![[2508.15555] HEAS: Hierarchical Evolutionary Agent Simulation Framework for Cross-Scale Modeling and Multi-Objective Search](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2508.15555] HEAS: Hierarchical Evolutionary Agent Simulation Framework for Cross-Scale Modeling and Multi-Objective Search
Summary
The HEAS framework integrates agent-based modeling with evolutionary optimization, enabling cross-scale modeling and multi-objective search in a reproducible manner.
Why It Matters
HEAS provides a unified approach to modeling complex systems, making it easier for researchers and practitioners to conduct reproducible studies across various domains. Its emphasis on modularity and standardization enhances the reliability of results, which is crucial for advancing multi-disciplinary research.
Key Takeaways
- HEAS combines agent-based modeling with evolutionary optimization for enhanced simulation.
- The framework supports reproducibility through standardized evaluation metrics and logging.
- Modularity allows for flexible model composition without extensive refactoring.
- Examples demonstrate HEAS's application in ecological and enterprise decision-making contexts.
- Facilitates cross-disciplinary research by providing a practical foundation for complex inquiries.
Computer Science > Multiagent Systems arXiv:2508.15555 (cs) [Submitted on 21 Aug 2025 (v1), last revised 14 Feb 2026 (this version, v2)] Title:HEAS: Hierarchical Evolutionary Agent Simulation Framework for Cross-Scale Modeling and Multi-Objective Search Authors:Ruiyu Zhang, Lin Nie, Xin Zhao View a PDF of the paper titled HEAS: Hierarchical Evolutionary Agent Simulation Framework for Cross-Scale Modeling and Multi-Objective Search, by Ruiyu Zhang and 2 other authors View PDF HTML (experimental) Abstract:Hierarchical Evolutionary Agent Simulation (HEAS) is a Python framework that unifies layered agent-based modeling with evolutionary optimization and tournament evaluation in a single, reproducible workflow. HEAS represents models as hierarchies of lightweight processes ("streams") scheduled in deterministic layers that read and write a shared context, making cross-scale couplings explicit and auditable. A compact API and CLI-simulate, optimize, evaluate-expose single- and multi-objective evolution, PyTorch policy integration via parameter flattening/unflattening, and general tournament tooling with user-defined scoring and voting rules. The framework standardizes evaluation through uniform per-step and episode metrics, persists seeds, logbooks, and hall-of-fame archives, and provides plotting helpers for traces, Pareto fronts, and comparative outcomes, reducing glue code and improving comparability across studies. HEAS emphasizes separation of mechanism from orchestration, ...
