dymad.sako.snapshot

Typed spectral snapshot records for checkpoint-backed analysis seams.

Functions

build_spectral_snapshot(*, model_class, ...)

Build a typed spectral snapshot from checkpoint-derived arrays.

Classes

KoopmanWeightSnapshot(mode[, full_matrix, ...])	Typed Koopman weight payload from checkpoint-backed models.
SpectralSnapshot(model_class, ...[, metadata])	Typed spectral inputs extracted from one checkpoint-backed model state.

class dymad.sako.snapshot.KoopmanWeightSnapshot(mode, full_matrix=None, left_factor=None, right_factor=None)

Bases: object

Typed Koopman weight payload from checkpoint-backed models.

full_matrix: ndarray | None = None

left_factor: ndarray | None = None

mode: str

right_factor: ndarray | None = None

class dymad.sako.snapshot.SpectralSnapshot(model_class, checkpoint_path, encoded_p0, encoded_p1, koopman_weights, input_dim, obs_dim, sample_count, metadata=<factory>)

Bases: object

Typed spectral inputs extracted from one checkpoint-backed model state.

checkpoint_path: str

encoded_p0: ndarray

encoded_p1: ndarray

input_dim: int

koopman_weights: KoopmanWeightSnapshot

metadata: dict[str, Any]

model_class: str

obs_dim: int

sample_count: int

dymad.sako.snapshot.build_spectral_snapshot(*, model_class, checkpoint_path, encoded_p0, encoded_p1, weights, input_dim, obs_dim, metadata=None)

Build a typed spectral snapshot from checkpoint-derived arrays.

Return type:

SpectralSnapshot