parallel_slicing

mqed.Dyadic_GF.parallel_slicing._assemble_sliced_results(all_results: list[list[tuple]], n_energy: int, n_rx: int, save_components: bool, worker_label: str) → tuple[numpy.ndarray, ...]

Validate sliced worker results and restore globally ordered tensors.

mqed.Dyadic_GF.parallel_slicing._task_slices(n_energy: int, n_rx: int, worker_count: int, integration_method: str, scheduler: str, rx_chunk_size: int | None = None) → list[tuple[int, numpy.ndarray]]

Build ordered energy/Rx work units for local and MPI executors.

The flattened scheduler uses one process pool or MPI communicator; it does not create nested parallel loops. Each task contains one energy and one contiguous slice of the global Rx grid. fixed_grid always receives the complete Rx row because its solver shares sampled q kernels across all Rx.

mqed.Dyadic_GF.parallel_slicing.assemble_sliced_results(all_results: list[list[tuple]], n_energy: int, n_rx: int, save_components: bool, worker_label: str) → tuple[numpy.ndarray, ...]

Validate sliced worker results and restore globally ordered tensors.

mqed.Dyadic_GF.parallel_slicing.task_slices(n_energy: int, n_rx: int, worker_count: int, integration_method: str, scheduler: str, rx_chunk_size: int | None = None) → list[tuple[int, numpy.ndarray]]

Build ordered energy/Rx work units for local and MPI executors.

The flattened scheduler uses one process pool or MPI communicator; it does not create nested parallel loops. Each task contains one energy and one contiguous slice of the global Rx grid. fixed_grid always receives the complete Rx row because its solver shares sampled q kernels across all Rx.