Elamur GroupantsโŒ•
โ˜ฐ

qagen-select-ranking

The groupant beside qagen/src/select.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

componentlabelwhere
approximantpartial source modelforward
astpartial../syntax.json
costpartial source modelcost
descriptantpartial source modelepoch-memory
invariantpartial source modelinvariant
testspartial source modelchecks

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.

idtypelabelwhere
qagen-select-ranking-complementbehavior_complementpartial source modelbehavior-complement
qagen-select-ranking-reversereverse_reconstructionpartial source modelreverse-reconstruction
qagen-select-ranking-specializationround_robin_prefixpartial source modelspecialization

Model text

forward

Group supplied pool indices by document; sort within each by (seeded key hash,index), label rank r; globally sort (r,document hash,index), take k. Sorting resolves randomized HashMap traversal.

behavior-complement

Exclude later per-document rank before an earlier available rank. No random iteration-order tie decides the result. Duplicate pool indices are outside the production pool construction invariant.

reverse-reconstruction

Output prefixes permit all compatible pools whose first k sorted rank triples are those indices. Unselected suffixes and hash collisions prevent unique reconstruction; fixed supplied all/pool/seed reconstruct the output exactly.

invariant

Source-bound operational relation over supplied finite inputs; deterministic observations under fixed order and seed. No corpus/distribution assumption. Integer quota arithmetic assumes 32/64-bit usize.

cost

P pool indices: O(P log P) sorting plus key hashing, O(P) temporary storage.

specialization

Type round_robin_prefix: scoped observation above, exact intended relation, no approximation-error budget asserted.

checks

selection_order_filter_skip_balance_and_rounds. See ../evidence.md for actual profiles and negative controls.

epoch-memory

2026-10-30: model written from the existing mechanism before numerical repair. Source hash "0c35caf4eda91c0a5d36bce728f3846b360b483a66fa4ca3c884964839cd2554". Finite witnesses constrain this relation, not a general proof.

This page in Markdown โ†’ https://md.groupants.elamur.ai/free/en/mova/qagen/src/select.rs.groupant/ranking.groupant/