Elamur Groupantswordgraph-rel
The groupant beside wordgraph/src/rel.rs in the Mova repository: what the code computes, what it never does, how to go back from its outputs, its invariants, cost and checks.
Facts
Implementation material. A Mova code model describes this code at one source revision: partial models, finite checks and their stated scope. It is not a reviewed general exposition, and its labels are shown as written, never raised.
Components
| component | label | where |
|---|---|---|
| Approximant | partial | Approximant |
| Ast | partial | syntax.json |
| Composition | partial | Composition |
| Cost | partial | Cost |
| Descriptant | present | Descriptant |
| Invariant | partial | Invariant |
| Tests | executed bounded checks | Tests |
Obernants
The typed relations around the forward model: the behaviour complement (what the code never does), the reverse reconstruction (every input compatible with an output) and specialised views.
| relation | type | label | where |
|---|---|---|---|
| Behavior complement | behavior complement | partial source derived model | Behavior complement |
| Reverse reconstruction | reverse reconstruction | partial source derived model | Reverse reconstruction |
Model text
Descriptant
Relation codes of the word graph: an 8-bit base relation (15 kinds) and a 24-bit preposition slot holding the en
lemma number of an obl/nmod case word (0 = none, 0xFFFFFF = outside the dictionary). Names print as obl:in.
Epoch 2026-11-06: first version; the dictionary has about 110 k lemmas, well under 2^24.
Approximant
code(b, p) = b << 24 | (p & 0xFFFFFF); base_of(c) = Base::from_u8(c >> 24); prep_of(c) = c & 0xFFFFFF.
parse accepts a bare base name (prep 0), or obl|nmod:<word> with the word looked up in en::dict, or :?.
name(code(b, p)) prints b, b:? or b:<lemma text>.
Invariant
Codes sort by base first, so all codes of one base are contiguous. parse(name(c)) = c for every code whose
preposition is 0, PREP_OTHER or a dictionary lemma whose text is unique. base_of(code(b, p)) = b for all b, p.
Cost
O(1) per code operation; parse/name do one dictionary hash probe and one allocation.
Composition
Used by extract (to emit codes), compile (bounds by base), graph (relation table, edges_base), eval and the
CLI. No state.
Tests
rel::tests::names_round_trip: every base round-trips; obl:in parses and prints; obj:in and unknown names are
refused (negative cases).
Behavior complement
Does not represent multiword prepositions («because of»: only the first case word), subtypes other than pass,
agent, poss, or prepositions on relations other than obl/nmod. Never panics; unknown bytes print rel#hex.
Reverse reconstruction
code is injective on (base, prep mod 2^24). From a code, base and preposition number are recovered exactly; the
preposition's surface form, case and any second case word are not. PREP_OTHER admits every out-of-dictionary word.
Provenance
This page in Markdown → https://md.groupants.elamur.ai/free/en/mova/wordgraph/src/rel.rs.groupant/