The Open Game Engine is our flagship open source software for computational game theory. It is an embedded domain specific language that extends the functional programming language Haskell with a powerful syntax for defining games in a compositional way, and then running computational analyses on them.

Supported by a grant from the Ethereum Foundation, we are developing applications of the Open Game Engine to game-theoretic analysis of smart contracts, for example for token auctions and DeFi contracts. You can read about this project here. We are also working on integration with the dapptools project to partially automate this process.

We are investigating the possible collusive behaviour of pricing algorithms in realistic training environments, using a reinforcement learning framework built on top of the Open Game Engine. You can read more about this project in this paper.

A team of academic collaborators based mainly at the MSP Group in Glasgow are studying the category-theoretic structure of general cybernetic systems, and the ways in which different real-world methods and problems map onto these structures. This includes deep learning, dynamic programming and reinforcement learning, Bayesian and variational learning, and economic game theory. You can read more in this paper.