Algorithm non-linear FC-to-SC completion is |
External input: empirical FC (FCemp) |
Output: non-linear virtual SC (SCMFM) |
Fixed parameters: FC* fitting quality (CCtarget), initial guess SC*(0), learning rate λ, noise level (σ), simulation time (T), range to scan Gstart ≤ G ≤ Gstop, range to scan τstart ≤ τ ≤ τstop, other frozen Wong-Wang neural mass parameters |
Begin |
1. FC*(0) = non-linear SC-to-FC completion starting from SC*(0) |
2. Dist = corr(FC*(0), FCemp) |
3. Iteration = 0 |
While (Dist ≤ CCtarget) |
Iteration = iteration + 1 |
SC*(iteration) = SC*(iteration – 1) + λ*(FC*(iteration) – FC*(iteration)) |
FC*(iteration) = non-linear SC-to-FC completion starting from SC*(iteration) |
Dist = corr(FC*(iteration), FCemp) |
End |
Return SCMFM = SC*(iteration) |
End |