world-induce-gpu-subset-kernel

The groupant beside world/src/induce_gpu.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`
composition	partial	composition
cost	partial	cost
descriptant	present	descriptant
invariant	partial	invariant
tests	unexecuted gpu boundary	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
`world_induce_gpu_subset_kernel_invariants`	world_induce_gpu_subset_kernel_invariants	partial source derived model	world_induce_gpu_subset_kernel_invariants
`world_induce_gpu_subset_kernel_resource_cost`	world_induce_gpu_subset_kernel_resource_cost	partial source derived model	world_induce_gpu_subset_kernel_resource_cost
`world_induce_gpu_subset_kernel_representation_join`	world_induce_gpu_subset_kernel_representation_join	partial source derived model	world_induce_gpu_subset_kernel_representation_join
`world_induce_gpu_subset_kernel_domain_partition`	world_induce_gpu_subset_kernel_domain_partition	partial source derived model	world_induce_gpu_subset_kernel_domain_partition

Model text

descriptant

world-induce-gpu-subset-kernel

w=blockIdx*blockDim+threadIdx; w>=nwords returns. Iterate i<j<k in off word range; probe(ci),probe(ci|cj<<21),probe(ci|cj<<21|ck<<42). Match key+rel+1 then atomicAdd1 and return.

Modeling epoch 2026-10-31

Prospective source/branch/cost hypotheses were saved externally before source edits. This partial source-derived model retains observed boundaries rather than claiming complete verification. GPU source/embedded CUDA modeled, no native compile, execution or compiled negative control. No surrogate/equivalence measurement.

approximant

Forward relation

w=blockIdx*blockDim+threadIdx; w>=nwords returns. Iterate i<j<k in off word range; probe(ci),probe(ci|cj<<21),probe(ci|cj<<21|ck<<42). Match key+rel+1 then atomicAdd1 and return.

Domain: Well-formed host arrays, reachable sentinel; static-only evidence.

No executable percent surrogate: these categorical/effect observations lack a useful approximation metric. Future supported caps1/2/5/10% need a defined domain/metric and real reference evidence; no error or speed bound is claimed.

invariant

No off/cnt bounds checks;21bit atom slots differ from flat offsets. Counts exact only below wrap threshold and trusted valid host table.

Cost component: cost.md.

cost

256threads/block, cubic subsets per word, linear worst probe and serialized contested atomics.

composition

Host must match hash constants/mask/sentinel; this scoped CUDA string is not executed by lexical Rust AST projection.

models

behavior_complement

Out-of-range thread writes nothing. Missing lookup terminates at sentinel; malformed full table may not terminate. Duplicate equal slots route only first match.

reverse_reconstruction

Observed c[g] admits all input words producing same c[g] queried subsets modulo2^32, no unique recovery of feature grouping or unused values.