Skip to content

v1.4.10

Latest

Choose a tag to compare

@lerdmann1601 lerdmann1601 released this 14 Nov 00:12

HERA

HERA is a MATLAB based scientific ranking framework for paired benchmarking. It provides a hierarchical-compensatory logic that integrates effect-size statistics, and automated reporting capabilities.

For more information please refer to the Readme and Project Documentation.

What's New in v1.4.10?

⚙️ Enhancements

  • Python 3.13+ Compatibility (Stable ABI) & Batch Ergonomics: Relaxed PyPI python_requires to >=3.9 so Python 3.13+ users can install and run hera-matlab using MATLAB Runtime's Stable ABI (abi3) while receiving Runtime's compatibility UserWarning (Python 3.9–3.12 remain officially validated). Suppressed blocking interactive console pauses in batch mode and added bootstrap configuration fallbacks.
  • CSV Delimiter & Simulation Reporting Consistency: Standardized remaining convergence and output CSV tables to comma delimiters across +HERA/+analysis/+convergence/ and +HERA/+output/, refactored data/utils/simulation/reporting.py and plotting helpers, updated benchmark summary reports/plots for consistency, and updated example outputs and header descriptions.

📚 Documentation & Usability

🐛 Bug Fixes

  • PyPI Package Initialization (hera_matlab.initialize()): Fixed .github/scripts/prepare_pypi.py calling the non-existent _pir.get_matlab_instance() method (replaced with _pir.initialize_package() and removed unreachable code after return), restoring standalone PyPI usage without local MATLAB fallbacks.
  • run_ranking Python Return Marshaling: Added recursive cell-array sanitization in deploy/wrappers/run_ranking.m so 2D cell tables inside results.swap_details convert cleanly to nested 1 x N cells instead of raising ValueError: cell arrays returned from MATLAB must be 1-by-N or M-by-1 during hera.run_ranking(config, nargout=1).
  • Numerical Precision, Workflow Paths & Build Automation: Added 1e-12 floating-point tolerance to zero checks in relative_difference.m and calculate_thresholds.m, corrected REPO_ROOT and deploy/output/matlab discovery in data/utils/run_workflow.py, made eval_rank_stability.py symlink-safe, fixed build_hera_python() invocation in .github/workflows/build_release.yml, and made deploy/build_HERA_matlab.m architecture checks dynamic via computer('arch').

Recent Major Features (from v1.4.0)

✨ Convergence Analysis Enhancements:

  • New Scenario: Added a "Sensitivity" scenario with small effect sizes.
  • JSON Configuration: All scenarios are fully customizable through a JSON configuration file.

(For a full history of previous features and changes, see the Changelog)

🖥️ Platform Support

📦 Compatibility

  • MATLAB: Compatible with R2020a and newer.
  • Python: Compatible with Python 3.9 through 3.12 (officially supported; Python 3.13+ supported via Stable ABI with a runtime compatibility warning) and MATLAB Runtime R2025b (25.2).

Full Changelog: v1.4.9...v1.4.10