Variable IDs

CITs are identified relative to variables involved. The variables themselves may be addressed by any totally ordered (wrt the less-equal operator) hashable type for example integers or tuples of integers.

var_index: TypeVar

CIT Identifiers

class CI_Identifier

Bases: Generic[var_index]

A multi-index defining a conditional independence-statement

Note

The index type var_index must be comparable (totally ordered) and hashable, eg int or tuple of int, type-annotations are given for the type-var var_index; for time-series, see also CI_Identifier_TimeSeries.

var_index = var_index

__init__(idx_x: var_index, idx_y: var_index, idx_list_z: list[var_index])

Construct representation of CIT for single-variables X, Y and a list of indices for the variables in the set of conditios Z. The representation is undirected and for a (logical) set Z: Reordering X and Y or changing the order in the list of inices for Z will result in the same representation, see also __hash__() and __eq__().

Parameters:

idx_x (var_index) – Index of variable X
idx_y (var_index) – Index of variable Y
idx_list_z (list[var_index]) – List of indices of variables in set of conditions Z.

undirected_link() → tuple[var_index, var_index]

Get the associated (undirected) link as a tuple.

Returns:: The associated undirected link.
Return type:: tuple[var_index,var_index]

__hash__() → int

Hash for unordered containers (dict, set).

Returns:: A hash value.
Return type:: int

__eq__(other: CI_Identifier) → bool

Equality compare two CI-identifiers as undirected (X and Y are exchangable) test with a set (order does not matter) of condtions.

Parameters:: other (CI_Identifier) – Other CI-Identifier to compare to.
Returns:: Equality
Return type:: bool

z_dim() → int

Get dimension of (number of variables in) conditioning set Z.

Returns:: dim(Z)
Return type:: int

conditioning_set() → set[var_index]

Get the conditioning set Z (as set).

Returns:: Z
Return type:: set[var_index]

CIT Identifiers Timeseries

class CI_Identifier_TimeSeries

Bases: CI_Identifier[tuple[int, int]]

A multi-index defining a conditional independence-statement for timeseries, using tigramite’s indexing-convention: Individual nodes are indexed by a pair (index, -lag).

max_abs_timelag() → int

Get the maximum (abolute) time-lag of any variable involved in the test. This means, the maximum over -lag for the lags stored in X, Y or any member of Z.

Returns:: Maximum absolute time-lag.
Return type:: int