their
  condor submit description files, dependencies in
  the DAG.
• This file is submitted to the Condor.
• DAGMan is then responsible for scheduling,
  recovery, and reporting for the set of programs
  submitted to Condor.

10
Some General Issues

• Technically one can include all the processes
  involved in the analysis in the DAG, but..
• that may not be the most efficient thing to
  do.
• Our DAG includes
• LSCdataFind
• lalapps_binj optional
• lalapps_power
• lalapps_ burca optional

11
Conclusion

• So we are at a stage where we are almost ready to
  analyze the data in a completely automated way
• AS_Q data
  xml files containing
  
• 
  triggers
• Immediate things to do
• Implement time shifted background estimation (
  almost done )
• Implement a correlation test in the pipeline
  

DAG