Struct McProgram 

pub struct McProgram { /* private fields */ }

Implementations

impl McProgram

pub fn compile_evidence_forcing(&self) -> Result<EvidenceForcing>

impl McProgram

pub fn evaluate_resident_with_provider( &self, cfg: McEvalConfig, provider: Arc<CudaKernelProvider>, ) -> Result<McResidentResult>

Evaluate this program with the GPU-resident engine. Returns device-resident counts plus no-host instrumentation for the measured region. Fails closed (typed ResidentRejection wrapped into [XlogError]) for programs outside the supported fragment.

pub fn evaluate_resident(&self, cfg: McEvalConfig) -> Result<McResidentResult>

Convenience: evaluate with a fresh provider.

impl McProgram

pub fn compile_source(source: &str) -> Result<Self>

pub fn compile_source_with_gpu(source: &str, config: GpuConfig) -> Result<Self>

pub fn compile_from_program( program: &Program, config: GpuConfig, ) -> Result<Self>

Compile an already parsed program with the requested GPU configuration.

Imports must already be resolved and merged. This method does not load unresolved use declarations from the filesystem.

pub fn num_vars(&self) -> usize

pub fn evaluate(&self, cfg: McEvalConfig) -> Result<McResult>

Host-facing MC evaluation: runs the resident megakernel engine (Self::evaluate_gpu_device_with_provider) and then materializes the result on the host by downloading the final query/evidence counts after the measured region. The download is a host-result materialization, not part of the measured region — the no-host property belongs to the resident engine, not to this convenience wrapper. Use Self::evaluate_gpu_device when you want device-resident counts with no host download at all.

Fail-closed contract: if the resident engine rejects the program (negation, aggregates, unbounded terms, …), this returns the typed rejection error. The CPU oracle is reachable only via the explicit McEvalConfig::allow_cpu_oracle_fallback opt-in and its result is labeled McEngine::CpuOracle.

pub fn evaluate_cpu(&self, cfg: McEvalConfig) -> Result<McResult>

CPU oracle / debug MC path. Downloads the full sampled-bit matrix to the host and evaluates every sampled world on a host relation store.

This is intentionally not GPU-native: it performs a large DtoH of the sample matrix and runs the deterministic core on the CPU. It exists solely as a deterministic, seed-matched oracle for validating the GPU-native device counts (the GPU sampler is shared, so for the same program/seed the two paths see identical samples). It must never be used as zero-host / GPU-native release evidence, and the acceptance matrix excludes it and the tests that call it (tests/gpu_mc_vs_cpu.rs, tests/mc.rs).

pub fn evaluate_gpu(&self, cfg: McEvalConfig) -> Result<McResult>

Alias for Self::evaluate: GPU device evaluation followed by host-result materialization (final-count download after the measured region). The _gpu suffix denotes that the compute runs on the GPU — it does not imply a zero-host result, since it returns a host McResult. For the device-resident, no-host-download API use Self::evaluate_gpu_device.

pub fn evaluate_gpu_device(&self, cfg: McEvalConfig) -> Result<McDeviceResult>

GPU-native device-resident MC evaluation via the resident megakernel engine ([resident]). Returns McDeviceResult with counts left on the device (no host download). The engine evaluates all worlds in a single launch with no host interaction in the measured region (no host sample loop, no per-sample/per-operator host launches or allocations, no tracked transfers, and no untracked metadata reads); see McResidentResult::no_host / McNoHostStats::is_no_host for the full measured contract.

pub fn evaluate_gpu_device_with_provider( &self, cfg: McEvalConfig, provider: Arc<CudaKernelProvider>, ) -> Result<McDeviceResult>

GPU-native device-resident MC evaluation using a caller-supplied provider (enables provider/buffer reuse across calls). Static setup (arena allocation, plan upload, sampling) happens before the measured region; the measured region is a single resident-engine launch with zero host interaction — no host loop over samples or fixpoint iterations, no per-sample/per-operator host launches or device allocations, no tracked HtoD/DtoH transfers, and no untracked metadata reads. The full contract is measured by McNoHostStats (McResidentResult::no_host) and gated by tests/mc_resident.rs. Counts remain device-resident; the caller decides whether/when to download them.

Trait Implementations

impl Clone for McProgram

fn clone(&self) -> McProgram

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.