SQUARE Symsim is a test-bed for implementing reinforcement learning algorithms, formalizing their correctness properties, and testing them. It is implemented in Scala 3, in purely functional style, and uses property-based testing.