ThesisPlots

Collection of plots of dissertation thesis, simplified version of the original repository

Use case

If you want to organize your thesis plots in an easily reproducible manner, this might be a nice example for you.

Getting started

# Fork the repository to your own github account, and copy it locally (or on a cluster)

# clone the repository
git clone [email protected]:<YOUR_GITHUB_ACCOUNT>/thesis_plots.git

# install the repository and it's dependencies. You can either do this in your own or  
pip install -e thesis_plots

Additionally, you want to setup latex such that you can use it to rended axis labels etc. There is an example in the pytest script for linux and macos (I'm sorry Windows users maybe you want to install a WSL anyway 😉). The latex rendering on macos is still untested at the time of writing.

Now you can add new notebooks in the notebooks folder. If you are happy with a new plot you made, you can upload it using the normal git commands:

cd thesis_plots

# Check which changes you made
git status
git diff

# Make a new branch
git checkout -b <NEW_PROJECT>

# Now add and commit to the repository
git add <THE FILES YOU WANT TO ADD>
git commit -m "<SOME DISCRIPTION>"
git push --set-upstream origin <NEW_PROJECT>

This will crate a new branch on your repository which you can now view on you page https://github.com/<YOUR_GITHUB_ACCOUNT>/thesis_plots

Changing the setup

In the file main.py you can customize the default style of plots that are set by the function setup_plt. The documentation on this is on Matplotlib.

Example usage

Imagine that you made a new notebook with the following code in the notebook notebooks/normal_distribution.ipynb:

import matplotlib.pyplot as plt
import numpy as np
from matplotlib.colors import LogNorm
import thesis_plots
distribution = np.random.normal(loc = 1, scale = 2, size = (2, 100_000))

# Load your plotting style
thesis_plots.setup_plt()

# Make the plot
plt.hist2d(*distribution, 
           norm=LogNorm(), 
           bins=25,
           cmap='custom_map');
plt.colorbar(label=thesis_plots.mathrm('Counts/bin'))
plt.xlabel(thesis_plots.mathrm('X (some unit)'))
plt.ylabel(thesis_plots.mathrm('Y (some unit)'))
plt.title(thesis_plots.mathrm('Normal distribution'))

You can upload this notebook directly (following the commands above) or you can create a new module:

mkdir thesis_plots/normal_distribution
touch thesis_plots/normal_distribution/normal_distribution_plot.py
touch thesis_plots/normal_distribution/__init__.py

In thesis_plots/__init__.py you would add a line from .normal_distribution import *. In touch thesis_plots/normal_distribution/__init__.py you will import the normal_distribution_plot.py file similar to

from . import normal_distribution_plot
from .normal_distribution_plot import *

And in thesis_plots/normal_distribution/normal_distribution_plot.py you would make a function/class for the code you just wrote

import matplotlib.pyplot as plt
import numpy as np
from matplotlib.colors import LogNorm
import thesis_plots

def plot_normal_distribution():
    distribution = np.random.normal(loc = 1, scale = 2, size = (2, 100_000))
    
    # Make the plot
    plt.hist2d(*distribution, 
               norm=LogNorm(), 
               bins=25,
               cmap='custom_map');
    plt.colorbar(label=thesis_plots.mathrm('Counts/bin'))
    plt.xlabel(thesis_plots.mathrm('X (some unit)'))
    plt.ylabel(thesis_plots.mathrm('Y (some unit)'))
    plt.title(thesis_plots.mathrm('Normal distribution'))

This greatly simplifies your notebook, where you can now reduce the code to:

import thesis_plots
# Load your plotting style
thesis_plots.setup_plt()
thesis_plots.normal_distribution.plot_normal_distribution()

Advanced features

• Every time you make (or merge) a pull request, you will test making the plots on github actions. This prevents you from writing buggy code or forgetting to upload a data file.
• If you want to add data files, you can for instance do that in data, see the Lambda CDM example.
• You can setup a coverage report on https://coveralls.io/github/<YOUR_GITHUB_USER>/thesis_plots this allows you to view if all your code is actually used (and therefore useful)
• You can add tests in the tests folder, these are run as well when you commit to github.