Elamur Groupantsqagen-lemma-coverage
The groupant beside qagen/src/lemma.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 source model | forward |
| ast | partial | ../syntax.json |
| composition | partial source model | composition |
| cost | partial source model | cost |
| descriptant | partial source model | epoch-memory |
| invariant | partial source model | invariant |
| tests | partial source model | checks |
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.
| id | type | label | where |
|---|---|---|---|
qagen-lemma-coverage-behavior-complement | behavior_complement | partial source model | behavior-complement |
qagen-lemma-coverage-reverse-reconstruction | reverse_reconstruction | partial source model | reverse-reconstruction |
qagen-lemma-coverage-zero-case | zero_case | partial source model | zero-case |
qagen-lemma-coverage-membership | membership | partial source model | membership |
qagen-lemma-coverage-threshold | threshold | partial source model | threshold |
qagen-lemma-coverage-error-erasure | error_erasure | partial source model | error-erasure |
Model text
forward
Count content answers, build set of all sentence form/lemma strings when needed, count hits against either answer field; validate and maximize per-anchor shares.
invariant
Each answer occurrence counts separately; sentence duplicates do not increase hits. Invalid anchor is any zero or value greater than sentence length.
behavior-complement
No implicit stemming or casefolding inside cover, no skipping duplicate callbacks, no last-anchor tie winner.
reverse-reconstruction
Coverage observes membership only: reorder sentence lemmas or add duplicate entries without changing result. Winner output admits many distinct subthreshold losing anchors.
cost
Two answer scans and one sentence-pool construction; no content-reference vector. Hash worst-case not claimed linear; allocation can fail.
composition
Checked anchors prove sents[k-1] safe. Strictly increasing best coverage preserves stable tie choice without sorting.
zero-case
No content -> None before sentence pool allocation; try_doubt still annotates all anchors on valid input for callback compatibility.
membership
hit(w) iff w.content and either form or lemma belongs to union of sentence forms and lemmas.
threshold
DOUBT=0.8 inclusive, compared with f64 quotient inherited from reference.
error-erasure
try_doubt retains invalid-anchor error context; doubt maps Err to None for compatibility with its Option return type.
checks
Synthetic embedded Rust tests, 1024 independent coverage cases, preflight sentinel artifacts, callback trace and compiled branch/arithmetic controls; debug and release twice, model-loading test explicitly skipped. See parent evidence.md for observed outcomes, not proof.
epoch-memory
Prospective model was saved before repair on 2026-10-30. Old source failed four library witnesses and one CLI witness. The tie control initially survived because [2,1,2] selects 2 for both first and last ties; the corrected [2,1] witness distinguishes them. Preserve this observation: duplicate-endpoint fixtures can hide tie regressions.
references
Original project implementation and rule v2: docs/context.md, qagen/src/gate.rs and qagen/src/run.rs. Model conventions: docs/groupants.md, docs/mathematical-modeling.md and docs/mathematical-modeling.md. No external solution code copied.
limitations
Finite local tests and partial syntax projections, no semantic AST or universal proof. Allocation failure and callback panic are outside successful-return model; filesystem races and write failures can interrupt publication. No live model/teacher/database/corpus used. No useful certified numeric surrogate is attached to these categorical/effectful interfaces; no invented error budget, substitution or speed claim.
This page in Markdown → https://md.groupants.elamur.ai/free/en/mova/qagen/src/lemma.rs.groupant/coverage.groupant/