grf_gp.sampler module

GRF random-walk sampler built around torch sparse CSR tensors.

class grf_gp.sampler.GRFSampler(adjacency_matrix: Tensor, walks_per_node: int = 10, p_halt: float = 0.5, max_walk_length: int = 10, seed: int | None = None, use_tqdm: bool = True, n_processes: int | None = None)

Bases: object

Generate GRF random-walk matrices as sparse linear operators.

sample_random_walk_matrices() → List[SparseLinearOperator]

Sample per-step random-walk matrices.

Returns:

A list of per-step random-walk matrices wrapped as SparseLinearOperator objects.