Release Notes 3.1.0
Overview
This release deepens EDAF's research and usability focus with stronger algorithm documentation, grammar-based symbolic optimization capabilities, expanded UI analytics, and release/publishing hardening.
Highlights
- Grammar-based symbolic regression and classification workflow:
- auto/custom grammar modes
- BNF parsing and validation
- deterministic derivation encoding for discrete EDA drivers
- ERC support with reproducible sampling
- tree rendering/export in dashboard
- Expanded benchmark and domain support:
- disjunct matrix family workflows (DM/RM/ADM)
- boolean/cryptography optimization set
- broader campaign orchestration patterns for multi-run experiments
- Richer web analytics and UX:
- experiment-level convergence CI, success-vs-budget, ECDF, time-to-target views
- sortable tables, export actions, focus mode enhancements
- run/experiment lifecycle controls
- Documentation quality wave:
- strong JavaDoc coverage across algorithms and models
- formula-level and reference-level class documentation
- improved docs IA and release/publishing guidance
- Release automation improvements:
- tag-driven GitHub release artifact workflows
- package publication workflows for repository and central ecosystems
Quality and Engineering
- Additional tests around metric correctness and deterministic behavior.
- Documentation-driven cleanup of operator behavior and configuration clarity.
- Improved consistency across alias drivers and plugin-family descriptions.
Operational Notes
- Existing datasets/config suites remain runnable; new suites extend capabilities for grammar and benchmark campaigns.
- For heavy campaign workloads, prefer sequential SQLite or PostgreSQL-backed multi-worker setups.
Estimation of Distribution Algorithms Framework
Copyright (c) 2026 Dr. Karlo Knezevic
Licensed under the Apache License, Version 2.0.