A matrix capacitance correction reduced low-frequency transfer error in passive models of two partial adult worm electrical maps

We tested frequency transfer after removal of 170 neurons from passive electrical models of two partial adult worm maps. Ten named locomotion interneurons were retained. Unit leak and capacitance and count-weighted symmetric conductance were assumed. At the primary coupling scale and dimensionless angular frequency 0.1, a matrix capacitance correction reduced relative transfer error from 0.6215% to 0.0322% in adult 7 and from 0.5393% to 0.0295% in adult 8. A diagonal correction left errors of 0.2989% and 0.2390%. The matrix correction matched the first frequency derivative, but did not give a uniformly wider 5% error range than the diagonal correction. It became less accurate than original capacitance at high frequency. Full-system and independent spectral checks agreed. This is a reproducible model benchmark, not a measured activity result. Partial electrical annotation, assumed cell parameters, one retained set, and two maps limit biological interpretation.

C. elegans · electrical circuit · Kron reduction · hidden neurons · frequency response · passive model · model reduction