Fig. 6

Comparison of optimization methods for NGF-induced ERK activation model. a Final objective function values (darker color in steady-state, lighter color not in steady-state), b comparison of convergence criteria with respect to steady-state constraint, c computation time for 100 runs and d average computation time per converged start of unconstrained optimization method (fmincon), constrained optimization algorithm (fmincon), the hybrid optimization method and the proposed simulation-based optimization methods with gradient descent and newton method based updates and e the best fit to the data are depicted