Elamur Groupantslearning-memo-execution-ordered-visits
The groupant beside world/src/knowledge/learning.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 | model.md |
| ast | partial | ../../syntax.json |
| descriptant | present | model.md |
| invariant | partial | model.md |
| tests | references only | 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 |
|---|---|---|---|
excluded-traces | behavior_complement | partial source derived | Behavior complement |
compatible-antecedents | reverse_reconstruction | partial source derived | Reverse reconstruction |
trace-correspondence | trace_correspondence | partial source derived | Specialized relation |
Model text
model
execution/ordered-visits fragment
Source coverage: UTF-8 bytes [2236, 4310) in learning.rs at the manifest digest. learn_lesson_with function, excluding the surrounding impl closing brace. This is a source-derived partial model, not a new runtime unit.
Forward approximant
Return default work for P=0. Otherwise shuffle options once, allocate Empty memos, and traverse P passes in exercise order. A valid Inert increments inert only; valid Scored increments reused then learns; a miss evaluates, increments evaluated, snapshots before learning when reuse is enabled. Observe after every visit.
Invariant
Under normal return and nonoverflowing counts, evaluated+reused+inert=P*E. Each call begins with Empty slots; its nonempty first pass evaluates every exercise. Reusing an Answer never bypasses the scored learning update.
Cost approximation
O(E) preparation plus P*E validity/dispatch work and reached evaluator/learner costs; timing appends at most P values. E=0 still permits P timing samples.
Behavior complement
Let U be finite helper traces with fixed inputs, valid indices, normal completion and nonoverflowing counters, and R the actual relation restricted to U. The candidate inner approximant C_hat to U \ R contains traces whose work counters do not sum to PE, whose observer count differs from PE, or whose scored snapshot occurs after its associated learning update. This is an explicit excluded relation, not merely a list of negative tests; it does not enumerate the entire complement or prove completeness.
Reverse reconstruction
B(work,observed weights) contains all ordered visit branch sequences and input states compatible with those counts and vectors. Totals lose exercise attribution; P=0 and E=0 can both yield zero counters. Observations do not uniquely recover answers or lesson seeds.
Specialized relation
The target relation equates every ordered weight vector to reuse=false recomputation for the same exercise order. It requires fixed evaluator dependencies and that Inert answers cannot update reference state; equal final scores alone are weaker.
Epoch memory and limits
2026-10-29: decomposed existing source without moving code. Existing root observations and all twenty root obernants remain authoritative. Child anchors select source-bound partial syntax nodes; name resolution and macro expansion remain absent. No new runtime test or measured performance gain is claimed.
tests
Existing bounded oracles
Existing every_ordered_update_matches_recomputation_over_multiple_passes checks a nontrivial 30-visit trace. Existing bounded_cache_reference_traces_cover_weights_orders_seeds_and_passes compares ordered and final vectors over 270 cases, exercising pass counts, seeds, orders and initial weights.
Run the existing suite from the repository root:
cargo test -p world knowledge::learning::tests --lib --locked
Test definitions are in ../../../learning.rs. These are references to existing oracles; this fragment adds no executable test. Historical execution evidence is retained in ../../tests.md. The fixture's simplified evaluator does not establish full parser/inference correspondence.
This page in Markdown → https://md.groupants.elamur.ai/free/en/mova/world/src/knowledge/learning.rs.groupant/execution.groupant/ordered-visits.groupant/