Elamur Groupantsmlab-numerics-poly-convolution
The groupant beside mlab/src/builtins/poly.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 |
product_order | product_order | partial source derived model | product_order |
division_fixture | division_fixture | partial source derived model | division_fixture |
short_dividend | short_dividend | partial source derived model | short_dividend |
zero_divisor | zero_divisor | partial source derived model | zero_divisor |
Model text
descriptant
mlab-numerics-poly-convolution
conv flattens nonempty inputs, accumulates complex products into length P+Q-1 and uses first-input orientation. deconv requires nonempty divisor with nonzero first re; long-divides real coefficients, explicitly zeroing eliminated remainder positions.
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
conv flattens nonempty inputs, accumulates complex products into length P+Q-1 and uses first-input orientation. deconv requires nonempty divisor with nonzero first re; long-divides real coefficients, explicitly zeroing eliminated remainder positions.
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
Output orientation follows first input for conv, dividend for deconv. Remainder remains dividend length in long-division branch.
Cost
O(PQ) conv; O((Y-D+1)D) division; linear output and remainder storage.
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-poly-convolution
behavior_complement
conv excludes empty success but does not validate vector-only shapes despite error text. deconv excludes imaginary participation except short-dividend clone; neither guarantees no overflow.
reverse_reconstruction
For real divisor d with leading nonzero, (q,r) reconstructs y=conv(d,q)+r with leading padding to dividend length; floating rounding prevents exact general identity. Different divisor/quotient factorizations share convolution.
product_order
conv [1,-1],[1,-2] gives [1,-3,2].
division_fixture
[1,-3,2]/[1,-1] yields q[1,-2] and r[0,0,0].
short_dividend
Y shorter than D returns scalar0 quotient and original cloned Y if remainder requested.
zero_divisor
D empty or D.re[0]==0 fails even if imaginary first coefficient is nonzero.
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/poly.rs.groupant/poly-convolution.groupant/