zoomy_core.model.models.legacy.basis_cache module

Hash-based cache for basis function matrices.

Replaces the old flat-file cache in Basismatrices with a manifest-based system: - Each (basis_type, level, weight) combination gets a unique hash - All matrices stored in a single .npz file per hash - Central manifest.json tracks all entries with metadata - Stale detection via basis definition fingerprint

class zoomy_core.model.models.legacy.basis_cache.BasisMatrixCache(cache_root='.cache/basismatrices')

Bases: object

Hash-based cache for basis matrices.

Usage:

cache = BasisMatrixCache() matrices = cache.get_or_compute(basis, level) # matrices is a dict: {“M”: np.array, “A”: np.array, …}

get_or_compute(basis, level, compute_fn)

Load cached matrices or compute and cache them.

Parameters:

• basis (Basisfunction) – The basis function object.

• level (int) – Polynomial level.

• compute_fn (callable) – Function that computes matrices. Called as compute_fn(level) and should set self.M, self.A, etc. on the Basismatrices object. Returns a dict of {name: np.ndarray}.

Returns:

dict

Return type:

{name: np.ndarray} for each matrix

list_entries()

clear()