MOGA optimization

perform_moga_optimization (evaluate/funs_evaluate.py) drives Dakota's Multi-Objective Genetic Algorithm (MOGA) method (create_moga_optimization_conffile in config/) over one or more objective functions defined by a fitted surrogate, and returns the Pareto-optimal set.

Pareto-front support (data/funs_data_processing.py)

• is_dominated / get_non_dominated_indices — standard Pareto dominance check/filter, used both to validate MOGA's own output and (independently) as a reusable utility.
• get_bounds_uniform_distribution(s) — resolves variable bounds from a uniform-distribution spec, feeding the variables block MOGA searches over.

Verified behavior

(Verification & Validation Category G)

• Single-objective minimization finds the known optimum within tolerance.
• A bi-objective front spans the expected trade-off curve and every returned point is genuinely non-dominated (cross-checked against get_non_dominated_indices).
• The front improves (dominates the lower-iteration front) as the iteration budget increases.
• Every returned point respects the input variable bounds.

Known limitation

max_function_evaluations is accepted by the composer but not enforced by the underlying Dakota MOGA method in the pinned engine version (deprecated parameter on the Dakota side) — don't rely on it as a hard evaluation-count cap; use the iteration/generation controls instead. See Known limitations.