Feat: Multi-Objective API Integration (BiEGO, MF-BiEGO) & Advanced Profiling

[Paul-Saves]

This Pull Request introduces full Multi-Objective optimization support into the core smt-optim API. It ports and
integrates the BiEGO (Bio-objective Efficient Global Optimization) strategy, extends it to Multi-Fidelity (MF-BiEGO)
with constraint handling, and provides comprehensive post-processing tools for Multi-Objective comparisons.
Key Features & Additions
• BiEGO Strategy: Fully integrated the BiEGO strategy with vectorized EI formulations. Includes both the standard
"Composite" approach (with adaptive Nadir point tracking) and a "Naive" BiEGO formulation for benchmarking.
• Multi-Fidelity BiEGO (MF-BiEGO): Added support for multi-fidelity multi-objective optimization with constraints.
• Public API Support: Enabled native multi-objective support in the public minimize API, State , and
OptimizationDataset .
• Advanced Post-Processing ( smt_optim.utils.profiles ): Integrated tools based on oihanc and amaug's work to
generate Data Profiles, Performance Profiles, and Hypervolume Convergence Profiles.
Documentation & Notebooks

• Multi-Objective Tutorial Overhaul: Completely updated docs/source/getting_started/multi_objective_optim.ipynb to
reflect the new v0.2.0 API.
• Interactive Animations: Added rich matplotlib animations visualizing the Pareto Front discovery process on the
ZDT1 benchmark for BiEGO Composite , BiEGO Naive , and MOSEGO .
• Hypervolume Analysis: Added concrete examples in the tutorial for plotting Hypervolume Convergence and Performance
Profiles.

Testing & Bug Fixes
• Added robust integration and convergence tests for BiEGO and MOSEGO on the ZDT1 problem, including assertions for
Hypervolume decay and IGD+ decay.
• Fixed PymooWrapper compatibility issues with multi-fidelity evaluation.
• Resolved various ruff formatting warnings (E402 module level imports, E902 encoding) and removed deprecated API
calls ( smt_optim.adapters , _points attribute).
• Cleaned up duplicate functions (e.g., accuracy_profile in profiles.py ).

Checklist

[✓] Code passes ruff check cleanly.
[✓] Multi-objective and Multi-fidelity convergence tests are passing.
[✓] Documentation and notebooks have been updated and verified.
[✓] Coverage remains stable or has increased (+0.5%).

[coveralls]

Coverage Report for CI Build 28764842813

Coverage increased (+0.5%) to 86.58%

Details

• Coverage increased (+0.5%) from the base build.
• Patch coverage: 89 uncovered changes across 9 files (297 of 386 lines covered, 76.94%).
• 14 coverage regressions across 2 files.

Uncovered Changes

File	Changed	Covered	%
src/smt_optim/benchmarks/base.py	31	4	12.9%
src/tests/test_integration/test_biego.py	24	4	16.67%
src/tests/test_convergence/test_multi_objective.py	147	128	87.07%
src/smt_optim/acquisition_strategies/biego.py	72	55	76.39%
src/smt_optim/utils/logger.py	5	3	60.0%
src/smt_optim/acquisition_functions/multi_obj.py	29	28	96.55%
src/smt_optim/acquisition_strategies/mosego.py	5	4	80.0%
src/smt_optim/frameworks.py	1	0	0.0%
src/smt_optim/utils/initial_design.py	3	2	66.67%
Total (17 files)	386	297	76.94%

Coverage Regressions

14 previously-covered lines in 2 files lost coverage.

File	Lines Losing Coverage	Coverage
src/smt_optim/acquisition_strategies/mosego.py	10	79.73%
src/smt_optim/acquisition_strategies/biego.py	4	79.1%

---

Coverage Stats

Relevant Lines:	5872
Covered Lines:	5084
Line Coverage:	86.58%
Coverage Strength:	1.73 hits per line

---

💛 - Coveralls