Wong Wang working example to produce FCD matrix

[Ziaeemehr]

Dear developers,

I am trying to run Wong Wang model, original and reduced version on ds.Cmat network and my purpose is to get proper FC and FCD matrices, but I could not find a suitable configuration for the model parameters that does not give nan. I tried also putting k_gl the global coupling to zero, still getting nan. Have you ever tried to get FCD from WW model?
I also could not figure out how to get the reduced version, apparently the
from neurolib.models.ww.model import WWModel return the original version,
could you please guide?

Cheers
A

[HussainAther]

Hi @Ziaeemehr, great question!

It sounds like you're encountering instability or parameter sensitivity in the Wong-Wang (WW) model. A few things that might help. .. just a few things...

NaN Values: These often come from numerical instabilities. Have you tried reducing the integration step size (dt)? Also, setting initial conditions explicitly (e.g., small non-zero values) can sometimes stabilize the integration.
Zero Global Coupling: With k_gl = 0, the WW nodes decouple, so FCD/FC matrices may not yield meaningful network dynamics. You might try small non-zero values (e.g., 0.05) and gradually increase to identify a working range.
Reduced WW Model: You’re right — importing WWModel from neurolib.models.ww.model currently gives the original. A reduced version (if available) might live under a different submodule or branch. If it's undocumented, it may not be implemented yet. You could tag the maintainers (@caglorithm or @lenasal) for clarification.
For FCD: If you get stable simulations, try checking model.BOLD.BOLD output and compute sliding-window FC matrices over time. The utils.func.fc and func.fcd methods may help.