An alternative approach to building degenerate parameter sets allows the effect of homogeneous scaling to be separated from variability in conductance ratios. A) Scatter matrices of custom generated populations in the conductance spaces for the STG model (left) and the DA model (right) along with the directions of PC1 and PC2. The 2D subspaces shown here do not represent all conductances of the models and are randomly chosen. All conductances are expressed in mS/cm^2. The bottom left corner of each 2D subspace represents the origin of the conductance space. For the STG model, $g¯A$ ranges up to 600, $g¯CaS$ to 50, $g¯H$ to 0.7 and $g¯Na$ to 8,000. For the DA model, $g¯CaN$ ranges up to 0.12, $g¯Kd$ to 20, $g¯ERG$ to 0.25 and $g¯CaL$ to 0.1. The dotted line on the voltage traces corresponds to the 0mV line. B) Scatter matrices of custom generated populations in the conductance spaces for the STG model (left) and the DA model (right), isolating the effects of homogeneous scaling only (resp. variability in conductance ratios) shown as triangles (resp. crosses). The 2D subspaces are the same as in a). All conductances are expressed in mS/cm^2. The bottom left corner of each 2D subspace represents the origin of the conductance space. For the STG model, $g¯A$ ranges up to 600, $g¯CaS$ to 50, $g¯H$ to 0.7 and $g¯Na$ to 8,000. For the DA model, $g¯CaN$ ranges up to 0.12, $g¯Kd$ to 20, $g¯ERG$ to 0.25 and $g¯CaL$ to 0.1.