Table 2 Optimal parameter points, goodness-of-fit and the used computation time for the results presented in Fig. 2.

From: Towards an efficient validation of dynamical whole-brain models

 

Optimal coupling C

Optimal delay τ

Optimal noise σ

Goodness-of-fit

Invested core-h

GS3D

0.2400

3.0000

0.8000

0.3310

1200.0

NMA3D

0.3287

0.2883

0.7079

0.3173

0020.2

PSO3D

0.1603

7.5195

0.5717

0.3489

1382.5

CMAES3D

0.1603

0.1554

0.5230

0.3479

0303.3

BO3D

0.2358

0.0042

0.6991

0.3298

0016.4

  1. The indicated parameter values (coupling, delay and noise) and the goodness-of-fit pertain to the best solution (highest goodness-of-fit) found across all 15 algorithm runs for the four optimization methods (NMA3D, PSO3D, CMAES3D, BO3D). For the grid search in 3Dim (GS3D), the best parameter constellation obtained from a single parameter space scan is shown. The invested core-hours (core-h) for GS3D and the optimization methods represent the required resources for one thorough scan and all 15 runs in total, respectively.
Back to article page