Elamur Groupantsworld-retention-probe
The groupant beside world/examples/retention_probe.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_retention_probe_invariants | world_retention_probe_invariants | partial source derived model | world_retention_probe_invariants |
world_retention_probe_resource_cost | world_retention_probe_resource_cost | partial source derived model | world_retention_probe_resource_cost |
world_retention_probe_representation_join | world_retention_probe_representation_join | partial source derived model | world_retention_probe_representation_join |
world_retention_probe_domain_partition | world_retention_probe_domain_partition | partial source derived model | world_retention_probe_domain_partition |
Model text
descriptant
world-retention-probe
main gathers args then probe requires4. Read/parse train then exam, Retained::new(train.course), train(false), live exam, exclusive save, load, cold exam. Delete seconds only from reports, assert JSON equality, blank exam, print timings/no_training_right/checkpoint bytes.
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. Two early-failure tests only; successful retention path and process isolation remain unexecuted/delegated; no surrogate.
approximant
Forward relation
main gathers args then probe requires4. Read/parse train then exam, Retained::new(train.course), train(false), live exam, exclusive save, load, cold exam. Delete seconds only from reports, assert JSON equality, blank exam, print timings/no_training_right/checkpoint bytes.
Domain: Tests bad arity and empty synthetic train with missing exam/state in tempfile. No train/load/exam success or school/model effects executed.
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
Material parse order preserved by main->probe extraction; errors propagate and prior effects may remain. Retained owns create_new/checksum semantics; success stdout after equality/blank exam/metadata.
Cost component: cost.md.
cost
Two material reads+train+3exams+save/load+metadata, sizes unbounded. Durations seconds, checkpoint bytes; no measured performance claim.
composition
knowledge::retention::Retained owns graph semantics; removing seconds is report projection not graph equality proof. This file delegates all successful retention mechanics.
models
behavior_complement
Bad arity cannot read train; malformed train cannot read exam/write state. Save precedes equality so mismatch can leave checkpoint. Equality ignores seconds only; one-process diagnostic does not establish separate-process isolation.
reverse_reconstruction
Success admits all train histories with same retained graph/report and arbitrary timing values. Early parse failure admits any absent exam/state paths; stdout cannot recover transient learner cache or original material bytes.
world_retention_probe_invariants
Material parse order preserved by main->probe extraction; errors propagate and prior effects may remain. Retained owns create_new/checksum semantics; success stdout after equality/blank exam/metadata.
world_retention_probe_resource_cost
Two material reads+train+3exams+save/load+metadata, sizes unbounded. Durations seconds, checkpoint bytes; no measured performance claim.
world_retention_probe_representation_join
knowledge::retention::Retained owns graph semantics; removing seconds is report projection not graph equality proof. This file delegates all successful retention mechanics.
world_retention_probe_domain_partition
Tests bad arity and empty synthetic train with missing exam/state in tempfile. No train/load/exam success or school/model effects executed.
tests
Finite actual-code evidence — 2026-10-31
Names: ["wave29_worldlearn::arity_gate_precedes_input_reads", "wave29_worldlearn::malformed_training_precedes_missing_exam_and_state_write"]
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-probe: 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.
Two early-failure tests only; successful retention path and process isolation remain unexecuted/delegated; no surrogate.
Source SHA256: 0c83df4c221af39fa86e068aee0ca4e1688efd03062a2551195e154495732cb3. 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/examples/retention_probe.rs.groupant/