Companion repository for the paper of the same title.
Interested in trying this for your own system? Jump right to the demo folder and you'll have a fast example that's easy to modify to suit your approach.
Want to dig into how something was done in the paper? Read on for the organization of the code to reproduce the results in the paper, although most of the hierarchy is self-explanatory.
This folder contains a basic demo of the entire pipeline. This isn't included as a main result in the paper, but is described in the Supplemental Materials and provides a basic overview of all the parts needed if users would like to apply the method to their own dynamical systems. This shows an example of learning a trigonometric terms in a simple 2D system of ODEs.
This section contains the code to reproduce the figures and supplemental materials for the first section of the results (Rössler system). Includes Figure 2 and Supplemental Figures.
This section contains the code to reproduce the figures for the second section of the results (Petrzela-Polak circuit). Includes Figure 3 and Table 1.
This section contains the code to reproduce the figures for the third section of the results (deriving new terms for next-generation neural mass models. Reproduces the results for Figure 4.
This section contains the code to reproduce Figure 5. The data are not included as they are downloadable from CRCNS.org, and the Matlab scripts in this section can work with the downloaded files directly as the data files are all in Matlab. Once they exist they will generate the data used to plot the subpanels of Figure 5 in Julia.
Note: Make sure to fix paths throughout for local paths only!!
Still have questions? Open an issue here for us to discuss or email [email protected].