We design an RL environment for automated mathematical theory formation and present an evolutionary synthesis methodology for interestingness measures to guide exploration in it