Elamur Groupantsmlab-numerics-reduce-order
The groupant beside mlab/src/builtins/reduce.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 | approximant |
| ast | partial | ../syntax.json |
| cost | symbolic source derived | invariant |
| descriptant | present | descriptant |
| invariant | partial | invariant |
| tests | executed bounded checks | 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 |
|---|---|---|---|
behavior_complement | behavior_complement | partial source derived model | behavior_complement |
reverse_reconstruction | reverse_reconstruction | partial source derived model | reverse_reconstruction |
strict_tie | strict_tie | partial source derived model | strict_tie |
nan_sort | nan_sort | partial source derived model | nan_sort |
binary_dispatch | binary_dispatch | partial source derived model | binary_dispatch |
complex_key | complex_key | partial source derived model | complex_key |
Model text
descriptant
mlab-numerics-reduce-order
better ignores NaN best when candidate finite and uses strict >/<. minmax reduction scans slices in order; complex keys are modulus but returned cell keeps components. sort computes stable local index permutation by real/modulus keys, with NaNs last ascending and first descending.
Modeling epoch 2026-10-31
Prospective source reading preceded inline tests. This node records implementation behavior, not a complete MATLAB compatibility contract. Models retain limits discovered in source reading, including real-only projections and warning-only result gates. Synthetic CPU fixtures are independent of frozen suites. No external effects, teacher, corpora or benchmark main are executed.
approximant
Forward relation
better ignores NaN best when candidate finite and uses strict >/<. minmax reduction scans slices in order; complex keys are modulus but returned cell keeps components. sort computes stable local index permutation by real/modulus keys, with NaNs last ascending and first descending.
Executable Rust surrogate gap: no calibrated replacement is attached. Exact test calls are evidence for the reference, not a fast approximant. Numerical approximants would need a named domain, zero convention and metric at supported caps 1/2/5/10%; categorical errors/effects have no percent surrogate here.
invariant
Invariants and cost
Coordinates are1-based local slice positions; sort preserves Class and complex storage, minmax casts Char/Logical result class to Double.
Cost
O(N+S) extrema; O(N log(max(L,2))+S) sort and O(N+S) index/scratch memory, including headers for empty slices.
Allocation failure, unchecked usize dimension arithmetic and IEEE nonfinite propagation are outside finite fixture guarantees. No measured throughput or global accuracy claim.
models
Typed source models: mlab-numerics-reduce-order
behavior_complement
Ties exclude winner replacement; sort excludes independent real/imag sorting. Empty0x0 minmax returns two empty values regardless nargout, while other empty slice shapes are unchecked.
reverse_reconstruction
Extremum and position identify selected cell and constrain other keys; earlier equal keys are incompatible with later winner. Sorted values plus 1-based per-slice permutation recover original cells when the permutation is valid.
strict_tie
[NaN,4,4] max index2, never3.
nan_sort
Two NaNs compare equal in stable sort; descending moves NaNs before finite keys.
binary_dispatch
Exactly2 arguments can take elementwise max/min even for empty second Mat; three-argument empty placeholder uses dim reduction.
complex_key
Equal magnitudes retain first minmax cell without phase tiebreak; returned value may have different phase than another compatible winner.
tests
Evidence and limits
Selected wave30_tests exercise this source; fragment-specific limits are stated in its operators.
Native command specifications, numeric exits, profile-paired logs, compiled controls and exact-restoration receipts are retained externally in the wave30 numerics run. Tests exercise actual source, not a copied model. The selected finite fixtures do not prove full convergence, backend accuracy on all matrices, resource exhaustion or side-effectful callback behavior.
This page in Markdown → https://md.groupants.elamur.ai/free/en/mova/mlab/src/builtins/reduce.rs.groupant/reduce-order.groupant/