Module aggregation 

Aggregation monoids and posting-list aggregate operators.

Aggregation functions form a monoid (Section 5.1, Paper 1) so a parallel executor can split the input, fold each shard with AggregationMonoid::accumulate, merge the partial states with AggregationMonoid::combine, and emit the final value via AggregationMonoid::finalize.

Concrete monoids

• CountMonoid – (0, +) over u64.
• SumMonoid – (0.0, +) over f64.
• AvgMonoid – ((0.0, 0), pair-wise +) over (sum, count).
• MinMonoid, MaxMonoid – (infinity, min) / (-infinity, max) over f64.
• QuantileMonoid – collect-then-finalize quantile estimator.

Operators

• AggregateOperator folds a posting list with the supplied monoid and emits a single-entry posting list whose payload carries _aggregate_field / _aggregate metadata.
• GroupByOperator groups by a field, folds a second field with the monoid per group, and emits one entry per group.

Structs

AggregateOperator

Apply a monoid over a single field across the rows the source operator produces. Emits a one-entry posting list whose payload carries _aggregate_field / _aggregate so downstream callers can pick the result by name.

AvgMonoid

AvgState

Pair-wise additive state used by AvgMonoid.

CountMonoid

GroupByOperator

Group documents by group_field and fold agg_field per group.

MaxMonoid

MinMonoid

QuantileMonoid

Quantile aggregation: collects observed values and computes the requested quantile at finalize. quantile = 0.5 is the median.

SumMonoid

Enums

AggState

Generic aggregation state. Concrete monoids consume / emit the variants they care about; a mismatched variant is an execution error rather than a silently accepted partial fold.

Traits

AggregationMonoid

Aggregation function with monoid structure for parallel decomposition. See Section 5.1, Paper 1.