pyphi.analysis.Analysis#

class pyphi.analysis.Analysis(system, sia, ces)[source]#

Bases: Displayable, Serializable

A single system’s analysis: its SIA, its CES, and their scalar values.

The cause-effect structure embeds the system irreducibility analysis, and sia, ces, and phi are always populated.

Notes

phi and big_phi are different quantities. phi is φₛ, the system integrated information, which decides whether the system exists as one whole; big_phi is Φ, the structure integrated information, the sum of φ over the Φ-structure’s distinctions and relations. A system can have φₛ = 0 — it is reducible — while its distinctions still sum to a nonzero Φ.

For results computed under an earlier version of IIT (see Reproduce results from earlier versions of IIT): under IIT 3.0 the CES has no relations, the system-level quantity phi is that version’s Φ, and big_phi is not defined.

Parameters:
property phi: float#

φₛ, the system integrated information.

Type:

float

property formalism: str#

The version of IIT that produced this analysis: "IIT_4_0_2026" (the default), "IIT_4_0_2023", or "IIT_3_0". Read from the configuration snapshot the system irreducibility analysis carries.

Type:

str

property big_phi: float#

Φ, the structure integrated information — the sum of φ over the Φ-structure’s distinctions and relations.

Raises:

AttributeError – For a result computed under IIT 3.0, which has no relations and so no structure integrated information. That version’s Φ is phi.

Type:

float

to_pandas()[source]#

A one-row DataFrame that identifies the analysis and gives its values.

The identifiers are formalism (the version name), subset (the system’s sorted unit indices) and state (the substrate’s state), as in a sweep’s rows, and current_state (the system’s state), as in the SIA’s own row. The values are φₛ (phi), normalized_phi, Φ (big_phi), n_distinctions, sum_phi_d and sum_phi_r, the same columns as a sweep’s rows for compute="ces" or compute=None.

Return type:

DataFrame