Skip to content

Simulation Guide

James Daley edited this page Jun 26, 2026 · 6 revisions

Simulation and Evidence Guide

ASH simulations are deterministic reference tools for finite-state behavior. They are useful for checking implementation semantics, controlled ablations, and artifact generation. They are not external physical validation by themselves.

Generate Repository Evidence

python tools/generate_artifacts.py
python tools/run_proof_suite.py
python tools/check_generated_outputs.py . --include-manuscript

Run the Simulation Entrypoints

python simulation.py
python src/simulate.py --agents 1000 --ticks 250 --noise 0.01 --seed 20260624

Controlled Ablations

Tracked ablations compare:

Condition Purpose
Uniform starts Confirm uniform-hypercube baseline behavior
Zero starts Test convergence behavior from concentrated initial condition
ASH transforms Reference transform behavior
No transforms Control without code action
Random weight-four transforms Matched finite transform control
Bit-flip noise Noise sensitivity check

Interpreting Histograms

The proof certificate records that the binomial Hamming-weight envelope is the marginal of uniform occupancy. A bell-shaped Hamming histogram is therefore a finite-hypercube baseline unless a separate test proves stronger structure.

Sector-Mixing Reproduction

python tools/reproduce_sector_mixing.py --output-dir .
python -m pytest tests/test_sector_mixing.py -q

Background-Bridge Reproduction

python tools/run_background_bridge_validation.py
python -m pytest tests/test_background_bridge_remediation.py -q

Evidence Flow

flowchart LR
    A["Simulation command"] --> B["Data output"]
    B --> C["Figure output"]
    C --> D["Artifact manifest"]
    D --> E["Repository verification"]
Loading

Boundary

A simulation output can support implementation correctness only within its configured scope. It cannot replace proof obligations, external data likelihoods, or prediction-lock discipline.

Clone this wiki locally