Bridges

In bridges we provide wrappers for seamless integration with CD-algorithm implemenations in other packages.

To integrate further third-party or your own algorithms and implemenations, they only have to be wrapped to satisfy the format abstract_cd_t, for more details see Abstract CD Specification.

Tigramite Integration

Bridge for tigramite integration. Tigramite specializes to causal methods (including causal discovery) for time-series. Our output-graphs (see graph_t) are in tigramite’s internal format, and can be used directly e.g. with effect-estimation or mediation code from tigramite.

alg_pcmci(data_format: IManageData, mci_transition_callback: IHandleExplicitTransitionToMCI = None, pcmci_obj_init_args: dict = {}, pcmci_obj_run_args: dict = {}) → abstract_cd_t

Get PCMCI [RNK+19] implementation from tigramite. Use with ControllerTimeseries.

Parameters:
Returns:

PCMCI as abstract CD-algorithm.

Return type:

abstract_cd_t

alg_pcmciplus(data_format: IManageData, mci_transition_callback: IHandleExplicitTransitionToMCI = None, pcmci_obj_init_args: dict = {}, pcmci_obj_run_args: dict = {}) → abstract_cd_t

Get PCMCI+ [R20] implementation from tigramite. Use with ControllerTimeseries.

Parameters:
Returns:

PCMCI+ as abstract CD-algorithm.

Return type:

abstract_cd_t

alg_lpcmci(data_format: IManageData, lpcmci_obj_init_args: dict = {}, lpcmci_obj_run_args: dict = {}) → abstract_cd_t

Get LPCMCI [GR20] implementation from tigramite. Use with ControllerTimeseriesLPCMCI.

Note: In this case a transition callback IHandleExplicitTransitionToMCI is notified by the controller ControllerTimeseriesLPCMCI.

Parameters:

  • data_format (IManageData) – data manager

  • lpcmci_obj_init_args (dict) – forwarded to tigramites LPCMCI constructor

  • lpcmci_obj_run_args (dict) – forwarded to tigramites LPCMCI.run_lpcmci

Returns:

LPCMCI as abstract CD-algorithm.

Return type:

abstract_cd_t

Causal Learn Integration

Bridge for causal-learn [ZHC+24] integration. Causal Learn provides a broad range of causal methods, currently our framework is ready for use with their basic IID-data constraint-based CD-method implementations.

alg_pc(data_format: IManageData, **runtime_args) → abstract_cd_t

Get PC [SG91] implementation from causal learn.

Parameters:

  • data_format (IManageData) – data manager

  • runtime_args – forwarded to causal-learns run_pc (together with “stable=False”)

Returns:

PC as abstract CD-algorithm.

Return type:

abstract_cd_t

alg_pc_stable(data_format: IManageData, **runtime_args) → abstract_cd_t

Get PC-stable [CM+14] implementation from causal learn.

Parameters:

  • data_format (IManageData) – data manager

  • runtime_args – forwarded to causal-learns run_pc (together with “stable=True”)

Returns:

PC-stable as abstract CD-algorithm.

Return type:

abstract_cd_t

alg_fci(data_format: IManageData, **runtime_args) → abstract_cd_t

Get FCI [SGS01] implementation from causal learn.

Parameters:

  • data_format (IManageData) – data manager

  • runtime_args – forwarded to causal-learns run_fci

Returns:

FCI as abstract CD-algorithm.

Return type:

abstract_cd_t