Elamur Groupantsmath-eval-gate
The groupant beside math/src/eval.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 |
| descriptant | present | descriptant |
| invariant | partial | invariant |
| tests | 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 |
temperature_kind | temperature_kind | partial source derived model | temperature_kind |
failure_precedence | failure_precedence | partial source derived model | failure_precedence |
gate_boundary | gate_boundary | partial source derived model | gate_boundary |
empty_nonfinite | empty_nonfinite | partial source derived model | empty_nonfinite |
Model text
descriptant
Scoped gate fragment, source bytes [34232,35748).
Expr -> eval using Rat/units -> Q plus Ctx -> gate using fcheck. Raw eval success is insufficient: existing checks or SI mismatch can reject. Context mutations survive raw errors; answer owns fresh context. No global relative-error budget follows.
Inherited family observations are parent context; only the explicit byte range belongs to this fragment.
approximant
eval recursively produces Q and appends Ctx evidence. Exact pairs use Rat; mixed approximate pairs use f64. Aggregates convert to first operand units. answer uses fresh context then gate checks independent fcheck SI value and recorded booleans.
O(N) traversal plus numeric costs; aggregate O(k log k); exact probability enumeration O(T*2^T) for T<=16. Trace size includes formatted intermediates; stack/allocation unbounded. Gate tolerance is policy, not universal numerical assurance.
Executable approximant gap: no useful percentage surrogate for exact arithmetic and categorical branch contracts. Existing float pruning/tolerances do not establish an approximation guarantee.
invariant
Point+point and point scaling are excluded, dimension mismatch excludes conversion. Gate rejects any false check or SI disagreement; raw eval does not enforce gate. Empty Add/Mul indexing, nonfinite handling, unchecked count sum and exponent narrowing remain gaps.
Output Q admits many Expr/environment/trace antecedents and equivalent units. Fixed-scale exact conversion has unique inverse only when all arithmetic succeeds. Gate accepts SI-equivalent representations, not a unique syntax or provenance.
Cost: O(N) traversal plus numeric costs; aggregate O(k log k); exact probability enumeration O(T*2^T) for T<=16. Trace size includes formatted intermediates; stack/allocation unbounded. Gate tolerance is policy, not universal numerical assurance.
Finite fixtures do not prove all-domain correctness or resource behavior.
models
behavior_complement
Answer is excluded by any false Ctx check, fcheck None, or false SI agreement. Raw eval success is insufficient. Equivalent display units are permitted; upstream structural validity is assumed.
reverse_reconstruction
Accepted Answer has a set of expression/value/context antecedents producing its retained Q/steps/notes and checks augmented with independent comparison. Numeric SI agreement cannot identify original syntax or provenance uniquely.
temperature_kind
Q.base transports affine points to Kelvin and multiplicative units by scale. Gate compares transported numbers and does not independently validate point/delta structural legality in publicly constructed Q.
failure_precedence
answer evaluates before gate. Gate calls fcheck before Q.base; base error propagates before failed-check aggregation. fcheck None appends failure without evaluating base. Previous checks are inspected after independent processing.
gate_boundary
Tolerance is1e-9*(1+max(abs(b),abs(fv))). Checks append, preserving previous false values. The compiled control bypasses failed-list rejection and must fail the wrong-answer test. This is a gate policy, not a universal approximation guarantee.
empty_nonfinite
An empty prior check list still receives the independent check. NaN comparisons fail ordinarily; infinities require separate analysis, so no finite-answer guarantee is claimed. Ctx variable bindings are not passed through this interface to fcheck.
tests
Actual source module filter: eval::. Inspected synthetic module tests only, selected independently; numeric run statuses, sorted named outcomes and compiled controls retained in wave27-math external receipts. No full crate suite, corpus, teacher, GPU or training. See source-test-map.json for exact evidence and limitations.
This page in Markdown → https://md.groupants.elamur.ai/free/en/mova/math/src/eval.rs.groupant/gate.groupant/