Elamur Groupantsprag-fbench-alignment
The groupant beside prag/src/fbench.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 |
|---|---|---|---|
prag-fbench-alignment-behavior-complement | behavior_complement | partial source model | behavior-complement |
prag-fbench-alignment-reverse-reconstruction | reverse_reconstruction | partial source model | reverse-reconstruction |
prag-fbench-alignment-specialization | fbench_alignment_projection | partial source model | specialization |
Model text
model
prag-fbench-alignment
forward
Eligibility: nonempty gold verb with matching first word OR at least two matching agent/patient/to/place entities. Weight=4head equality+full verb equality+2role matches+kind equality. Sort descending weight then ascending gold/system positions and greedily accept unused endpoints.
invariant
Partial injection, not maximum-weight matching. Entity same is nonempty ID intersection and is not transitive equivalence.
cost
O(GSC+P log P+P log(G+S)), O(P+G+S) memory.
behavior-complement
Reuse of either endpoint, eligibility from one role with different verb heads, and reverse positional tie preference are excluded. Universe: bounded supplied inputs and candidate observable outputs at this source revision. Partial source-derived inner exclusions with finite witness evidence; unsampled semantics remain unknown.
reverse-reconstruction
Mapping alone leaves many possible ranks and frames. A compatible rank trace must have each skipped earlier pair blocked by a used endpoint. Target is all x in the stated domain with F(x)=y, without unique recovery or an implemented complete solver.
specialization
Eligibility intervention from two roles to three compiles and fails alignment_ties_and_threshold.
composition
Children share source identity. Preserve frame indices through alignment/link mapping and batch references through prompt/gate numbering. Independent marginal inverses can lose shared correlations; overlap costs are counted once.
checks
Six fbench and five silver wave16_models tests: finite canonical role oracle, independent ten-role counts, old-code witnesses and compiled controls. See receipts at file root.
epoch-memory
2026-10-30: prospective inspection predicted predecessor defects and duplicate-edge recall inflation. Three old-code witnesses confirm these. Valid adjacency and unique links retain semantics. No score assertions or deduplication introduced. Historical metrics retain original semantics.
limitations
Partial syntax is not expanded semantic AST; macros opaque. Scoped models and finite observations, not full-file verification or proof. No live parser, teacher, corpus, filesystem aggregate, GPU, full course, speed or approximation-cap claim. External effects and resource-extreme malformed inputs remain unexecuted.
evidence
Evidence scope
See root receipts.md for source and raw-log hashes. Pure supplied fixtures only; effects are source inspected.
This page in Markdown → https://md.groupants.elamur.ai/free/en/mova/prag/src/fbench.rs.groupant/alignment.groupant/