Elamur Groupantsworld-tune-orchestration
The groupant beside world/src/tune.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 | bounded native and inspected | ../model.md#evidence |
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-tune-orchestration-behavior-complement | behavior_complement | partial source model | behavior-complement |
world-tune-orchestration-reverse-reconstruction | reverse_reconstruction | partial source model | reverse-reconstruction |
world-tune-orchestration-zip-projection | information_loss | partial source model | zip-projection |
world-tune-orchestration-effect-boundary | effect_partition | partial source model | effect-boundary |
Model text
Subject: world/src/tune.rs SHA-256 932ebce85aebb89467cb9da0d82b136debed8b77f6a9f0b8e852336480db0adb. All entries share this exact revision. Status: partial source-derived relations, finite native evidence for pure seams, source inspection for effects. Domain: finite supplied inputs under the stated resource/arithmetic premises. Exclusions are inner source-derived families, not a complete behavioral complement. Reverse relations retain compatible antecedents but no complete inverse solver is implemented.
forward
Open journal/code key before loop; score base; build/rank proposals; atomic index distributes candidate work to threads. Reuse journal or score/zip differences/bootstrap; append fresh entries, sort/report, and maybe adopt. run loads domains/expressions paths; domains adoption writes hot configuration.
behavior-complement
Universe: successful traces under admitted IO/callback/arithmetic premises. Exclude claiming one candidate index twice per generation by atomic fetch, changing shipped constants in tune, another generation with adopt=false, or run domains hot write with adopt=false. Callback internals remain unconstrained.
reverse-reconstruction
Final config admits different worker completion/journal orders and score tails ignored by zip. Same-size file revisions alias path/size data keys. Missing effect trace prevents persistence/provenance recovery.
invariant
Each index claimed once. Sorted observation order deterministic only for fixed callback outputs/indices. Journal append follows completion order. Sync does not imply callback purity. Length mismatch truncates and arithmetic overflow remains unchecked.
cost
Bounded by gens in generations only: no time, thread or dataset cap. Callback/bootstrap/IO costs unknown; journal scan repeatedly participates in sorting.
zip-projection
For vectors a,b, only indices below min lengths enter difference. Extra tails vanish. Equal length/aligned item identity and representable differences/sums are caller premises, not verified.
effect-boundary
Even gens=0 opens journal and computes code identity; adopt=false may still score and journal. Tests therefore call pure helpers, never presumed read-only tune/run. No live parser/teacher/corpus/bootstrap training/GPU executed.
composition
Root file model connects these inspected effects with tested pure neighbor/selection relations. world/main.rs passes file paths/gens/threads/adopt. Callback/source identity and cache freshness remain outside supplied-input evidence.
epoch-memory
Wave14 first modeling epoch, 2026-10-30: preserves rendering and tuning selection semantics. Two public pure tuning seams are now actual production-path calls. Six rendering tests and four tuning tests record field loss, tolerance/duplicate behavior, strict selection thresholds and domain limits. No bug repair or speed claim. Existing semantics retained include newline passthrough, first key/last MId collision, zip truncation and unchecked score arithmetic. Evidence is finite; no parser, teacher, filesystem corpus, journal-backed tune, optimizer training or GPU execution.
limitations
Partial syntax binds source bytes; macro bodies remain opaque and semantic AST is absent. Native/structural checks are not universal proofs or full-system coverage. No approximation error budget or automatic substitution is claimed. File evidence distinguishes executed pure paths and inspected effects.
This page in Markdown → https://md.groupants.elamur.ai/free/en/mova/world/src/tune.rs.groupant/orchestration.groupant/