Data-driven entropic spatially inhomogeneous evolutionary games

We introduce novel multi-agent interaction models of entropic spatially inhomogeneous evolutionary undisclosed games and their quasi-static limits. These evolutions vastly generalise first- second-order dynamics. Besides the well-posedness these forms interactions, we are concerned with learnability individual payoff functions from observation data. formulate learning as a variational problem, minimising discrepancy between observations predictions by function. The inferred function can then be used to simulate further evolutions, which fully data-driven. prove convergence solutions obtained finite number mean-field limit, minimal value provides quantitative error bound on data-driven evolutions. abstract framework is constructive numerically implementable. illustrate this computational examples where ground truth known not case, including model for pedestrian movement.