Elamur Groupantsstreaming-statistics
The groupant beside models/streaming-statistics.groupant 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 source model | forward |
| ast | not materialized | |
| composition | partial source model | composition |
| cost | partial source model | cost |
| descriptant | partial source model | epoch-memory |
| invariant | partial source model | invariant |
| tests | bounded executed | checks |
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 |
|---|---|---|---|
streaming-statistics-behavior-complement | behavior_complement | partial source model | behavior-complement |
streaming-statistics-reverse-reconstruction | reverse_reconstruction | partial source model | reverse-reconstruction |
streaming-statistics-shared-count | accepted_prefix_transport | partial source model | shared-count |
streaming-statistics-conservation | stream_conservation | partial source model | conservation |
streaming-statistics-summary-loss | information_loss_relation | partial source model | summary-loss |
Model text
model
streaming-statistics
forward
Summarize one accepted stream using bounded heavy-hitter counters, histogram rank/value intervals and online or merged moments.
invariant
Accepted prefix counts agree under shared legal updates; counter, bucket and moment statistics retain different conservation and error relations.
cost
Misra-Gries and Space-Saving O(k) worst update and O(k) state with deterministic BTreeMap order; histogram O(1) update/O(b) query/state, b<=10^6; moments O(1) update/merge/state.
composition
Only shared successfully accepted inputs yield common N. There is no transaction across separate components. Exact counts, sorted ranks and direct finite mean/variance serve as independent fixture observations. The harness tests MG=SS jointly on 800 updates; histogram and moments use separate fixtures. A shared accepted-prefix execution across all four is not exercised.
epoch-memory
Wave19, 2026-10-30. Models retain real-arithmetic assumptions separately from native finite checks. Kernels are library tools, not automatically adopted into training or inference.
checks
["merged_moments_join_compensated_centered_sums","frequency_summaries_join_literal_histogram_certificates","histogram_intervals_join_sorted_rank_and_exact_counts"]
limitations
Frequency error/N, quantile value/rank intervals and floating variance are different metrics. Moment arithmetic has no universal relative-error certificate. Map allocation failure lies outside the Result contract; histogram allocation is fallible.
catalog
Typed family relations
Each relation keeps its own domain and assurance.
behavior-complement
Type: behavior_complement.
Exclude frequency intervals omitting the exact accepted-stream count, rank/value rectangles omitting the sorted requested rank, mismatched accepted-prefix N and successful nonfinite moment output. A rejected individual update must not mutate that component; no cross-component rollback is promised.
reverse-reconstruction
Type: reverse_reconstruction.
Compatible streams lie in the intersection of frequency intervals, histogram bucket occupancies and moment constraints with execution consistency. Permutations and unrecorded identities/within-bucket values remain ambiguous; rounded moments may distinguish some permutations.
shared-count
Type: accepted_prefix_transport.
Under legal updates for every component, MG.len=SS.len=histogram.len=moments.count=N. Rejection in one component invalidates this common-prefix premise unless caller coordinates updates.
conservation
Type: stream_conservation.
MG counter sum+(k+1)*cancellations=N; SS estimate sum=N; histogram bucket sum=N; moments count=N. These equalities concern accepted updates and do not convert all output errors into one percentage.
summary-loss
Type: information_loss_relation.
Frequency summaries omit most identities; histogram retains only bucket order/value intervals; moments retain low-order statistics. Equal low-order moments do not establish equal distributions or equal quantiles.
wave19-evidence
Family evidence — 2026-10-30
This family binds eight implementation/test sources through its manifest; views are partial source-derived models. Selected named joins are ["merged_moments_join_compensated_centered_sums","frequency_summaries_join_literal_histogram_certificates","histogram_intervals_join_sorted_rank_and_exact_counts"] in algorithms/tests/composition_wave19.rs. Those checks pass four times as part of the 616-check suite. Leaf oracles and compiled production controls remain separately bound in adjacent evidence.
Integrated evidence records receipts; conditional laws outside these fixture scopes are not promoted to executed checks. No joint universal percentage bound follows from family membership.
This page in Markdown → https://md.groupants.elamur.ai/free/en/mova/models/streaming-statistics.groupant/