Figure 5:

Stochastic structure distance profiles in the parameter space. (A) Three realizations of the distance hyper-surface projection along a dimensionless parameter λ of the SSM, controlling the apical dominance of a tree (the shown fragment of the projection with the step of 0.001 approximates 30% of the allowed variability of the parameter during optimization, which was in the range [0.0, 0.65]). (B) Structural distance values (with U = {S^0,1, B^2,3,4}) for 100 randomly generated SSM trees for each value of a discrete SSM parameter, i.e., number of growth iterations (red line connects the median points of the distance distributions for each parameter value; blue line shows the same median distance profile but for the distance with U = S^0,1, see (C)). (C) Same as in (B), but U = S^0,1 (blue line is the median profile; red line is from (B)). The SSM is the best-fit SSM obtained in the experimentation reported in Fig. 4; the black arrow indicates the parameter value of the best-fit SSM found in the experimentation.