examples

examples/ — a guided tour

Each script is runnable as uv run python examples/<name>.py and doubles as a self-checking assertion suite (every file ends with assert result is Answer.X). Read them in the order below.

Tier 1 — Showcase

Script	What it shows
nixon_diamond.py	Equi-specific defeasible rules leave the query UNDECIDED. Shows: Garcia & Simari 2004 Def 5.3 four-valued answer.
innocent_until_proven_guilty.py	Presumption (innocent -< true) overridden by evidence and attacked by a blocking defeater. Shows: presumptions=[...] + defeaters=[...].
reviewer_assignment.py	Multi-axis peer-review COI (coauthor / institution / advisor). An undercutting defeater waives only the institution axis; superiority keeps advisor-COI above the waiver. Shows: defeaters + incomparable disqualifiers — not scalar-reducible.
access_control_break_glass.py	Financial authorization where emergency overrides audit hold but is itself overridden by the self-dealing bar, and deliberately cannot trump four-eyes. Shows: partial-order superiority + UNDECIDED when policy is silent.
gdpr_lawful_basis.py	Two independent lawful-basis paths; withdrawal beats consent via explicit superiority=[("d2", "d0")]. Shows: user superiority layered on specificity.

Tier 2 — Domain depth

Script	What it shows
platypus.py	Strict rule wins over defeasible default — mammals don't fly, except when they do (no: platypuses still don't). Shows: strict-over-defeasible precedence.
clinical_drug_safety.py	Layered contraindications with explicit superiority. Shows: defeasible clinical reasoning.
looks_red_under_red_light.py	Pollock's perception defeater: looks_red defeasibly supports red, but red-light undercuts the perception rule. Shows: undercutting defeaters.
data_fusion_sources.py	Conflicting sensor readings resolved by source-preference superiority. Shows: data fusion with trust.
config_precedence.py	Site < project < user < flag cascade. Shows: precedence via specificity + superiority.
explanations_gallery.py	Six canonical cases each rendered with explain(tree, criterion). Shows: prose explanations (F-B).

Tier 3 — Engine breadth

Script	What it shows
git_ancestry.py	Recursive ancestor relation via SemiNaiveEvaluator. Shows: stratified Datalog.
klm_config_defaults.py	KLM rational/lexicographic/relevant closure on propositional defaults. Shows: ClosureEvaluator.
safe_vs_nemo.py	Same theory, two negation semantics, different answers. Shows: NegationSemantics.SAFE vs NegationSemantics.NEMO.

Tier 4 — Visuals

Script	What it shows
mermaid_gallery.py	Emits .mmd files for five dialectical trees via render_tree_mermaid (F-C).
mermaid/*.mmd	Rendered Mermaid diagrams: nixon_diamond, innocent_coerced, reviewer_waiver_overruled, break_glass_vs_four_eyes, clinical_aspirin, opus.