Phase 39 · self-propagating geometry corridor

Light one match, and watch it go out

AT-0313 → AT-0322. Deterministic toy benchmark seed 313: a seventy-site chain, sixty-nine edges, and a single bounded local trigger that was supposed to launch a corridor all the way to the far end without anyone pushing it. It did not. This page records the failure, the sweep that shows it is not one unlucky setting, and the two other media we tried instead.

The lesson before the numbers

A one-time local trigger did not launch a full corridor in the medium tested.

Plain language

Controls

SeedSeed 313 reproduces the recorded self-propagation benchmark.Sites — 70Places on the chain. Edges are one fewer.Excitation threshold — 0.12How hard the medium is to ignite locally. Lower is easier, and eventually unstable.Diffusion — 3.00How strongly a site pulls its neighbours along. Higher spreads further and thinner.Damping — 0.040Linear pull of the geometry field back to baseline.Recovery rate — 0.008How fast the refractory variable catches the front and extinguishes it.Trigger amplitude — 1.00Strength of the one-shot local trigger. Applied once, never again.Trigger width — 2.5 sitesSpatial half-width of the trigger. Wider is a bigger local intervention, not a remote one.Opening threshold — 0.20Compression a link must reach before the traveller may cross it.Horizon — 90 time unitsHow long we watch before declaring the pulse dead.Preparation price — 0.80/siteOne-time cost of making the route excitable at all. Never folded into the per-trip figure.Repeat pulses — 5How many pulses are fired through the same medium in the degradation test.

Running the self-propagating-corridor package…