A known uniform partial gap block fixed full-removal responses in linear models with exact maps, fixed properties and access to all changed endpoints

We tested if a measured partial loss of electrical coupling fixes the response after complete loss. In a linear circuit, exact maps fix this forecast under three conditions. The removed conductance fraction is known and uniform. Cell and recording properties stay fixed. All changed endpoints have input and output access. Other cells can stay hidden. We give a proof, a prediction tool and a deterministic error certificate. We used the corrected July 2020 Cook C. elegans neural contact support. It has 302 cells, 3,671 chemical directions and 1,091 electrical pairs. We chose stable models with varied capacitances, gains and chemical filters. The solver did not receive these parameters. Across 96 exact settings, the largest relative forecast error was 2.13e-12. Global removal had 298 changed endpoints. Removal of 32 internal pharyngeal gaps needed only 20 ports under the sufficient rule. Hiding a changed endpoint gave an exact 14.3% error in a passive two-cell example. In the chosen full model, 0.5% spectral-norm response errors gave median DC half-block forecast error of 20.58% for global removal. The pharyngeal error was 1.42%. Dose and gain changes also caused bias. Agreement of inverse slopes at two doses did not prove selectivity. The package keeps code, proofs, input tables, source hashes, model parameters, panels, error seeds, results, figures and repeat steps. Source checks and a clean numerical repeat passed. These are mathematical and synthetic method results. They are not measured conductances or animal intervention effects.

connectome · electrical synapse · gap junction · intervention prediction · partial block · hidden neurons · response matrix · error bound · Caenorhabditis elegans