Extend singular equilibrium to strategic environments

Develop singular-equilibrium concepts for games with multiple players and mixed actions, extending the single-agent framework studied for pure actions.

Background

The paper defines singular equilibrium for a single Bayesian learning agent whose actions affect the data-generating process. The concept restricts long-run beliefs to parameters in the concentration set selected by the local learning coefficient.

The authors explicitly identify extending this solution concept to multi-player games and mixed actions as an unresolved extension of their framework.