Reproduce “ADAGE: A generic two-layer framework for adaptive agent based modelling” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Benjamin Patrick Evans, Sihan Zeng, Sumitra Ganesh, Leo Ardon
ADAGE, a framework from JP Morgan AI Research, addresses the limitation of fixed agent behaviors in traditional agent-based models by integrating bi-level adaptation. It employs a two-layer architecture, formalizing the interaction as a Stackelberg game, and demonstrates improved performance across tasks like policy design, calibration, scenario generation, and robust behavioral learning, achieving more equitable outcomes in economic models and better data fit in financial simulations.
Key insights
- Traditional agent-based models (ABMs) are limited by fixed agent behaviors, which fail to adapt to changing environments, leading to inaccurate predictions (Lucas critique).
- ADAGE addresses this by integrating bi-level adaptation: agents learn and adapt to their environment, and the environment simultaneously adapts to the collective behavior of agents.
- The framework unifies traditionally separate ABM tasks, such as policy design, calibration, scenario generation, and robust behavioral learning, under a single mathematical formulation.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.