Abstract: |
We demonstrate a method to infer the topology of a hidden nonlinear network of cable-like links (elements that can sustain only tensile loads) given limited data on the hidden network. The hidden networks are based on two classical representations of the tendon network of a human finger, containing 11 and 16 links, respectively. The phenotype consists of a regular, mesh-like network of 72 links whose properties are modified by our evolutionary algorithm. Using 30 data points, randomly obtained from the hidden network, we evolved networks whose functional behavior reproduced that of the hidden network, and whose topology closely matched that of the hidden network. |
