Elamur Groupantsworld-rreport-sampling
The groupant beside world/src/rreport.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 |
|---|---|---|---|
world-rreport-sampling-behavior-complement | behavior_complement | partial source model | behavior-complement |
world-rreport-sampling-reverse-reconstruction | reverse_reconstruction | partial source model | reverse-reconstruction |
world-rreport-sampling-coupling | constraint_relation | partial source model | coupling |
world-rreport-sampling-observation-fiber | information_loss | partial source model | observation-fiber |
Model text
forward
sample groups known ATTRS, excludes exact (story,n), then stable-sorts salted wrapping FNV/finalizer keys and takes k per type. read_verdicts skips malformed records and overwrites duplicate identities. manual averages only rows with exact verdict identity matches.
invariant
Sample size per known type <=k and eligible count; duplicate input identities can survive and hash collisions retain input order. Verdict floats are permissive, including NaN/out-of-range, and repeated rows count repeatedly.
cost
N rows, G groups, V verdicts, total emitted bytes L: aggregation O(N log G + L), sampling O(N log N) with repeated hash-string work and seven attribute scans; manual O(N log V), formatting O(L). Answering and ROUGE dependency costs are separate. No measured speed claim.
composition
main.rs reader evaluation calls reader::read_all on qs, then rows(own,&qs,ans), report/tsv and optional sample/manual. read_all maps qs in unchanged order. Native sample->read_verdicts->manual/report joins preserve (story,n) but bypass read_all and nonempty rows scoring. Ordered positional answer identity is a caller obligation.
epoch-memory
2026-10-30 first source-bound epoch. Before repair, native invalid-score and cardinality assertions failed on old production behavior. Narrow input checks now prevent undefined NaN ordering and silent report undercount. Infinite parser scores remain supported. Empty conventions and malformed-metadata/repair limits are retained explicitly. No optimization or percentage-error surrogate is proposed.
checks
cargo test -p world --lib rreport::wave10_models -- --test-threads=1. See root evidence.md for raw evidence hashes, repeated profiles and compiled controls. Syntax is partial; macro bodies are opaque and semantic AST/name/type resolution is absent.
behavior-complement
Subject/revision: bound source hash and this scope. Domain/observation/operator: Exclude sampled identities in the exclusion set and contribution by unmatched rows to manual denominator. Equal n in different stories is not the same identity.
Status: partial source-derived relation; finite evidence only as named in checks. Assumptions: bounded counts without integer or floating arithmetic overflow, successful allocations and fixed dependencies. Unknown external behavior is not excluded. Reverse relations retain ambiguity; finite witnesses do not establish universal completeness.
reverse-reconstruction
Subject/revision: bound source hash and this scope. Domain/observation/operator: Sample/aggregate output retains every input multiset containing chosen rows and any omitted ineligible/lower-ranked rows. Overwritten verdict histories and unselected rows are not recoverable.
Status: partial source-derived relation; finite evidence only as named in checks. Assumptions: bounded counts without integer or floating arithmetic overflow, successful allocations and fixed dependencies. Unknown external behavior is not excluded. Reverse relations retain ambiguity; finite witnesses do not establish universal completeness.
coupling
Subject/revision: bound source hash and this scope. Domain/observation/operator: Two stories with the same question number retain distinct manual verdicts; supplied fixture yields 50% per metric and sentence count 2. Exclusion removes only one story's identity.
Status: partial source-derived relation; finite evidence only as named in checks. Assumptions: bounded counts without integer or floating arithmetic overflow, successful allocations and fixed dependencies. Unknown external behavior is not excluded. Reverse relations retain ambiguity; finite witnesses do not establish universal completeness.
observation-fiber
Subject/revision: bound source hash and this scope. Domain/observation/operator: For 20 unique clean identities and k=0..22 native checks enforce cardinality, uniqueness and deterministic order; these do not prove hash distribution or uniqueness for duplicate input identities.
Status: partial source-derived relation; finite evidence only as named in checks. Assumptions: bounded counts without integer or floating arithmetic overflow, successful allocations and fixed dependencies. Unknown external behavior is not excluded. Reverse relations retain ambiguity; finite witnesses do not establish universal completeness.
This page in Markdown → https://md.groupants.elamur.ai/free/en/mova/world/src/rreport.rs.groupant/sampling.groupant/