Elamur Groupantslearning-memo-execution-memo-policy
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 |
weight-subspace-quotient | weight_subspace_quotient | partial source derived | Specialized relation |
Model text
model
execution/memo-policy fragment
Source coverage: UTF-8 bytes [886, 1702) in learning.rs at the manifest digest. Memo enum and remember/valid implementation. This is a source-derived partial model, not a new runtime unit.
Forward approximant
remember yields Inert when !understood, count exists, or correct option has no best support; otherwise Scored stores the Answer and its current ask block. valid maps Empty to false, Inert to true, and Scored to exact equality of the stored/current selected block.
Invariant
A supported currently-correct answer is still Scored. Inert eligibility is structural under the fixed-world assumption. Equality of a block is a validity predicate, not a proof about every dependency of an arbitrary evaluator.
Cost approximation
Scored remember copies W weights and valid compares up to W entries; Empty/Inert operations are O(1). Answer and Vec allocations contribute storage.
Behavior complement
Let U be normal remember/valid input-output observations with valid correct-option indices and readable weight blocks, and R the actual relation restricted to U. The candidate inner approximant C_hat to U \ R contains valid(Empty)=true, valid(Scored)=true for unequal selected blocks, or remember classifying a supported understood non-count answer as Inert. 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(Inert) includes all answers satisfying at least one eligibility disjunct, without selecting which disjunct held. B(valid=true) includes Inert and matching Scored states. A boolean cannot reconstruct the answer, world or weight history.
Specialized relation
For a fixed ask, w~v iff their VIAS.len()*NF selected blocks agree. Scored validity is constant on these classes; answer equivalence additionally requires evaluator dependence only on that block and fixed world/options.
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 changed_winner_invalidates_but_another_weight_block_does_not probes an unchanged selected block, a changed winner and the stale cached-answer negative control. This is a concrete fixture, not a universal dependency proof.
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/memo-policy.groupant/