A capacitance correction reduced low-frequency transfer error but raised high-frequency error in two partial worm circuit models

We tested static circuit reduction and a first-order capacitance correction in passive models of two partial adult worm electrical maps. The graphs had 145 and 147 neurons. Thirty random half-observed masks were tested per map. With unit capacitance, leak 0.1 and angular frequency 0.1 in model units, the correction reduced median full transfer error from 9.57% to 1.75% and from 8.61% to 1.82%. It improved all 60 masks. At frequency 1 it raised error in all 60 masks. We give a sufficient continuous frequency band for a chosen full-matrix error limit and prove a high-frequency failure limit. Full-network calculations and 100 random-network tests checked the code. Counts are not measured conductances. Partial annotation, passive dynamics, assumed cell parameters and arbitrary time units prevent a claim about measured neuronal responses. The package includes a proof, graph arrays, masks, source hashes, code and all 6600 model cases.

Kron reduction · hidden neurons · electrical connectome · capacitance · frequency response · error bound · C. elegans