TABU-STABLE Algorithm

Overview

Tabu-Stable combines the tabu list mechanism of the Tabu algorithm with the deterministic tie-breaking introduced by HC-Stable. Like HC-Stable, it resolves equal-scoring candidate moves using a canonical variable ordering, ensuring that two runs on identical data always produce identical graphs regardless of platform or iteration order.

The tabu memory gives Tabu-Stable the same local-optima escape capability as standard Tabu, while the stability guarantee makes its output fully reproducible. This combination makes it the recommended default for score-based structure learning in CausalIQ Discovery when reproducibility is important.

Class: score · DAG Package: CausalIQ

Reference

Kitson N.K. and Constantinou A.C. (2025) – Stable structure learning with HC- Stable and Tabu-Stable algorithms. Int. J. Approx. Reason. 186, 109522. https://doi.org/10.1016/j.ijar.2025.109522

Hyperparameters

Hyperparameter	Type	Default	Values	Description
iss	float	1.0	—	Imaginary Sample Size weighting the prior in Bayesian scores.
max_elapsed	int	No limit	—	Maximum allowed execution time in seconds.
max_iterations	int	No limit	—	Maximum number of iterations.
no_increase	int	10	—	Iterations permitted without a score improvement.
penalty_weight	float	1.0	—	Weight of the penalty component in AIC and BIC scores.
score	str	bic	aic, bdeu, bge, bic, k2, loglik	Scoring function for score-based learning.
tabulist_len	int	10	—	Length of the tabu list.

Variants

Variant	Package
causaliq	CausalIQ