pytams.diagdb

A class for the TAMS data as an SQL database using SQLAlchemy.

Classes

DiagBase	A base class for the tables.
DiagnosticEntry	Table for recording model data at specific score levels.
DiagDB	A database to keep track of the diagnostics data.

Module Contents

class DiagBase

Bases: sqlalchemy.orm.DeclarativeBase

A base class for the tables.

class DiagnosticEntry

Bases: DiagBase

Table for recording model data at specific score levels.

id: sqlalchemy.orm.Mapped[int]

traj_id: sqlalchemy.orm.Mapped[int]

level_crossed: sqlalchemy.orm.Mapped[float]

time: sqlalchemy.orm.Mapped[float]

weight: sqlalchemy.orm.Mapped[float]

model_data: sqlalchemy.orm.Mapped[bytes]

active: sqlalchemy.orm.Mapped[bool]

diaglabel: sqlalchemy.orm.Mapped[str]

class DiagDB(file_name: str, in_memory: bool = False, ro_mode: bool = False)

Bases: pytams.sqlmanager.BaseSQLManager

A database to keep track of the diagnostics data.

Diagnostic entries are agregated in single table. Each entry is associated to a trajectory and its weight in the ensemble.

add_diagnostic_entry(diaglabel: str, traj_id: int, level: float, time: float, weight: float, ldata: bytes) → None

Atomic insert of a diagnostic snapshot.

The data schema assumes that any new addition to the database is made on an active trajectory.

Parameters:

• diaglabel – the label of the diagnostic inserting the entry

• traj_id – the ID of the traj adding the entry

• level – the score level of the entry

• time – the trajectory time at which the diagnostic was triggered

• weight – the weight of the trajectory

• ldata – the actual model data stored in the database

get_highest_recorded_level(traj_id: int, label: str) → float

Return the maximum level already recorded for this traj/label.

Parameters:

• traj_id – the ID of a trajectory

• label – the label of the diagnostic targeter

Returns:

the highest value of level_crossed

duplicate_diagnostic_history(ancestor_id: int, discarded_id: int, new_id: int, new_weight: float, threshold: float) → int

Copy diagnostic entries from an ancestor to a descendant.

Copies all entries where level_crossed <= threshold. Returns the number of entries duplicated.

The entries belonging to the discarded trajectory are set to inactive.

Parameters:

• ancestor_id – the ID of the ancestor to copy data from

• discarded_id – the ID of the discarded trajectory (during TAMS iterations)

• new_id – the ID of the new child trajectory

• new_weight – the weight of the new child trajectory

• threshold – the score threshold up to which copy must be performed

update_all_active_weights(new_weight: float) → int

Update all the active trajectories weight.

Parameters:

new_weight – the updated weight

Returns:

the number of trajectory updated

get_diagnostic_data(label: str) → dict[float, list[tuple[Any, float]]]

Retrieve all diagnostic snapshots for a specific label.

Parameters:

label – the label of the diagnostic of interest

Returns:

A dictionary mapping each iso-level (float) to a list of tuples. Each tuple contains (unpickled_data, trajectory_weight).

dump_to_json(json_path: str) → None

Export the entire diagnostic database to a JSON file.

Note that the content of the data stored in the database is omitted. Only the metadata of each stored data is dumpe for debuggin purposes.

close() → None

Dispose of the engine and clear connections.