pyphi.formalism.iit3.formalism.IIT3Formalism#

class pyphi.formalism.iit3.formalism.IIT3Formalism[source]#

Bases: object

IIT 3.0 (Oizumi et al. 2014) — distribution-distance phi computation.

evaluate_mechanism(system, direction, mechanism, purview, **kwargs)[source]#

Public mechanism-level evaluation. Calls back through queries.find_mip to preserve the short-circuit logic (empty purview, unreachable state) the public dispatcher owns.

Parameters:
Return type:

Any

evaluate_mechanism_partition(system, direction, mechanism, purview, partition, repertoire=None, partitioned_repertoire=None, repertoire_distance=None, partitioned_repertoire_kwargs=None, **kwargs)[source]#

IIT 3.0 mechanism-partition integration: distribution-distance between unpartitioned and partitioned repertoires.

repertoire_distance is a Protocol-typed measure callable (resolved here from config if not provided); mechanism_measure is threaded through to the partitioned-repertoire helper.

Parameters:

  • system (Any)

  • direction (Any)

  • mechanism (Any)

  • purview (Any)

  • partition (Any)

  • repertoire (Any)

  • partitioned_repertoire (Any)

  • repertoire_distance (Any)

  • partitioned_repertoire_kwargs (Any)

  • kwargs (Any)

Return type:

Any

evaluate_system(system, **kwargs)[source]#

Delegate to the IIT 3.0 sia in pyphi.formalism.iit3.

IIT 3.0 has no specified-state phase, so measure kwargs are not threaded through this method. The system-level measure is read from config.formalism.iit.mechanism_phi_measure inside the underlying sia implementation; compatibility is checked against the active formalism’s compatible_measures here. Callers attempting to pass system_measure / specification_measure receive a TypeError rather than a silent no-op.

Parameters:
Return type:

Any

build_ces(system, **kwargs)[source]#

IIT 3.0 CES is exactly the set of distinctions (no relations).

Returns the distinctions specified by the system; this is the IIT 3.0 cause-effect structure.

Parameters:
Return type:

Any