📐rt-mixed-report

The groupant beside tools/roundtrip/src/mixed/report.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

Components

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.

Parts

• rt-mixed-report-metrics

Model text

descriptant

Descriptant

Prospective epoch

2026-10-31 source review predicted: bits=sum(non-Meta selected residual bits)/sampled items, NaN when items=0. chrf=100sum/sents or NaN. acc uses token-confusion denominator separately from route denominator max(1). paired computes per-item occurrence-normalized means on shared item keys; TSV writes every residual class. Resource model: For I run/base item entries, C total cells, Q distinct layer/class keys and A varieties, the leak section first collects keys by scanning item maps, then invokes paired once per key per variety and stream. A conservative bound is O(AQ*(I log(I+1)+C) + K log(K+1)), including base item lookups and sorting K emitted leak rows; key-set insertion adds O(A*C log(Q+1)). The current finite Layer/Class universe bounds Q by a source constant, but repeated item/cell scans remain explicit. Other report sections scan cells/confusion maps and perform ordered lookups; output memory O(rendered bytes+K+Q). External prospective.md predates test additions and preserves APIs, effect order, hazards and interaction hypotheses.

Observed epoch

Bounded real helper checks in tests.md constrain these models; syntax is a partial lexical projection with no macro expansion, name resolution or proof. Production expressions unchanged; six inline test modules added. Compiled mutants are temporary and exactly restored. Existing malformed-input and nonfinite domains remain explicit, not silently repaired. Parent composes this scope with siblings without adding overlapping costs twice.

Finite synthetic helper evidence only; live corpora, parser/model loading, full orchestration and filesystem failure races excluded.

approximant

Forward relation

bits=sum(non-Meta selected residual bits)/sampled items, NaN when items=0. chrf=100*sum/sents or NaN. acc uses token-confusion denominator separately from route denominator max(1). paired computes per-item occurrence-normalized means on shared item keys; TSV writes every residual class.

Domain: Finite metrics and valid ctx item indices for paired/items_tsv; counts fit u64. NaN denotes absence, not zero quality. mono folds all mono-prefix runs of chosen mode without filtering link mode.

invariant

Invariant and cost

Invariant

Finite metrics and valid ctx item indices for paired/items_tsv; counts fit u64. NaN denotes absence, not zero quality. mono folds all mono-prefix runs of chosen mode without filtering link mode.

Cost

For I run/base item entries, C total cells, Q distinct layer/class keys and A varieties, the leak section first collects keys by scanning item maps, then invokes paired once per key per variety and stream. A conservative bound is O(AQ(I log(I+1)+C) + K log(K+1)), including base item lookups and sorting K emitted leak rows; key-set insertion adds O(A*C log(Q+1)). The current finite Layer/Class universe bounds Q by a source constant, but repeated item/cell scans remain explicit. Other report sections scan cells/confusion maps and perform ordered lookups; output memory O(rendered bytes+K+Q).

Limits

Finite synthetic helper evidence only; live corpora, parser/model loading, full orchestration and filesystem failure races excluded. No percentage surrogate is implemented: categorical parsing/effect observations lack an admitted numerical error metric. Any future profile must define its metric/domain and use only1/2/5/10 percent caps.

composition

Local composition

bits=sum(non-Meta selected residual bits)/sampled items, NaN when items=0. chrf=100*sum/sents or NaN. acc uses token-confusion denominator separately from route denominator max(1). paired computes per-item occurrence-normalized means on shared item keys; TSV writes every residual class.

Children select narrow source functions with their exact byte intervals. Remaining branches stay in the file model; fragments are not exhaustive partitions. Calls and state joins follow the forward relation, not folder order. Shared scans/helpers are counted once in whole-file costs. Complement of composition is not union of complements; reverse composition retains compatible intermediates and caller premises.

models

Typed source-bound relations

Subject: exact source SHA-256 in manifest. Domain and resource premises are in invariant.md. Each model is partial, source-derived, over a single helper invocation or the described finite trace. Complements are proposed inner exclusions under those premises; reverse relations retain all compatible antecedents, with no unique inverse claim. The oracle is the named checks in tests.md; unexecuted cases remain hypotheses. No numerical approximation certificate.

behavior_complement

Excluded: Meta contributes to bits() denominator-normalized numerator; absent variety returns numeric zero. TSV intentionally does include Meta.

reverse_reconstruction

Displayed rounded means admit many cell/count inputs; NaN fiber includes missing or zero counts. Paired deltas cannot uniquely reconstruct per-item residuals.

denominator_domain

Items, parsed sentences, language tokens and route tokens are distinct denominators.

rounding_fiber

One decimal TSV/display loses subdecimal differences; f only special-cases NaN, not infinity.

paired_intersection

Only item keys in both runs of requested variety contribute; occurrence divisor clamps at1.

schema_separation

Mixed TSV is nine columns stream/mode/link/...; compare expects its own residual/normal schemas, no direct adapter exists.

tests

Checks and evidence

Named source/helper checks: mixed::report::wave32_tests::denominators_and_meta.

Wave32 on2026-10-31:10 distinct exact tests passed twice debug and twice release, with equal sorted named outcomes. This node claims only its listed subset; empty subset means source review only. Five actual source-behavior mutants compiled, failed assertions, restored exact bytes/hash, and passed focused green reruns. No assertion-only mutants; existing swapped-input controls are separate. Receipt bundle: $MOVA_DATA/runs/octopus-wave32-rt_mixed-2026-10-31 (native-commands.json, source-test-map.json, mutation-results.json, .status/.log). No live corpus, teacher, frozen assessment, download or production data used.

Finite synthetic helper evidence only; live corpora, parser/model loading, full orchestration and filesystem failure races excluded.

This page in Markdown → https://md.groupants.elamur.ai/free/en/mova/tools/roundtrip/src/mixed/report.rs.groupant/