Temporal Mapper

This repository hosts code related to the Temporal Mapper project written by Mengsen Zhang. The associated paper is the following:

Zhang, M., Chowdhury, S., & Saggar, M. (2022). Temporal Mapper: transition networks in simulated and real neural dynamics. Network Neuroscience.

Overview

Here's a schematic (Fig. 3 from the paper) showing how Temporal Mapper operates on a dataset.

Interface

To run the code, you only need a (time_frames x nodes) matrix! Try starting with the code provided in run_sim.m. You should get the following image:

When running with your own data, you only need to supply two parameters:

• k: locality parameter for kNN graph construction
• delta: distance parameter for performing asymmetric graph compression

Generating simulated data

We have provided some helper files to help you generate additional trials of simulated data. See the script in code/gen_sim.m and also the original paper for more details. If everything runs fine, you should see the following image:

Dependencies

The core Temporal Mapper code has been tested on Matlab R2020b.

Extras

To facilitate comparison of transition networks of different sizes, we also provide a GPU implementation of a variant of the Gromov-Wasserstein distance. Please see the readme inside the GWTLBgpu folder for more details.