LIGO_practical_intro_doc.md

LIGO practical intro doc

Some useful links for registering to LVK and test-runs of GWs events.

Step 1: Become a LIGO member

• Register for a LIGO membership here.

After your membership is accepted:

• Register for LIGO's Mattermost here.

  Many channels are open for joining $\rightarrow$ you can find them by clicking the $+$ sign, next to LIGO at the top-left. This will open a list of available public channels (beware, the list continues for multiple pages - click Next at bottom-right, to check them all).

• Apply for a LIGO cluster account here.

• Create a GitLab LIGO account here.

Step 2: Background for GWs analysis

• The GWs Open Data Workshops (here) are very useful to understand the basics of GWs analysis. There is also a dedicated forum (here) with Q&A about the workshops, but also more general issues.

• The IGWN Public Alerts page (here) has a very useful website describing the whole process, but also many of the relevant terminology.

Step 3: LIGO Observing runs

• A useful compilation of FAQ can be found here.

• A quick tutorial, using bilby to analyse, i.e. do a parameter estimation (PE), of a GW event is here. Beware that there could be some typos on the event's naming, so try to be consistent with the original event's naming at the top, i.e. the command that creates the bilby configuration file.

• The general idea is the following:

  1. We create a directory for the PE run
  2. Activate relevant python environments
  3. Make all the necessary (physical) changes in the filename_config.ini
  4. Secure permission (If problems with permission occur, check (here) for alternative authorisation options. Sometimes the error log files of a failed job can have useful links.)
  5. Run the analysis: bilby_pipe filename_config.ini --submit
  6. Check analysis: condor_q --nobatch. More on HTCondor workload manager here.

• More details on bilby configuration options can be found here.

  • For user-interface, i.e. options in the config file, check here.
  • Bilby can be used to create a configuration .ini file based on GraceDB events. For a specific example (here) and more generally (here).
  • Note: There is a dedicated channel in LIGO mattermost (Bilby Help) for issues with bilby.

• Where to run?

  To access the LIGO clusters, follow the ways here. For UNLV GWs group, we use the CIT site, with the following hosts:

LIGO has a number of workstations - the HTCondor system will distribute the run accordingly. The ldas-grid and ldas-pcdev* are good for hosting your files and analyses (check the system configuration, in case you are interested in a job with specific characteristics, like high memory etc).

Step 4: Checking your bilby runs' results

Of course, the files could be downloaded and inspected locally. But there is also the possibility to examine some of the plots online:

• Follow the links here. Either Jupytel Lab or the public_html page organise the produced plots.

• To create a summary webpage (shown in public_html above), the following options must be selected in the configuration file:

  create-summary = True
  email = [email protected]
  webdir = /home/albert.einstein/public_html/project
  

• The dag_name.submit file and bash_name.sh file have useful information on the order of jobs submissions, but also on the commands used at each step that initialise the specific part of the run (these can be useful, if you want to check)

• The different .log files show details (duration, memory etc) about the runs.