PTPlot

PTPlot is a plotting tool for visualising the gravitational wave power spectrum from first-order phase transitions.

Prerequisites

• django
• dulwich [pure python git implementation]
• matplotlib
• scipy

Getting started

1. Set up "ptplot" as a django project by running:

   django-admin startproject ptplot_site
   

   This will create a new folder called ptplot_site. It is useful to create a symbolic link to your main ptplot folder (i.e. ptplot/ within this repo) here:

   ln -s <path to this repo>/ptplot ptplot_site/ptplot
   

   Alternatively, update the path in step 3.

2. Add "ptplot" to your INSTALLEDAPPS (in `ptplotsite/ptplot_site/settings.py`) like this:

   INSTALLED_APPS = [
       ...
       'ptplot',
   ]
   

3. Include the ptplot URLconf in your project urls.py (i.e. ptplot_site/ptplot_site/urls.py) like this:

   path('ptplot/', include('ptplot.urls')),
   

   You will first need to import the include function:

   from django.conf.urls import include
   

4. Run python manage.py makemigrations ptplot and python manage.py migrate.

5. Finally, run python manage.py runserver --nothreading and visit http://127.0.0.1:8000/ptplot/ to test the thing.

Benchmark points

These are created in the database by running python manage.py populate which calls ptplot/management/commands/populate.py. Edit that file to add benchmark points.

Threading issues when running locally

Since commit 5f861e5 we have removed all threading from the PTPlot code. When deploying the production server we use gunicorn to serve requests using the default 'sync' worker type (https://docs.gunicorn.org/en/latest/design.html#sync-workers) (see below for more details about deploying). This helps to prevent issues with non-thread-safe code. When running locally using the manage.py runserver command (see https://docs.djangoproject.com/en/4.2/ref/django-admin/) we recommend adding the argument --nothreading to avoid the use of threads. Errors when using threading seem mostly to originate in matplotlib's mathtext library (https://matplotlib.org/stable/tutorials/text/mathtext.html), so an alternative workaround would be to change how text is rendered in plots.

Deploying

For best performance, run using gunicorn behind a buffering proxy (e.g. nginx). Ensure that gunicorn has enough worker threads to handle separate plotting computations for 2-3 plots per page.

Elastic Beanstalk

If deploying to Elastic Beanstalk, use the following in .ebextensions/django.config:

packages:
  yum:
    texlive: []
    texlive-dvipng-bin: []

option_settings:
  aws:elasticbeanstalk:container:python:
    WSGIPath: ebdjango/wsgi.py
  aws:elasticbeanstalk:container:python:staticfiles:
    /static/: "ptplot/static/"