Editing Openai/691c1dba-9228-800f-8463-13b3a9006306 (section)

==== 1. Pick integrator: start with small dt and explicit integrator to prototype, then switch to symplectic Verlet or implicit schemes for stiff coupling (exponentials and delta impulses create stiffness). ====
# Parameter sweep: run batches changing eta_ucf, eta_ts, gamma to see qualitative regimes: stable memory, oscillatory ritual, runaway transfer to geometry.
# Observability: log CompassionMetric(t) (e.g., inverse of |X-Y| weighted by S_ent) and create ingots when compassion crosses thresholds.
# HPL -> manifest tests: feed the JSON to your compiler.py and run with cli.py to verify step-by-step behavior; make steps idempotent and side-effect-free to satisfy ZCP guard.
# Make it symbolic: annotate each parameter with human semantic tags (e.g., eta_ucf: "memory_transduction_strength"), so manifest-to-LaTeX and manifest-to-paper mapping is automatic.

If you want, I can:
* Generate a ready-to-run .py file with the ODE simulation and plotting code (matplotlib) using the supplied skeleton and sensible defaults.
* Produce a runnable HPL .hpl file and a sample manifest in your manifests/ format so you can drop it into fenix_core.
* Create a short README that documents parameter meanings and experimental recipes.

Which of those should I produce next (one-file runnable script, HPL file, or README)?