The substrate remembers through an accumulating topological debt D, and a feedback multiplier sets how fast the debt is serviced each cycle. The original law ran away; v2.0 replaces it with a logistic parabola that rises, peaks at debt-50, and falls back — so the substrate self-limits.
Left: the feedback multiplier across the whole debt range — the old law (amber) versus the v2.0 logistic (green). Right: the topological debt D(t) accumulating under dD/dt = r · feedback_new(D). Set the extraction rate and run it.
The portable C11 model — debt, the logistic correction, and the ODE step. Cross-compiles for Cortex-M4 and RV32IMC via the Makefile.
The embedded fixed-point form — Q64 in unsigned __int128 — the same logistic feedback with no floating point.
make → memristive_arm.o + memristive_rv.o. ARM -Os -mcpu=cortex-m4 -mthumb; RISC-V -march=rv32imc.
The full React + Recharts operator dashboard (needs a bundler). This page is the dependency-free version of the same kernel.
A substrate kernel and dynamical model — a corrected feedback law with an accumulating memory term, plus its fixed-point and cross-compile scaffolding. The physical realization (Ag₂S memristor wells, the attractor array) is the gravity-processor application built on this substrate. The topological debt and tensor tags (T133 phase-shadow · T057 negative-evidence · T128 root) are STOICHEION constructs, not measured device data.