Generate the figures in my paper Phantom oscillations in principal component analysis.

Standard scientific Python stack:

pip install numpy scipy scikit-learn matplotlib seaborn statsmodels scikit-image imageio

CanD for pretty scientific figures:

pip install cand

Colorednoise for easy 1/f noise simulations:

pip install colorednoise

Networkx for the branching manifold:

pip install networkx

Wbplot for rendering the MRI images:

pip install wbplot

I used Python 3.6 but it should work on a more modern version.

Run each Python script from the terminal. After running all scripts, edit the "REVERSE = False" line to be "REVERSE = True" to generate the remaining figures.

Missing data files can in theory be rendered by uncommenting the commented-out sections of each file, but that requires installing more software (specifically this for fMRI and this for NHP electrophysiology) so I uploaded the pre-generated data.

Data from Roitman and Shadlen (2002) is available from the authors.

All code copyright 2024 Max Shinn, available under the GNU GPLv3.

ames-task.png is copyright Ames and Churchland 2019 from their eLife paper.