Elamur Groupantsworld-know
The groupant beside world/src/know.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.
| id | type | label | where |
|---|---|---|---|
behavior_complement | behavior_complement | partial source derived model | behavior_complement |
reverse_reconstruction | reverse_reconstruction | partial source derived model | reverse_reconstruction |
world_know_invariants | world_know_invariants | partial source derived model | world_know_invariants |
world_know_resource_cost | world_know_resource_cost | partial source derived model | world_know_resource_cost |
world_know_representation_join | world_know_representation_join | partial source derived model | world_know_representation_join |
world_know_domain_partition | world_know_domain_partition | partial source derived model | world_know_domain_partition |
Parts
Model text
descriptant
world-know
KnowRules::from_english overlays defaults linewise, filters invalid numeric weights. Knowledge indexes extracted facts after kind remapping and derive(original). about exact lowercase subject wins, else head-word fallback. order uses13weights iff length13: score=sum(features*w), descending then source byte length/lex; otherwise link/definition/opening/length/lex. Learn audits verb mappings before ideas; new idea needs gd>0,gg>0 and controls nonincreasing.
Modeling epoch 2026-10-31
Prospective source/branch/cost hypotheses were saved externally before source edits. This partial source-derived model retains observed boundaries rather than claiming complete verification. Serialization/rank fallback/derivation/inheritance runtime evidence only; extraction/questions/cache/learning effects partial. No general logic truth proof or surrogate.
approximant
Forward relation
KnowRules::from_english overlays defaults linewise, filters invalid numeric weights. Knowledge indexes extracted facts after kind remapping and derive(original). about exact lowercase subject wins, else head-word fallback. order uses13weights iff length13: score=sum(features*w), descending then source byte length/lex; otherwise link/definition/opening/length/lex. Learn audits verb mappings before ideas; new idea needs gd>0,gg>0 and controls nonincreasing.
Domain: Synthetic Fact/KnowRules tests only, no annotator/manual/qa corpus/current state/train calls. Finite UTF-8 and representable arithmetic.
No executable percent surrogate: these categorical/effect observations lack a useful approximation metric. Future supported caps1/2/5/10% need a defined domain/metric and real reference evidence; no error or speed bound is claimed.
invariant
BTree and explicit ties deterministic. CountIn unit subject/holder object, show ignores negation/frequency. derive instrument branch filters !negated but not Often::Never. Cache one Arc keyed exact rules, rebuilt under lock. Prepared lookup does not validate current stamp.
Cost component: cost.md.
cost
Fact index O(F log S); derivation indices and kind scans, up to12 per accepted definition (not global subject cap). Rank O(F log F*13); i64 score multiplication/sum unchecked. Parsing and learning delegated and unmeasured.
composition
family stores English brain section and consumes knowledge answers; worldbook perception classifies missed sentences via facts. anticipate_if_stale stamps rules+code but prepared reads current file without revalidation.
models
behavior_complement
Exact subject never augments fallback head matches. Missing English fields leave defaults. Bad weight length bypasses learned scoring. Inherit does not recursively consume output; kind needs3 eligible Has/Can, own verb blocks inherited ability. right_answer unknown/yes can be true; check_half owns answered filtering.
reverse_reconstruction
Valid generated English reverses clean rule fields, but arbitrary ignored prose shares defaults. Same answer admits all lower-ranked and unrelated facts. Derived cannot preimage includes matching negativeHas instrument/action pairs; duplicate premises and irrelevant facts unrecoverable.
world_know_invariants
BTree and explicit ties deterministic. CountIn unit subject/holder object, show ignores negation/frequency. derive instrument branch filters !negated but not Often::Never. Cache one Arc keyed exact rules, rebuilt under lock. Prepared lookup does not validate current stamp.
world_know_resource_cost
Fact index O(F log S); derivation indices and kind scans, up to12 per accepted definition (not global subject cap). Rank O(F log F*13); i64 score multiplication/sum unchecked. Parsing and learning delegated and unmeasured.
world_know_representation_join
family stores English brain section and consumes knowledge answers; worldbook perception classifies missed sentences via facts. anticipate_if_stale stamps rules+code but prepared reads current file without revalidation.
world_know_domain_partition
Synthetic Fact/KnowRules tests only, no annotator/manual/qa corpus/current state/train calls. Finite UTF-8 and representable arithmetic.
tests
Finite actual-code evidence — 2026-10-31
Names: ["know::wave29_worldlearn::derivation_requires_missing_instrument_and_deduplicates", "know::wave29_worldlearn::english_defaults_and_malformed_weights_fallback", "know::wave29_worldlearn::inheritance_threshold_and_own_verb_override"]
Selected CPU command: CARGO_BUILD_JOBS=2 CARGO_TARGET_DIR=ISOLATED_TARGET cargo test --offline --locked -p world --lib --example retention_probe [--release] wave29_worldlearn -- --test-threads=1
15 selected cases passed twice debug and twice release with identical sorted names/outcomes, exits0. control-know: compiled source mutation failed with exit101; source restored byte-for-byte and15positive tests passed. No live teacher, corpus, state activation, GPU, or full-school run. Temporary synthetic files use unique process-id paths and are removed after passing tests.
Serialization/rank fallback/derivation/inheritance runtime evidence only; extraction/questions/cache/learning effects partial. No general logic truth proof or surrogate.
Source SHA256: 6e24c7d3793f5e08860e5549345b328dfa3916c89259520194cf95ba5166417e. Syntax generated independently twice and compared; lexical only, no semantic proof or CUDA compilation. Strict whole-crate Clippy was not repeated merely to reproduce existing dependency/source lints; Cargo warnings retained in receipts.
This page in Markdown → https://md.groupants.elamur.ai/free/en/mova/world/src/know.rs.groupant/