Struct EpistemicGpuExecutionResult 

pub struct EpistemicGpuExecutionResult {

    pub provider_identity: EpistemicGpuProviderIdentity,
    pub prepared: EpistemicGpuPreparedExecution,
    pub candidate_generation: EpistemicGpuCandidateGenerationTrace,
    pub propagation: EpistemicGpuPropagationTrace,
    pub candidate_validation: EpistemicGpuCandidateValidationTrace,
    pub model_membership: EpistemicGpuModelMembershipTrace,
    pub world_view_validation: EpistemicGpuWorldViewValidationTrace,
    pub constraint_world_view_validation: EpistemicGpuConstraintWorldViewValidationTrace,
    pub materialization: EpistemicGpuMaterializationTrace,
    pub final_result_materialization: EpistemicGpuFinalResultMaterializationTrace,
    pub final_tuple_materialization: EpistemicGpuFinalTupleMaterializationTrace,
    pub transfer_budget: EpistemicGpuTransferBudgetTrace,
    pub final_result_transfer: EpistemicGpuFinalResultTransferTrace,
    pub constraint_validation: EpistemicGpuConstraintValidationTrace,
    pub semantic_trace: EpistemicGpuSemanticTrace,
    pub tuple_membership_bindings: Vec<EpistemicTupleMembershipBinding>,
    pub final_output: CudaBuffer,
    pub additional_head_outputs: Vec<(String, CudaBuffer)>,
    pub tuple_evidence_output: Option<CudaBuffer>,
    pub output: CudaBuffer,
    pub trace: EpistemicGpuRuntimeTrace,
}

Output from executing the reduced production runtime plan for an epistemic program.

Fields

provider_identity: EpistemicGpuProviderIdentity

CUDA provider identity that owns this result’s device-resident buffers.

prepared: EpistemicGpuPreparedExecution

Prepared workspace and preflight state.

candidate_generation: EpistemicGpuCandidateGenerationTrace

Candidate-generation trace captured before reduced-plan dispatch.

propagation: EpistemicGpuPropagationTrace

Candidate-propagation trace captured before reduced-plan dispatch.

candidate_validation: EpistemicGpuCandidateValidationTrace

Candidate-validation trace captured before reduced-plan dispatch.

model_membership: EpistemicGpuModelMembershipTrace

Model-membership staging trace captured after reduced-plan dispatch.

world_view_validation: EpistemicGpuWorldViewValidationTrace

World-view validation trace captured after model-membership staging.

constraint_world_view_validation: EpistemicGpuConstraintWorldViewValidationTrace

World-view integrity-constraint validation trace captured after world-view validation.

materialization: EpistemicGpuMaterializationTrace

Accepted-candidate materialization trace captured after world-view validation.

final_result_materialization: EpistemicGpuFinalResultMaterializationTrace

Final result materialization trace captured from reduced output metadata.

final_tuple_materialization: EpistemicGpuFinalTupleMaterializationTrace

Final query tuple materialization trace captured after final-result gating.

transfer_budget: EpistemicGpuTransferBudgetTrace

Hot-path host-transfer budget trace for epistemic GPU execution.

final_result_transfer: EpistemicGpuFinalResultTransferTrace

Final-result transfer accounting after the GPU hot path.

constraint_validation: EpistemicGpuConstraintValidationTrace

Reduced integrity-constraint validation after production runtime dispatch.

semantic_trace: EpistemicGpuSemanticTrace

Device-derived semantic summary after world-view validation.

tuple_membership_bindings: Vec<EpistemicTupleMembershipBinding>

Tuple-membership bindings that were validated and executed for this result.

final_output: CudaBuffer

Device-resident final query output buffer.

For a single epistemic output head this is the only materialized relation. For a JOINT-SOLVED coalesced multi-head component this is the PRIMARY head’s output (the last reduction’s head); the remaining coupled heads, each materialized against the SAME accepted world view, are in Self::additional_head_outputs.

additional_head_outputs: Vec<(String, CudaBuffer)>

Additional coupled-head outputs for a JOINT-SOLVED multi-head component.

Empty for single-head execution. Each entry is (head_predicate, buffer) for a distinct epistemic output head OTHER than the primary head, filtered against the shared accepted world view via that head’s row-filter bindings.

tuple_evidence_output: Option<CudaBuffer>

Device-resident final tuple evidence buffer before public projection.

output: CudaBuffer

Output buffer returned by the reduced production execution plan.

trace: EpistemicGpuRuntimeTrace

Runtime counter trace for the reduced production plan dispatch.

Implementations

impl EpistemicGpuExecutionResult

pub fn tuple_evidence_output(&self) -> &CudaBuffer

Device-resident output used to derive concrete tuple-membership evidence.

pub fn require_runtime_dispatch_certification(&self) -> Result<()>

Require that the retained runtime trace certifies the prepared execution.

pub fn aggregate_kernel_timing(&self) -> EpistemicGpuKernelTimingTrace

Aggregate CUDA-event timing from all epistemic GPU hot-path kernels.

pub fn try_aggregate_kernel_timing( &self, ) -> Result<EpistemicGpuKernelTimingTrace>

Checked CUDA-event timing aggregation for certification paths.