Below a critical mass threshold M_c, the accumulated structure is inert — diffuse scatter
(sim06's null). Above M_c, the structure activates: it sources a transport field T
that vents pheromone away from saturated regions (the negative feedback sim06 lacked), producing
consolidation into a few large vented pillars and the crossing detector firing. The transition should
be sharp in M_c.
Prediction: a phase transition in M_c —
morphology (scatter → few pillars) and crossing coincide.
Simulation Grid
The mechanism (minimal lumped transport)
The ONLY addition to sim06 is the transport field T and its coupling to pheromone
P. Agents are unchanged (Grassé stigmergy). This isolates the
environmental-physics-coupling variable.
Tis sourced by structure:T += transport_gain · max(0, M − M_c). BelowM_c, no sourcing (inert — sim06 regime).Tdiffuses (toroidal) and decays.T → Pcoupling:P += transport_coupling · (T_neighbor_avg − T_local). This vents pheromone from high-T (saturated) structure to low-T (gaps/flanks) — the negative feedback sim06 lacked.
Sign correction: the DESIGN sketch wrote (T_local − T_neighbor_avg), which would increase P at structure (positive feedback). The prose ("saturated pillars shed their pheromone to their flanks") describes venting, so the implemented sign is (T_neighbor_avg − T_local).
Baseline vs Transport (default run)
Loading results.json…
Time series
Dashed lines mark the crossing detector thresholds (stability ≥ 0.90; pheromone ≥ 0.5; deposit-on-structure ≥ 0.60). The detector needs all three for ≥4 consecutive samples.
M_c sweep — the phase transition test
Sweep M_c from ∞ (never active) to 0.5 (almost always active). If H7's phase
transition exists, a sharp jump should appear in pillars (down) and stability (up) at some
M_c, with the crossing detector firing only above it.
Loading sweep_data.json… (run sweep_plot to generate)
Perturbation / self-repair (the circularity safeguard)
Damage 25% of the grid at 60% of steps; measure recovery (final / pre-perturb mass). The H7 acid test:
repair should succeed when T is active (above M_c) and fail when
T is suppressed — proving the crossing is emergent from the physics, not imposed by the rule.
Loading…
What the null means
M_c, higher coupling) → more, smaller, less-stable pillars, not fewer larger ones.
Stability falls monotonically (0.876 → 0.739) as M_c drops. The crossing never fires.
Lineage
Built on sim06_termite_mound (null result). Design grounded in King, Ocko & Mahadevan (PNAS 2015) and Ocko, Heyde & Mahadevan (PNAS 2019). See the environmental-physics-coupling concept.