Resources

How to Run PyCBC on Any Cluster

This tutorial guides you through installing PyCBC, Pegasus, and all necessary dependencies on a cluster, enabling you to confidently execute full gravitational wave search pipelines.

Step 1: Install Conda

Download the latest installer from:

https://repo.anaconda.com/archive/

$ bash Anaconda3-2023.09-0-Linux-x86_64.sh

$ source ~/.bashrc

Step 2: Install PyCBC

                $git clone https://github.com/gwastro/pycbc
                $ cd pycbc/
                $ pip install .
                $ pip install -r requirements.txt
                $ pip install -r companion.txt
            

Step 3: Run a Test Example

Navigate to the PyCBC example directory and run:

$ bash gen.sh

This example:

• Generates a bank using bank.sh
• Downloads frames using get.sh
• Generates stats with stats.sh

Step 4: Install Java (for Pegasus)

1. Download Java 8 (JRE) from Oracle: Link
2. Copy it to the cluster:

$ scp jre-8u391-linux-x64.tar username@ldas-grid.gw.iucaa.in:~/soft/

3. Extract and configure:

            $ cd ~/soft
            $ tar -xf jre-8u391-linux-x64.tar
            $ vi ~/.bashrc
            export PATH="/home/username/soft/jre1.8.0_391/bin:$PATH"
            $ source ~/.bashrc
            $ java -version
            

Step 5: Install Pegasus

1. Check system architecture:

$ arch

2. Check OS version:

$ condor_status -long | grep '^OpSys'

3. Download the correct binary from: Pegasus Downloads
4. Copy and extract:

            $ scp pegasus-binary-5.0.6-x86_64_ubuntu_20.tar username@ldas-grid.gw.iucaa.in:~/soft/
            $ cd ~/soft
            $ tar -xf pegasus-binary-5.0.6-x86_64_ubuntu_20.tar
            $ vi ~/.bashrc
            export PATH="/home/username/soft/pegasus-5.0.6/bin:$PATH"
            $ source ~/.bashrc
            $ pegasus-version
            

Quick Checklist

If all commands return version info, you're good to go!

Happy GW searching!

How to Submit Your First HTCondor Job?

Need to run your analysis across a cluster? HTCondor is your friend. Here's a quick and clean guide to get your job running in no time.

Step 1: Create a Run Script

This is the script that HTCondor will execute. Make sure it runs your full pipeline or Python script.

            # run.sh
            #!/bin/bash
            set -e
            source /path/to/your/env.sh  # Activate your conda env, load modules, etc.
            python my_analysis.py --config config.yaml
            

Tip: Make it executable with chmod +x run.sh.

Step 2: Write a Submit File

This file tells Condor how to run your job, how many resources to request, and where to save logs.

                # job_submit.sub
                universe       = vanilla
                executable     = run.sh
                log            = job.log
                output         = job.out
                error          = job.err
                request_cpus   = 2
                request_memory = 8GB
                request_disk   = 4GB
                queue
            

Pro tip: Customize cpus and memory for your job size!

Step 3: Submit Your Job

Fire off your job with this one-liner:

condor_submit job_submit.sub

Step 4: Monitor and Debug

Watch your job queue or troubleshoot with these commands:

• condor_q — See your jobs in the queue
• condor_history — View completed jobs
• condor_q -better-analyze — Diagnose why a job isn’t running

Learn More

• HTCondor Official Docs
• Center for High Throughput Computing (CHTC)

Cool Vim Hack

1. Replace a word throughout a large file

Goal: Replace every instance of GWOSG with GWOSG_FREE.

            :%s/GWOSG/GWOSG_FREE/g
            

Explanation: % means the entire file, s is for substitution, and g makes it apply globally on each line.

Bonus: Remove all extra spaces at end of lines

            :%s/\s\+$//g
            

Explanation: This removes all trailing whitespace from every line in the file.