Skip to Content

Studies

Simulation experiments and studies curated in the ontology.

NeuroML Ex8: AdEx (bursting)

Reproduces NeuroML2 LEMS canonical example 8, adExIaFCell. Bursting variant (adExBurst2) simulated for 300ms at 0.025ms step. Uses constant I_ext=800 pA matching the pulse generator amplitude (0.8 nA).

EI Tuning: Functional Connectivity Fitting

Fitting functional connectivity using the two-population Reduced Wong-Wang model with explicit excitatory and inhibitory populations. Combines Feedback Inhibition Control (FIC) to maintain local E-I balance with EIB tuning to globally optimize network connectivity. Supports both iterative (FIC+EIB) and gradient-based optimization approaches.

TVBO Epileptor2D as LEMS

Export the 2D Epileptor model from the TVBO database as LEMS XML. Demonstrates ConditionalDerivedVariable for Piecewise expressions.

NeuroML: FitzHugh-Nagumo 1969

FitzHugh-Nagumo 1969 model matching NeuroML2 fitzHughNagumo1969Cell. Parameters from canonical NML2_AbstractCells.nml example. Uses TS=1ms (time_scale: ms) as defined in the ComponentType.

NeuroML Ex9: FitzHugh-Nagumo

Reproduces NeuroML2 LEMS canonical example 9 (LEMS_NML2_Ex9_FN.xml). Single FitzHugh-Nagumo cell with external input I=0.8, simulated for 200s at 0.01s step size. Initial conditions V=0, W=0 match the NeuroML2 default (no OnStart block).

TVBO Generic2dOscillator as LEMS

Export the full Generic2dOscillator model from the TVBO database as LEMS XML. This model generalises FitzHugh-Nagumo. All 13 parameters, coupling inputs, and 2 state variables (V, W) are exported.

Hopf Pareto: FC vs Frequency-Gradient (NSGA-II + parallel refinement)

Fit a delay-coupled SupHopf network to two competing targets — empirical FC and a spatial peak-frequency gradient — replicating tvboptim Hopf_Pareto_ParallelOpt. Stage 0: NSGA-II genetic pre-search over (G, a, log10 sigma). Stage 1: parallel gradient refinement from every Pareto seed. Byte-identical to the reference workflow.

NeuroML Ex0: Integrate-and-Fire (tau variant)

Reproduces NeuroML2 LEMS canonical example 0, iafTauCell variant. Single leaky integrate-and-fire cell with spike/reset event. No external input — fires rhythmically because leakReversal > thresh. Simulated for 300ms at 0.005ms step.

NeuroML Ex2: Izhikevich (tonic spiking)

Reproduces NeuroML2 LEMS canonical example 2, izhikevichCell (2003). Tonic spiking variant simulated for 200ms at 0.005ms step. Uses constant I_ext=14 (original has pulseGeneratorDL with 20ms delay).

JR Peak Frequency Optimization

Reproducing MEG Resting-State Frequency Gradients with Network Dynamics. This experiment fits region-specific Jansen-Rit parameters to reproduce the empirical frequency gradient from visual cortex (11 Hz) to association areas (7 Hz).

Jansen-Rit — bifurcation in external input p

Reproduce the bifurcation diagram from Jansen & Rit (1995) Figure 3. The external firing rate p (representing thalamic input) is the continuation parameter. Two fold bifurcations and a Hopf bifurcation delimit the oscillatory regime.

TVBO JansenRit as LEMS

Export the Jansen-Rit model from the TVBO database as LEMS XML. 6 state variables with derived variables.

QIF neural mass — bifurcation in η̄

Quadratic integrate-and-fire population model. Continuation of equilibria in the mean neural excitability η̄, reproducing the fold bifurcation diagram from Montbrió et al. (2015).

RWW Functional Connectivity Fitting

Fitting functional connectivity using the Reduced Wong-Wang model. Two-stage optimization: global parameters first, then regional heterogeneity. Based on tvboptim RWW.qmd tutorial.

RateML rwongwang ground truth

No description.

Bayesian Inference (Stimulation)

Recover stimulus amplitude and excitability I of a pulse-driven Generic2dOscillator from one noisy recording, via numpyro NUTS/MCMC. Three prior scenarios steer the posterior along the amplitude/excitability degeneracy ridge. Replicates tvboptim.

TBPTT Jansen-Rit FC Fitting

Truncated backprop-through-time for a chaotic Jansen-Rit network. Fits global coupling G to empirical FC; contrasts full-BPTT vs windowed (grad_horizon) gradients. Based on tvboptim TBPTT.qmd.

TVBO WilsonCowan as LEMS

Export the Wilson-Cowan model from the TVBO database as LEMS XML. Demonstrates derived variables and coupling inputs in LEMS.