Clamping Excitatory Firing Rate with Reduced Wong Wang Model

[Dominique Hughes]

Hi everyone,

I have seen in multiple papers that authors mention when they use the reduced Wong Wang model with coupled excitatory and inhibitory populations, they use a parameter, Ji, to clamp the excitatory firing rate at about ~3 Hz. When I code up the model outside of TVB, I can see that Ji does have a large influence on excitatory firing rate, but I am struggling to understand how they clamped the firing rate within TVB.

The reduced Wong Wang model tracks and gives data for the synaptic gating variable in the model. You can code what the excitatory firing rate would be from the synaptic gating variable at each time step, but only if you also had the values of the coupling function at each time step. 

I can see how I could backtrack to find an appropriate Ji for the firing rate, but only if TVB can give me the firing rate or the values of the coupling function. Does anyone know if I can set up my simulations so I can get either information? Or have insights/ideas as to how they clamped the firing rate in the papers (I can provide references if requested)?

Best,

Dominique Hughes

[WOODMAN Michael]

hi

there is an "afferent coupling" monitor which provides the coupling values at every time step liek other monitors.  there is also the possibility to directly clamp state variable values by constructing the integrator with values for clamped_state_variable_indices and clamped_state_variable_values.

cheers,

Marmaduke

[Dominique Hughes]

Hi,

Thank you that's very helpful. I am a bit confused about what I get back from the AfferantCoupling and Raw monitors. I have 96 nodes, and I am running a simulation with a period=2000 and a simulation_length=20000 as a trial. I was expecting the resulting coupling and raw data to have 10 data points given the period and simulation_length, but instead I get over a million data points back for each node? What exactly is recording and what information is it giving back to be that large of a size?

For context, my simulation code is:

sim = simulator.Simulator(
    model = models.ReducedWongWangExcInh(G=np.array([1.1]),J_i=np.array([2.5])),
    connectivity = sub,
    coupling = coupling.Linear(a=np.array([1.0])),
    integrator = integrators.HeunStochastic(noise=noise.Additive(nsig=np.array([1e-4]))),
    monitors = (monitors.Raw(period=2000),monitors.AfferentCoupling(variables_of_interest=np.array([0]),period=2000),),
    simulation_length = 20000,
)
sim.configure()
(time, dataRaw), (time, dataCoupling), = sim.run()

Thank you for your help!

Dominique Hughes

[Dominique Hughes]

Hi,

I just wanted to check in again about the AfferantCoupling monitor?  I have 96 nodes, and I am running a simulation with a period=2000 and a simulation_length=20000 as a trial. I was expecting the resulting coupling and raw data to have 10 data points given the period and simulation_length, but instead I get over a million data points back for each node? What exactly is recording and what information is it giving back to be that large of a size? How do I interpret/consolidate that information into the coupling data at each expected time point?

Best,

Dominique Hughes

[Spase Petkoski]

Dear Dominique,

As Marmaduke said, the monitor gives the afferent coupling at every time-step of the integration. 

I imagine that your period refers to the period of the BOLD monitor that you have. 

At the same time, the integration step that you use has to be much smaller, probably ~0.01ms if your numbers are correct, and that seems unnecessary small. 

Best,

Spase

--
You received this message because you are subscribed to the Google Groups "TVB Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to tvb-users+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/tvb-users/7c94d51a-3ede-4e63-902c-aa8625753cbfn%40googlegroups.com.

[WOODMAN Michael]

hi

the AfferentCoupling monitor is intended for debugging does not therefore respect the period argument.  There's a another version which does temporal averaging, which is perhaps what you want, named AfferentCouplingTemporalAverage.

Regarding dt, you don't specify and the default is ~0.012.  For RWW, dt=1.0 (at least) should be usable.

cheers,

Marmaduke

---

From: tvb-...@googlegroups.com <tvb-...@googlegroups.com> on behalf of Spase Petkoski <spase.p...@gmail.com>
Sent: Tuesday, October 1, 2024 09:00

> the monitor gives the afferent coupling at every time-step of the integration. 

I imagine that your period refers to the period of the BOLD monitor that you have. 

At the same time, the integration step that you use has to be much smaller, probably ~0.01ms if your numbers are correct, and that seems unnecessary small.