Mask lower triangle of a Representational Dissimilarity Matrix (RDM)

This recipe takes a Representational Dissimilarity Matrix (RDM) as a square numpy array, masks the diagonal and lower triangle, and outputs a flattened numpy array of the upper triangle.

Easily adaptable for Representational Similarity Analysis (RSA), Functional/Structural Connectivity analyses, or other analyses with related pipelines with symmetric, square matrices.

Requirements:

numpy>=1.18.1

1
2
3
4
5
6
7
def flattenRDM(square_matrix):
    assert square_matrix.shape[0] == square_matrix.shape[1], "Must be a square numpy array"
    
    # mask the diagonal and lower triangle and output flattened array
    flattened_out = square_matrix[np.triu_indices(len(square_matrix), k=1)] 

    return flattened_out

Example usage:

1
2
3
4
5
6
7
8
import numpy as np

square_matrix = np.array([[1,0,0,0],
                          [0,1,0,0],
                          [0,0,1,0],
                          [0,0,0,1]])

flattenRDM(square_matrix)

Recipe made by Shawn Rhoads

These recipes may also be of interest

Automatically Generated Recipe
python r bash
Explain briefly what the recipe does. Explain briefly how the recipe works. Provide a citation and link to documentation of any third party … Read More
Contributing to the Neuroimaging Cookbook
python markdown
The title entered above will automatically appear here, so just start by explaining the recipe purpose of your recipe. Here are some … Read More
Using 3dcalc to create a binary mask from a parcellation scheme
This recipe uses 3dcalc in AFNI to output a binarized 3D image file (nifti) of a region-of-interest in a parcellation scheme that is indexed … Read More

Mask lower triangle of a Representational Dissimilarity Matrix (RDM)

Example usage:

These recipes may also be of interest

Automatically Generated Recipe

Contributing to the Neuroimaging Cookbook

Using 3dcalc to create a binary mask from a parcellation scheme