morphogen.seeding

Reproducible random number generation.

Every random draw in morphoGen goes through a numpy.random.Generator built from a seed. The realizations of a study get seeds derived from the study seed and their indices (SeedSequence(seed, spawn_key=(size_index, realization_index))), so a result does not depend on how the study was split across processes or cluster jobs.

morphogen.seeding.derive_seed(base: int | None, *key: int) → SeedSequence[source]

Seed of the item key (e.g. (size_index, realization_index)) of a study.

morphogen.seeding.make_rng(seed: int | SeedSequence | Generator | None = None) → tuple[Generator, dict | None][source]

Return a generator and a JSON-able description that reproduces it.

seed may be an integer, a SeedSequence, an existing generator (used as is, not reproducible from the description) or None (fresh entropy, recorded so that the realization can be regenerated).

morphogen.seeding.describe_seed(ss: SeedSequence) → dict[source]

The entropy and spawn key of a seed sequence, which recreate it (recorded with the results).

morphogen.seeding.seed_from_description(desc: dict | int | None) → SeedSequence | int | None[source]

Inverse of describe_seed() (also accepts a plain integer).